2022-03-22 19:46:16 +00:00
using ARMeilleure.Translation ;
2019-09-02 16:03:57 +00:00
using Gtk ;
2020-03-25 17:09:38 +00:00
using LibHac.Common ;
2022-01-12 11:22:19 +00:00
using LibHac.Common.Keys ;
2022-03-22 19:46:16 +00:00
using LibHac.Ncm ;
2020-03-25 17:09:38 +00:00
using LibHac.Ns ;
2022-01-12 11:22:19 +00:00
using LibHac.Tools.FsSystem ;
2023-03-02 02:42:27 +00:00
using LibHac.Tools.FsSystem.NcaUtils ;
2021-02-26 00:11:56 +00:00
using Ryujinx.Audio.Backends.Dummy ;
using Ryujinx.Audio.Backends.OpenAL ;
2021-05-05 21:37:09 +00:00
using Ryujinx.Audio.Backends.SDL2 ;
2021-02-26 00:11:56 +00:00
using Ryujinx.Audio.Backends.SoundIo ;
using Ryujinx.Audio.Integration ;
2021-05-16 15:12:14 +00:00
using Ryujinx.Common ;
2020-08-30 16:51:53 +00:00
using Ryujinx.Common.Configuration ;
2019-09-02 16:03:57 +00:00
using Ryujinx.Common.Logging ;
2022-12-15 17:07:31 +00:00
using Ryujinx.Common.SystemInterop ;
2023-01-04 23:01:44 +00:00
using Ryujinx.Cpu ;
2020-01-21 22:23:11 +00:00
using Ryujinx.Graphics.GAL ;
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501)
* Initial Implementation
About as fast as nvidia GL multithreading, can be improved with faster command queuing.
* Struct based command list
Speeds up a bit. Still a lot of time lost to resource copy.
* Do shader init while the render thread is active.
* Introduce circular span pool V1
Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next.
* Refactor SpanRef some more
Use a struct to represent SpanRef, rather than a reference.
* Flush buffers on background thread
* Use a span for UpdateRenderScale.
Much faster than copying the array.
* Calculate command size using reflection
* WIP parallel shaders
* Some minor optimisation
* Only 2 max refs per command now.
The command with 3 refs is gone. :relieved:
* Don't cast on the GPU side
* Remove redundant casts, force sync on window present
* Fix Shader Cache
* Fix host shader save.
* Fixup to work with new renderer stuff
* Make command Run static, use array of delegates as lookup
Profile says this takes less time than the previous way.
* Bring up to date
* Add settings toggle. Fix Muiltithreading Off mode.
* Fix warning.
* Release tracking lock for flushes
* Fix Conditional Render fast path with threaded gal
* Make handle iteration safe when releasing the lock
This is mostly temporary.
* Attempt to set backend threading on driver
Only really works on nvidia before launching a game.
* Fix race condition with BufferModifiedRangeList, exceptions in tracking actions
* Update buffer set commands
* Some cleanup
* Only use stutter workaround when using opengl renderer non-threaded
* Add host-conditional reservation of counter events
There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made.
* Address Feedback
* Make counter flush tracked again.
Hopefully does not cause any issues this time.
* Wait for FlushTo on the main queue thread.
Currently assumes only one thread will want to FlushTo (in this case, the GPU thread)
* Add SDL2 headless integration
* Add HLE macro commands.
Co-authored-by: Mary <mary@mary.zone>
2021-08-26 22:31:29 +00:00
using Ryujinx.Graphics.GAL.Multithreading ;
2019-10-13 06:02:07 +00:00
using Ryujinx.Graphics.OpenGL ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
using Ryujinx.Graphics.Vulkan ;
2020-01-05 11:49:44 +00:00
using Ryujinx.HLE.FileSystem ;
2020-09-21 03:45:30 +00:00
using Ryujinx.HLE.HOS ;
2021-04-13 01:16:43 +00:00
using Ryujinx.HLE.HOS.Services.Account.Acc ;
2021-05-16 15:12:14 +00:00
using Ryujinx.HLE.HOS.SystemState ;
2021-04-14 10:28:43 +00:00
using Ryujinx.Input.GTK3 ;
using Ryujinx.Input.HLE ;
using Ryujinx.Input.SDL2 ;
2021-01-08 08:14:13 +00:00
using Ryujinx.Modules ;
2022-05-15 11:30:15 +00:00
using Ryujinx.Ui.App.Common ;
2021-01-08 08:14:13 +00:00
using Ryujinx.Ui.Applet ;
2022-05-15 11:30:15 +00:00
using Ryujinx.Ui.Common ;
using Ryujinx.Ui.Common.Configuration ;
using Ryujinx.Ui.Common.Helper ;
2021-01-08 08:14:13 +00:00
using Ryujinx.Ui.Helper ;
using Ryujinx.Ui.Widgets ;
using Ryujinx.Ui.Windows ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
using Silk.NET.Vulkan ;
using SPB.Graphics.Vulkan ;
2022-03-22 19:46:16 +00:00
using System ;
using System.Diagnostics ;
using System.IO ;
using System.Reflection ;
using System.Threading ;
using System.Threading.Tasks ;
2019-11-29 04:32:51 +00:00
using GUI = Gtk . Builder . ObjectAttribute ;
2021-03-22 18:40:07 +00:00
using ShaderCacheLoadingState = Ryujinx . Graphics . Gpu . Shader . ShaderCacheState ;
2019-11-29 04:32:51 +00:00
namespace Ryujinx.Ui
2019-09-02 16:03:57 +00:00
{
public class MainWindow : Window
{
2021-08-12 21:56:24 +00:00
private readonly VirtualFileSystem _virtualFileSystem ;
private readonly ContentManager _contentManager ;
private readonly AccountManager _accountManager ;
private readonly LibHacHorizonManager _libHacHorizonManager ;
2019-09-02 16:03:57 +00:00
2021-01-08 08:14:13 +00:00
private UserChannelPersistence _userChannelPersistence ;
2019-09-02 16:03:57 +00:00
2021-01-08 08:14:13 +00:00
private HLE . Switch _emulationContext ;
2019-09-02 16:03:57 +00:00
2021-01-08 08:14:13 +00:00
private WindowsMultimediaTimerResolution _windowsMultimediaTimerResolution ;
2020-09-21 03:45:30 +00:00
2021-01-08 08:14:13 +00:00
private readonly ApplicationLibrary _applicationLibrary ;
private readonly GtkHostUiHandler _uiHandler ;
private readonly AutoResetEvent _deviceExitStatus ;
private readonly ListStore _tableStore ;
2020-02-06 11:38:24 +00:00
2021-01-08 08:14:13 +00:00
private bool _updatingGameTable ;
private bool _gameLoaded ;
private bool _ending ;
2019-09-02 16:03:57 +00:00
2021-01-08 08:14:13 +00:00
private string _currentEmulatedGamePath = null ;
2021-03-18 20:40:20 +00:00
private string _lastScannedAmiiboId = "" ;
private bool _lastScannedAmiiboShowAll = false ;
2021-05-04 16:19:04 +00:00
public RendererWidgetBase RendererWidget ;
2021-04-14 10:28:43 +00:00
public InputManager InputManager ;
2020-05-03 02:00:53 +00:00
2021-06-23 21:44:09 +00:00
public bool IsFocused ;
2020-04-20 21:59:59 +00:00
#pragma warning disable CS0169 , CS0649 , IDE0044
2020-02-12 00:56:19 +00:00
2020-09-29 20:05:25 +00:00
[GUI] public MenuItem ExitMenuItem ;
[GUI] public MenuItem UpdateMenuItem ;
[GUI] MenuBar _menuBar ;
[GUI] Box _footerBox ;
[GUI] Box _statusBar ;
2021-03-18 20:40:20 +00:00
[GUI] MenuItem _optionMenu ;
2021-04-23 20:26:31 +00:00
[GUI] MenuItem _manageUserProfiles ;
2021-06-28 18:54:45 +00:00
[GUI] MenuItem _fileMenu ;
[GUI] MenuItem _loadApplicationFile ;
[GUI] MenuItem _loadApplicationFolder ;
[GUI] MenuItem _appletMenu ;
2021-03-18 20:40:20 +00:00
[GUI] MenuItem _actionMenu ;
2021-09-11 20:08:25 +00:00
[GUI] MenuItem _pauseEmulation ;
[GUI] MenuItem _resumeEmulation ;
2020-09-29 20:05:25 +00:00
[GUI] MenuItem _stopEmulation ;
2021-01-11 15:03:37 +00:00
[GUI] MenuItem _simulateWakeUpMessage ;
2021-03-18 20:40:20 +00:00
[GUI] MenuItem _scanAmiibo ;
2021-06-28 20:09:43 +00:00
[GUI] MenuItem _takeScreenshot ;
2021-08-04 21:28:19 +00:00
[GUI] MenuItem _hideUi ;
2020-09-29 20:05:25 +00:00
[GUI] MenuItem _fullScreen ;
2020-12-01 22:02:27 +00:00
[GUI] CheckMenuItem _startFullScreen ;
2022-03-15 01:35:41 +00:00
[GUI] CheckMenuItem _showConsole ;
2020-09-29 20:05:25 +00:00
[GUI] CheckMenuItem _favToggle ;
[GUI] MenuItem _firmwareInstallDirectory ;
[GUI] MenuItem _firmwareInstallFile ;
2023-01-22 17:39:00 +00:00
[GUI] MenuItem _fileTypesSubMenu ;
2020-10-13 20:54:42 +00:00
[GUI] Label _fifoStatus ;
2020-09-29 20:05:25 +00:00
[GUI] CheckMenuItem _iconToggle ;
[GUI] CheckMenuItem _developerToggle ;
[GUI] CheckMenuItem _appToggle ;
[GUI] CheckMenuItem _timePlayedToggle ;
[GUI] CheckMenuItem _versionToggle ;
[GUI] CheckMenuItem _lastPlayedToggle ;
[GUI] CheckMenuItem _fileExtToggle ;
[GUI] CheckMenuItem _pathToggle ;
[GUI] CheckMenuItem _fileSizeToggle ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
[GUI] Label _gpuBackend ;
2020-09-29 20:05:25 +00:00
[GUI] Label _dockedMode ;
2020-12-16 02:19:07 +00:00
[GUI] Label _aspectRatio ;
2020-09-29 20:05:25 +00:00
[GUI] Label _gameStatus ;
[GUI] TreeView _gameTable ;
[GUI] TreeSelection _gameTableSelection ;
[GUI] ScrolledWindow _gameTableWindow ;
[GUI] Label _gpuName ;
[GUI] Label _progressLabel ;
[GUI] Label _firmwareVersionLabel ;
2022-01-12 11:22:19 +00:00
[GUI] Gtk . ProgressBar _progressBar ;
2020-09-29 20:05:25 +00:00
[GUI] Box _viewBox ;
[GUI] Label _vSyncStatus ;
2021-12-23 16:33:56 +00:00
[GUI] Label _volumeStatus ;
2020-09-29 20:05:25 +00:00
[GUI] Box _listStatusBox ;
2021-03-03 00:39:36 +00:00
[GUI] Label _loadingStatusLabel ;
2022-01-12 11:22:19 +00:00
[GUI] Gtk . ProgressBar _loadingStatusBar ;
2021-03-03 00:39:36 +00:00
2020-04-20 21:59:59 +00:00
#pragma warning restore CS0649 , IDE0044 , CS0169
2019-09-02 16:03:57 +00:00
2019-11-29 04:32:51 +00:00
public MainWindow ( ) : this ( new Builder ( "Ryujinx.Ui.MainWindow.glade" ) ) { }
2019-09-02 16:03:57 +00:00
2022-12-06 22:00:25 +00:00
private MainWindow ( Builder builder ) : base ( builder . GetRawOwnedObject ( "_mainWin" ) )
2019-09-02 16:03:57 +00:00
{
2019-11-29 04:32:51 +00:00
builder . Autoconnect ( this ) ;
2021-01-08 08:14:13 +00:00
// Apply custom theme if needed.
ThemeHelper . ApplyTheme ( ) ;
2022-11-17 12:02:43 +00:00
Gdk . Monitor monitor = Display . GetMonitor ( 0 ) ;
2021-01-08 08:14:13 +00:00
// Sets overridden fields.
2022-11-17 12:02:43 +00:00
int monitorWidth = monitor . Geometry . Width * monitor . ScaleFactor ;
int monitorHeight = monitor . Geometry . Height * monitor . ScaleFactor ;
2020-03-30 22:10:13 +00:00
2021-01-08 08:14:13 +00:00
DefaultWidth = monitorWidth < 1280 ? monitorWidth : 1280 ;
DefaultHeight = monitorHeight < 760 ? monitorHeight : 760 ;
2019-11-29 04:32:51 +00:00
2022-05-15 11:30:15 +00:00
Icon = new Gdk . Pixbuf ( Assembly . GetAssembly ( typeof ( ConfigurationState ) ) , "Ryujinx.Ui.Common.Resources.Logo_Ryujinx.png" ) ;
2021-01-08 08:14:13 +00:00
Title = $"Ryujinx {Program.Version}" ;
2019-12-22 02:49:51 +00:00
2021-01-08 08:14:13 +00:00
// Hide emulation context status bar.
_statusBar . Hide ( ) ;
2020-01-05 11:49:44 +00:00
2021-08-12 21:56:24 +00:00
// Instantiate HLE objects.
_virtualFileSystem = VirtualFileSystem . CreateInstance ( ) ;
_libHacHorizonManager = new LibHacHorizonManager ( ) ;
_libHacHorizonManager . InitializeFsServer ( _virtualFileSystem ) ;
_libHacHorizonManager . InitializeArpServer ( ) ;
_libHacHorizonManager . InitializeBcatServer ( ) ;
_libHacHorizonManager . InitializeSystemClients ( ) ;
// Save data created before we supported extra data in directory save data will not work properly if
// given empty extra data. Luckily some of that extra data can be created using the data from the
// save data indexer, which should be enough to check access permissions for user saves.
// Every single save data's extra data will be checked and fixed if needed each time the emulator is opened.
// Consider removing this at some point in the future when we don't need to worry about old saves.
VirtualFileSystem . FixExtraData ( _libHacHorizonManager . RyujinxClient ) ;
2020-11-19 00:34:28 +00:00
_contentManager = new ContentManager ( _virtualFileSystem ) ;
2022-11-12 23:36:36 +00:00
_accountManager = new AccountManager ( _libHacHorizonManager . RyujinxClient , CommandLineState . Profile ) ;
2021-01-08 08:14:13 +00:00
_userChannelPersistence = new UserChannelPersistence ( ) ;
2019-09-02 16:03:57 +00:00
2021-08-12 21:56:24 +00:00
// Instantiate GUI objects.
2021-01-08 08:14:13 +00:00
_applicationLibrary = new ApplicationLibrary ( _virtualFileSystem ) ;
_uiHandler = new GtkHostUiHandler ( this ) ;
_deviceExitStatus = new AutoResetEvent ( false ) ;
2020-01-05 11:49:44 +00:00
2021-01-08 08:14:13 +00:00
WindowStateEvent + = WindowStateEvent_Changed ;
DeleteEvent + = Window_Close ;
2021-06-23 21:44:09 +00:00
FocusInEvent + = MainWindow_FocusInEvent ;
FocusOutEvent + = MainWindow_FocusOutEvent ;
2020-01-05 11:49:44 +00:00
2021-01-08 08:14:13 +00:00
_applicationLibrary . ApplicationAdded + = Application_Added ;
_applicationLibrary . ApplicationCountUpdated + = ApplicationCount_Updated ;
2021-06-28 18:54:45 +00:00
_fileMenu . StateChanged + = FileMenu_StateChanged ;
2021-03-18 20:40:20 +00:00
_actionMenu . StateChanged + = ActionMenu_StateChanged ;
2021-04-23 20:26:31 +00:00
_optionMenu . StateChanged + = OptionMenu_StateChanged ;
2021-03-18 20:40:20 +00:00
2021-01-08 08:14:13 +00:00
_gameTable . ButtonReleaseEvent + = Row_Clicked ;
_fullScreen . Activated + = FullScreen_Toggled ;
2020-01-21 22:23:11 +00:00
2021-05-04 16:19:04 +00:00
RendererWidgetBase . StatusUpdatedEvent + = Update_StatusBar ;
2019-09-02 16:03:57 +00:00
2021-05-16 15:12:14 +00:00
ConfigurationState . Instance . System . IgnoreMissingServices . Event + = UpdateIgnoreMissingServicesState ;
ConfigurationState . Instance . Graphics . AspectRatio . Event + = UpdateAspectRatioState ;
ConfigurationState . Instance . System . EnableDockedMode . Event + = UpdateDockedModeState ;
2021-12-26 14:17:13 +00:00
ConfigurationState . Instance . System . AudioVolume . Event + = UpdateAudioVolumeState ;
2021-05-16 15:12:14 +00:00
2020-12-01 22:02:27 +00:00
if ( ConfigurationState . Instance . Ui . StartFullscreen )
{
_startFullScreen . Active = true ;
}
2022-03-15 01:35:41 +00:00
_showConsole . Active = ConfigurationState . Instance . Ui . ShowConsole . Value ;
_showConsole . Visible = ConsoleHelper . SetConsoleWindowStateSupported ;
2021-03-18 20:40:20 +00:00
_actionMenu . Sensitive = false ;
2021-09-11 20:08:25 +00:00
_pauseEmulation . Sensitive = false ;
2021-09-18 12:31:44 +00:00
_resumeEmulation . Sensitive = false ;
2019-09-02 16:03:57 +00:00
2023-01-22 17:39:00 +00:00
_fileTypesSubMenu . Visible = FileAssociationHelper . IsTypeAssociationSupported ;
2019-12-21 19:52:31 +00:00
if ( ConfigurationState . Instance . Ui . GuiColumns . FavColumn ) _favToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . IconColumn ) _iconToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . AppColumn ) _appToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . DevColumn ) _developerToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . VersionColumn ) _versionToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . TimePlayedColumn ) _timePlayedToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . LastPlayedColumn ) _lastPlayedToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . FileExtColumn ) _fileExtToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . FileSizeColumn ) _fileSizeToggle . Active = true ;
if ( ConfigurationState . Instance . Ui . GuiColumns . PathColumn ) _pathToggle . Active = true ;
2019-11-29 04:32:51 +00:00
2021-02-09 09:24:37 +00:00
_favToggle . Toggled + = Fav_Toggled ;
_iconToggle . Toggled + = Icon_Toggled ;
_appToggle . Toggled + = App_Toggled ;
_developerToggle . Toggled + = Developer_Toggled ;
_versionToggle . Toggled + = Version_Toggled ;
_timePlayedToggle . Toggled + = TimePlayed_Toggled ;
_lastPlayedToggle . Toggled + = LastPlayed_Toggled ;
_fileExtToggle . Toggled + = FileExt_Toggled ;
_fileSizeToggle . Toggled + = FileSize_Toggled ;
_pathToggle . Toggled + = Path_Toggled ;
2019-11-29 04:32:51 +00:00
_gameTable . Model = _tableStore = new ListStore (
2019-12-21 19:52:31 +00:00
typeof ( bool ) ,
typeof ( Gdk . Pixbuf ) ,
typeof ( string ) ,
typeof ( string ) ,
typeof ( string ) ,
typeof ( string ) ,
typeof ( string ) ,
typeof ( string ) ,
typeof ( string ) ,
2020-03-25 17:09:38 +00:00
typeof ( string ) ,
typeof ( BlitStruct < ApplicationControlProperty > ) ) ;
2019-12-21 19:52:31 +00:00
2021-01-08 08:14:13 +00:00
_tableStore . SetSortFunc ( 5 , SortHelper . TimePlayedSort ) ;
_tableStore . SetSortFunc ( 6 , SortHelper . LastPlayedSort ) ;
_tableStore . SetSortFunc ( 8 , SortHelper . FileSizeSort ) ;
2020-06-26 10:30:16 +00:00
int columnId = ConfigurationState . Instance . Ui . ColumnSort . SortColumnId ;
bool ascending = ConfigurationState . Instance . Ui . ColumnSort . SortAscending ;
_tableStore . SetSortColumnId ( columnId , ascending ? SortType . Ascending : SortType . Descending ) ;
2019-11-29 04:32:51 +00:00
2020-04-25 13:02:44 +00:00
_gameTable . EnableSearch = true ;
_gameTable . SearchColumn = 2 ;
2021-06-24 00:21:52 +00:00
_gameTable . SearchEqualFunc = ( model , col , key , iter ) = > ! ( ( string ) model . GetValue ( iter , col ) ) . Contains ( key , StringComparison . InvariantCultureIgnoreCase ) ;
2020-04-25 13:02:44 +00:00
2021-08-04 21:28:19 +00:00
_hideUi . Label = _hideUi . Label . Replace ( "SHOWUIKEY" , ConfigurationState . Instance . Hid . Hotkeys . Value . ShowUi . ToString ( ) ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
UpdateGameTable ( ) ;
2020-01-12 02:10:55 +00:00
2020-07-03 23:29:36 +00:00
ConfigurationState . Instance . Ui . GameDirs . Event + = ( sender , args ) = >
{
if ( args . OldValue ! = args . NewValue )
{
UpdateGameTable ( ) ;
}
} ;
2020-01-12 02:10:55 +00:00
Task . Run ( RefreshFirmwareLabel ) ;
2021-04-14 10:28:43 +00:00
InputManager = new InputManager ( new GTK3KeyboardDriver ( this ) , new SDL2GamepadDriver ( ) ) ;
2019-11-29 04:32:51 +00:00
}
2021-05-16 15:12:14 +00:00
private void UpdateIgnoreMissingServicesState ( object sender , ReactiveEventArgs < bool > args )
{
if ( _emulationContext ! = null )
{
_emulationContext . Configuration . IgnoreMissingServices = args . NewValue ;
}
}
private void UpdateAspectRatioState ( object sender , ReactiveEventArgs < AspectRatio > args )
{
if ( _emulationContext ! = null )
{
_emulationContext . Configuration . AspectRatio = args . NewValue ;
}
}
private void UpdateDockedModeState ( object sender , ReactiveEventArgs < bool > e )
{
if ( _emulationContext ! = null )
{
_emulationContext . System . ChangeDockedModeState ( e . NewValue ) ;
}
}
2021-12-23 16:33:56 +00:00
private void UpdateAudioVolumeState ( object sender , ReactiveEventArgs < float > e )
{
_emulationContext ? . SetVolume ( e . NewValue ) ;
}
2021-01-08 08:14:13 +00:00
private void WindowStateEvent_Changed ( object o , WindowStateEventArgs args )
2020-07-23 13:12:19 +00:00
{
_fullScreen . Label = args . Event . NewWindowState . HasFlag ( Gdk . WindowState . Fullscreen ) ? "Exit Fullscreen" : "Enter Fullscreen" ;
}
2021-06-23 21:44:09 +00:00
private void MainWindow_FocusOutEvent ( object o , FocusOutEventArgs args )
{
IsFocused = false ;
}
private void MainWindow_FocusInEvent ( object o , FocusInEventArgs args )
{
IsFocused = true ;
}
2019-11-29 04:32:51 +00:00
private void UpdateColumns ( )
2019-09-02 16:03:57 +00:00
{
2019-11-29 04:32:51 +00:00
foreach ( TreeViewColumn column in _gameTable . Columns )
{
_gameTable . RemoveColumn ( column ) ;
}
2019-09-02 16:03:57 +00:00
2019-11-29 04:32:51 +00:00
CellRendererToggle favToggle = new CellRendererToggle ( ) ;
favToggle . Toggled + = FavToggle_Toggled ;
2019-12-21 19:52:31 +00:00
if ( ConfigurationState . Instance . Ui . GuiColumns . FavColumn ) _gameTable . AppendColumn ( "Fav" , favToggle , "active" , 0 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . IconColumn ) _gameTable . AppendColumn ( "Icon" , new CellRendererPixbuf ( ) , "pixbuf" , 1 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . AppColumn ) _gameTable . AppendColumn ( "Application" , new CellRendererText ( ) , "text" , 2 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . DevColumn ) _gameTable . AppendColumn ( "Developer" , new CellRendererText ( ) , "text" , 3 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . VersionColumn ) _gameTable . AppendColumn ( "Version" , new CellRendererText ( ) , "text" , 4 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . TimePlayedColumn ) _gameTable . AppendColumn ( "Time Played" , new CellRendererText ( ) , "text" , 5 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . LastPlayedColumn ) _gameTable . AppendColumn ( "Last Played" , new CellRendererText ( ) , "text" , 6 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . FileExtColumn ) _gameTable . AppendColumn ( "File Ext" , new CellRendererText ( ) , "text" , 7 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . FileSizeColumn ) _gameTable . AppendColumn ( "File Size" , new CellRendererText ( ) , "text" , 8 ) ;
if ( ConfigurationState . Instance . Ui . GuiColumns . PathColumn ) _gameTable . AppendColumn ( "Path" , new CellRendererText ( ) , "text" , 9 ) ;
2019-11-29 04:32:51 +00:00
foreach ( TreeViewColumn column in _gameTable . Columns )
2019-09-02 16:03:57 +00:00
{
2020-06-26 10:30:16 +00:00
switch ( column . Title )
{
case "Fav" :
column . SortColumnId = 0 ;
column . Clicked + = Column_Clicked ;
break ;
case "Application" :
column . SortColumnId = 2 ;
column . Clicked + = Column_Clicked ;
break ;
case "Developer" :
column . SortColumnId = 3 ;
column . Clicked + = Column_Clicked ;
break ;
case "Version" :
column . SortColumnId = 4 ;
column . Clicked + = Column_Clicked ;
break ;
case "Time Played" :
column . SortColumnId = 5 ;
column . Clicked + = Column_Clicked ;
break ;
case "Last Played" :
column . SortColumnId = 6 ;
column . Clicked + = Column_Clicked ;
break ;
case "File Ext" :
column . SortColumnId = 7 ;
column . Clicked + = Column_Clicked ;
break ;
case "File Size" :
column . SortColumnId = 8 ;
column . Clicked + = Column_Clicked ;
break ;
case "Path" :
column . SortColumnId = 9 ;
column . Clicked + = Column_Clicked ;
break ;
}
2019-09-02 16:03:57 +00:00
}
2019-12-21 19:52:31 +00:00
}
2021-04-14 10:28:43 +00:00
protected override void OnDestroyed ( )
{
InputManager . Dispose ( ) ;
}
2021-01-08 08:14:13 +00:00
private void InitializeSwitchInstance ( )
2019-12-21 19:52:31 +00:00
{
2021-08-12 21:56:24 +00:00
_virtualFileSystem . ReloadKeySet ( ) ;
2020-01-21 22:23:11 +00:00
2021-05-04 16:19:04 +00:00
IRenderer renderer ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
if ( ConfigurationState . Instance . Graphics . GraphicsBackend = = GraphicsBackend . Vulkan )
2021-05-04 16:19:04 +00:00
{
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
string preferredGpu = ConfigurationState . Instance . Graphics . PreferredGpu . Value ;
renderer = new VulkanRenderer ( CreateVulkanSurface , VulkanHelper . GetRequiredInstanceExtensions , preferredGpu ) ;
2021-05-04 16:19:04 +00:00
}
else
{
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
renderer = new OpenGLRenderer ( ) ;
2021-05-04 16:19:04 +00:00
}
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501)
* Initial Implementation
About as fast as nvidia GL multithreading, can be improved with faster command queuing.
* Struct based command list
Speeds up a bit. Still a lot of time lost to resource copy.
* Do shader init while the render thread is active.
* Introduce circular span pool V1
Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next.
* Refactor SpanRef some more
Use a struct to represent SpanRef, rather than a reference.
* Flush buffers on background thread
* Use a span for UpdateRenderScale.
Much faster than copying the array.
* Calculate command size using reflection
* WIP parallel shaders
* Some minor optimisation
* Only 2 max refs per command now.
The command with 3 refs is gone. :relieved:
* Don't cast on the GPU side
* Remove redundant casts, force sync on window present
* Fix Shader Cache
* Fix host shader save.
* Fixup to work with new renderer stuff
* Make command Run static, use array of delegates as lookup
Profile says this takes less time than the previous way.
* Bring up to date
* Add settings toggle. Fix Muiltithreading Off mode.
* Fix warning.
* Release tracking lock for flushes
* Fix Conditional Render fast path with threaded gal
* Make handle iteration safe when releasing the lock
This is mostly temporary.
* Attempt to set backend threading on driver
Only really works on nvidia before launching a game.
* Fix race condition with BufferModifiedRangeList, exceptions in tracking actions
* Update buffer set commands
* Some cleanup
* Only use stutter workaround when using opengl renderer non-threaded
* Add host-conditional reservation of counter events
There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made.
* Address Feedback
* Make counter flush tracked again.
Hopefully does not cause any issues this time.
* Wait for FlushTo on the main queue thread.
Currently assumes only one thread will want to FlushTo (in this case, the GPU thread)
* Add SDL2 headless integration
* Add HLE macro commands.
Co-authored-by: Mary <mary@mary.zone>
2021-08-26 22:31:29 +00:00
BackendThreading threadingMode = ConfigurationState . Instance . Graphics . BackendThreading ;
bool threadedGAL = threadingMode = = BackendThreading . On | | ( threadingMode = = BackendThreading . Auto & & renderer . PreferThreading ) ;
if ( threadedGAL )
{
renderer = new ThreadedRenderer ( renderer ) ;
}
Logger . Info ? . PrintMsg ( LogClass . Gpu , $"Backend Threading ({threadingMode}): {threadedGAL}" ) ;
2021-02-26 00:11:56 +00:00
IHardwareDeviceDriver deviceDriver = new DummyHardwareDeviceDriver ( ) ;
2021-01-08 08:14:13 +00:00
2021-05-05 21:37:09 +00:00
if ( ConfigurationState . Instance . System . AudioBackend . Value = = AudioBackend . SDL2 )
{
if ( SDL2HardwareDeviceDriver . IsSupported )
{
deviceDriver = new SDL2HardwareDeviceDriver ( ) ;
}
else
{
2021-08-26 21:03:19 +00:00
Logger . Warning ? . Print ( LogClass . Audio , "SDL2 is not supported, trying to fall back to OpenAL." ) ;
2021-12-26 14:17:13 +00:00
2021-08-26 21:03:19 +00:00
if ( OpenALHardwareDeviceDriver . IsSupported )
{
Logger . Warning ? . Print ( LogClass . Audio , "Found OpenAL, changing configuration." ) ;
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . OpenAl ;
SaveConfig ( ) ;
deviceDriver = new OpenALHardwareDeviceDriver ( ) ;
}
else
{
Logger . Warning ? . Print ( LogClass . Audio , "OpenAL is not supported, trying to fall back to SoundIO." ) ;
2021-12-26 14:17:13 +00:00
2021-08-26 21:03:19 +00:00
if ( SoundIoHardwareDeviceDriver . IsSupported )
{
Logger . Warning ? . Print ( LogClass . Audio , "Found SoundIO, changing configuration." ) ;
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . SoundIo ;
SaveConfig ( ) ;
deviceDriver = new SoundIoHardwareDeviceDriver ( ) ;
}
else
{
Logger . Warning ? . Print ( LogClass . Audio , "SoundIO is not supported, falling back to dummy audio out." ) ;
2021-12-26 14:17:13 +00:00
}
}
2021-05-05 21:37:09 +00:00
}
}
else if ( ConfigurationState . Instance . System . AudioBackend . Value = = AudioBackend . SoundIo )
2021-01-08 08:14:13 +00:00
{
2021-02-26 00:11:56 +00:00
if ( SoundIoHardwareDeviceDriver . IsSupported )
2021-01-08 08:14:13 +00:00
{
2021-02-26 00:11:56 +00:00
deviceDriver = new SoundIoHardwareDeviceDriver ( ) ;
2021-01-08 08:14:13 +00:00
}
else
{
2021-08-26 21:03:19 +00:00
Logger . Warning ? . Print ( LogClass . Audio , "SoundIO is not supported, trying to fall back to SDL2." ) ;
if ( SDL2HardwareDeviceDriver . IsSupported )
{
Logger . Warning ? . Print ( LogClass . Audio , "Found SDL2, changing configuration." ) ;
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . SDL2 ;
SaveConfig ( ) ;
deviceDriver = new SDL2HardwareDeviceDriver ( ) ;
}
else
{
Logger . Warning ? . Print ( LogClass . Audio , "SDL2 is not supported, trying to fall back to OpenAL." ) ;
if ( OpenALHardwareDeviceDriver . IsSupported )
{
Logger . Warning ? . Print ( LogClass . Audio , "Found OpenAL, changing configuration." ) ;
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . OpenAl ;
SaveConfig ( ) ;
deviceDriver = new OpenALHardwareDeviceDriver ( ) ;
}
else
{
Logger . Warning ? . Print ( LogClass . Audio , "OpenAL is not supported, falling back to dummy audio out." ) ;
2021-12-26 14:17:13 +00:00
}
2021-08-26 21:03:19 +00:00
}
2021-01-08 08:14:13 +00:00
}
}
else if ( ConfigurationState . Instance . System . AudioBackend . Value = = AudioBackend . OpenAl )
{
2021-02-26 00:11:56 +00:00
if ( OpenALHardwareDeviceDriver . IsSupported )
2021-01-08 08:14:13 +00:00
{
2021-02-26 00:11:56 +00:00
deviceDriver = new OpenALHardwareDeviceDriver ( ) ;
2021-01-08 08:14:13 +00:00
}
else
{
2021-08-26 21:03:19 +00:00
Logger . Warning ? . Print ( LogClass . Audio , "OpenAL is not supported, trying to fall back to SDL2." ) ;
2021-01-08 08:14:13 +00:00
2021-08-26 21:03:19 +00:00
if ( SDL2HardwareDeviceDriver . IsSupported )
2021-01-08 08:14:13 +00:00
{
2021-08-26 21:03:19 +00:00
Logger . Warning ? . Print ( LogClass . Audio , "Found SDL2, changing configuration." ) ;
2021-01-08 08:14:13 +00:00
2021-08-26 21:03:19 +00:00
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . SDL2 ;
2021-01-08 08:14:13 +00:00
SaveConfig ( ) ;
2021-08-26 21:03:19 +00:00
deviceDriver = new SDL2HardwareDeviceDriver ( ) ;
2021-01-08 08:14:13 +00:00
}
else
{
2021-08-26 21:03:19 +00:00
Logger . Warning ? . Print ( LogClass . Audio , "SDL2 is not supported, trying to fall back to SoundIO." ) ;
2021-12-26 14:17:13 +00:00
2021-08-26 21:03:19 +00:00
if ( SoundIoHardwareDeviceDriver . IsSupported )
{
Logger . Warning ? . Print ( LogClass . Audio , "Found SoundIO, changing configuration." ) ;
ConfigurationState . Instance . System . AudioBackend . Value = AudioBackend . SoundIo ;
SaveConfig ( ) ;
deviceDriver = new SoundIoHardwareDeviceDriver ( ) ;
}
else
{
Logger . Warning ? . Print ( LogClass . Audio , "SoundIO is not supported, falling back to dummy audio out." ) ;
}
2021-01-08 08:14:13 +00:00
}
}
}
2021-04-04 12:06:59 +00:00
var memoryConfiguration = ConfigurationState . Instance . System . ExpandRam . Value
2022-11-16 22:27:42 +00:00
? HLE . MemoryConfiguration . MemoryConfiguration6GiB
: HLE . MemoryConfiguration . MemoryConfiguration4GiB ;
2021-04-04 12:06:59 +00:00
2021-05-16 15:12:14 +00:00
IntegrityCheckLevel fsIntegrityCheckLevel = ConfigurationState . Instance . System . EnableFsIntegrityChecks ? IntegrityCheckLevel . ErrorOnInvalid : IntegrityCheckLevel . None ;
HLE . HLEConfiguration configuration = new HLE . HLEConfiguration ( _virtualFileSystem ,
2021-08-12 21:56:24 +00:00
_libHacHorizonManager ,
2021-05-16 15:12:14 +00:00
_contentManager ,
_accountManager ,
_userChannelPersistence ,
renderer ,
deviceDriver ,
memoryConfiguration ,
_uiHandler ,
( SystemLanguage ) ConfigurationState . Instance . System . Language . Value ,
( RegionCode ) ConfigurationState . Instance . System . Region . Value ,
ConfigurationState . Instance . Graphics . EnableVsync ,
ConfigurationState . Instance . System . EnableDockedMode ,
ConfigurationState . Instance . System . EnablePtc ,
2021-12-26 14:17:13 +00:00
ConfigurationState . Instance . System . EnableInternetAccess ,
2021-05-16 15:12:14 +00:00
fsIntegrityCheckLevel ,
ConfigurationState . Instance . System . FsGlobalAccessLogMode ,
ConfigurationState . Instance . System . SystemTimeOffset ,
ConfigurationState . Instance . System . TimeZone ,
POWER - Performance Optimizations With Extensive Ramifications (#2286)
* Refactoring of KMemoryManager class
* Replace some trivial uses of DRAM address with VA
* Get rid of GetDramAddressFromVa
* Abstracting more operations on derived page table class
* Run auto-format on KPageTableBase
* Managed to make TryConvertVaToPa private, few uses remains now
* Implement guest physical pages ref counting, remove manual freeing
* Make DoMmuOperation private and call new abstract methods only from the base class
* Pass pages count rather than size on Map/UnmapMemory
* Change memory managers to take host pointers
* Fix a guest memory leak and simplify KPageTable
* Expose new methods for host range query and mapping
* Some refactoring of MapPagesFromClientProcess to allow proper page ref counting and mapping without KPageLists
* Remove more uses of AddVaRangeToPageList, now only one remains (shared memory page checking)
* Add a SharedMemoryStorage class, will be useful for host mapping
* Sayonara AddVaRangeToPageList, you served us well
* Start to implement host memory mapping (WIP)
* Support memory tracking through host exception handling
* Fix some access violations from HLE service guest memory access and CPU
* Fix memory tracking
* Fix mapping list bugs, including a race and a error adding mapping ranges
* Simple page table for memory tracking
* Simple "volatile" region handle mode
* Update UBOs directly (experimental, rough)
* Fix the overlap check
* Only set non-modified buffers as volatile
* Fix some memory tracking issues
* Fix possible race in MapBufferFromClientProcess (block list updates were not locked)
* Write uniform update to memory immediately, only defer the buffer set.
* Fix some memory tracking issues
* Pass correct pages count on shared memory unmap
* Armeilleure Signal Handler v1 + Unix changes
Unix currently behaves like windows, rather than remapping physical
* Actually check if the host platform is unix
* Fix decommit on linux.
* Implement windows 10 placeholder shared memory, fix a buffer issue.
* Make PTC version something that will never match with master
* Remove testing variable for block count
* Add reference count for memory manager, fix dispose
Can still deadlock with OpenAL
* Add address validation, use page table for mapped check, add docs
Might clean up the page table traversing routines.
* Implement batched mapping/tracking.
* Move documentation, fix tests.
* Cleanup uniform buffer update stuff.
* Remove unnecessary assignment.
* Add unsafe host mapped memory switch
On by default. Would be good to turn this off for untrusted code (homebrew, exefs mods) and give the user the option to turn it on manually, though that requires some UI work.
* Remove C# exception handlers
They have issues due to current .NET limitations, so the meilleure one fully replaces them for now.
* Fix MapPhysicalMemory on the software MemoryManager.
* Null check for GetHostAddress, docs
* Add configuration for setting memory manager mode (not in UI yet)
* Add config to UI
* Fix type mismatch on Unix signal handler code emit
* Fix 6GB DRAM mode.
The size can be greater than `uint.MaxValue` when the DRAM is >4GB.
* Address some feedback.
* More detailed error if backing memory cannot be mapped.
* SetLastError on all OS functions for consistency
* Force pages dirty with UBO update instead of setting them directly.
Seems to be much faster across a few games. Need retesting.
* Rebase, configuration rework, fix mem tracking regression
* Fix race in FreePages
* Set memory managers null after decrementing ref count
* Remove readonly keyword, as this is now modified.
* Use a local variable for the signal handler rather than a register.
* Fix bug with buffer resize, and index/uniform buffer binding.
Should fix flickering in games.
* Add InvalidAccessHandler to MemoryTracking
Doesn't do anything yet
* Call invalid access handler on unmapped read/write.
Same rules as the regular memory manager.
* Make unsafe mapped memory its own MemoryManagerType
* Move FlushUboDirty into UpdateState.
* Buffer dirty cache, rather than ubo cache
Much cleaner, may be reusable for Inline2Memory updates.
* This doesn't return anything anymore.
* Add sigaction remove methods, correct a few function signatures.
* Return empty list of physical regions for size 0.
* Also on AddressSpaceManager
Co-authored-by: gdkchan <gab.dark.100@gmail.com>
2021-05-24 20:52:44 +00:00
ConfigurationState . Instance . System . MemoryManagerMode ,
2021-05-16 15:12:14 +00:00
ConfigurationState . Instance . System . IgnoreMissingServices ,
2021-12-23 16:33:56 +00:00
ConfigurationState . Instance . Graphics . AspectRatio ,
2023-01-29 11:37:52 +00:00
ConfigurationState . Instance . System . AudioVolume ,
ConfigurationState . Instance . System . UseHypervisor ) ;
2021-05-16 15:12:14 +00:00
_emulationContext = new HLE . Switch ( configuration ) ;
2019-09-02 16:03:57 +00:00
}
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
private SurfaceKHR CreateVulkanSurface ( Instance instance , Vk vk )
{
return new SurfaceKHR ( ( ulong ) ( ( VKRenderer ) RendererWidget ) . CreateWindowSurface ( instance . Handle ) ) ;
}
2021-03-03 00:39:36 +00:00
private void SetupProgressUiHandlers ( )
{
2023-03-31 19:16:46 +00:00
if ( _emulationContext . Processes . ActiveApplication . DiskCacheLoadState ! = null )
2023-01-04 23:01:44 +00:00
{
2023-03-31 19:16:46 +00:00
_emulationContext . Processes . ActiveApplication . DiskCacheLoadState . StateChanged - = ProgressHandler ;
_emulationContext . Processes . ActiveApplication . DiskCacheLoadState . StateChanged + = ProgressHandler ;
2023-01-04 23:01:44 +00:00
}
2021-03-03 00:39:36 +00:00
2021-03-22 18:40:07 +00:00
_emulationContext . Gpu . ShaderCacheStateChanged - = ProgressHandler ;
_emulationContext . Gpu . ShaderCacheStateChanged + = ProgressHandler ;
2021-03-03 00:39:36 +00:00
}
2021-03-22 18:40:07 +00:00
private void ProgressHandler < T > ( T state , int current , int total ) where T : Enum
2021-03-03 00:39:36 +00:00
{
2021-03-22 18:40:07 +00:00
bool visible ;
string label ;
2021-03-03 00:39:36 +00:00
2021-03-22 18:40:07 +00:00
switch ( state )
2021-03-03 00:39:36 +00:00
{
2023-01-04 23:01:44 +00:00
case LoadState ptcState :
visible = ptcState ! = LoadState . Loaded ;
2021-03-22 18:40:07 +00:00
label = $"PTC : {current}/{total}" ;
break ;
case ShaderCacheLoadingState shaderCacheState :
visible = shaderCacheState ! = ShaderCacheLoadingState . Loaded ;
label = $"Shaders : {current}/{total}" ;
break ;
default :
throw new ArgumentException ( $"Unknown Progress Handler type {typeof(T)}" ) ;
}
2021-03-03 00:39:36 +00:00
Application . Invoke ( delegate
{
2021-03-22 18:40:07 +00:00
_loadingStatusLabel . Text = label ;
_loadingStatusBar . Fraction = total > 0 ? ( double ) current / total : 0 ;
_loadingStatusBar . Visible = visible ;
_loadingStatusLabel . Visible = visible ;
2021-03-03 00:39:36 +00:00
} ) ;
}
2021-01-08 08:14:13 +00:00
public void UpdateGameTable ( )
2019-09-02 16:03:57 +00:00
{
2020-07-03 23:29:36 +00:00
if ( _updatingGameTable | | _gameLoaded )
2019-09-02 16:03:57 +00:00
{
2019-11-29 04:32:51 +00:00
return ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
_updatingGameTable = true ;
_tableStore . Clear ( ) ;
2020-01-31 18:21:46 +00:00
Thread applicationLibraryThread = new Thread ( ( ) = >
{
2021-01-08 08:14:13 +00:00
_applicationLibrary . LoadApplications ( ConfigurationState . Instance . Ui . GameDirs , ConfigurationState . Instance . System . Language ) ;
2019-11-29 04:32:51 +00:00
2020-01-31 18:21:46 +00:00
_updatingGameTable = false ;
} ) ;
2021-01-08 08:14:13 +00:00
applicationLibraryThread . Name = "GUI.ApplicationLibraryThread" ;
2020-01-31 18:21:46 +00:00
applicationLibraryThread . IsBackground = true ;
applicationLibraryThread . Start ( ) ;
2019-09-02 16:03:57 +00:00
}
2021-01-08 08:14:13 +00:00
[Conditional("RELEASE")]
public void PerformanceCheck ( )
2019-09-02 16:03:57 +00:00
{
2022-02-18 00:08:07 +00:00
if ( ConfigurationState . Instance . Logger . EnableTrace . Value )
2019-09-02 16:03:57 +00:00
{
2021-01-08 08:14:13 +00:00
MessageDialog debugWarningDialog = new MessageDialog ( this , DialogFlags . Modal , MessageType . Warning , ButtonsType . YesNo , null )
{
Title = "Ryujinx - Warning" ,
2022-02-18 00:08:07 +00:00
Text = "You have trace logging enabled, which is designed to be used by developers only." ,
SecondaryText = "For optimal performance, it's recommended to disable trace logging. Would you like to disable trace logging now?"
2021-01-08 08:14:13 +00:00
} ;
if ( debugWarningDialog . Run ( ) = = ( int ) ResponseType . Yes )
{
2022-02-18 00:08:07 +00:00
ConfigurationState . Instance . Logger . EnableTrace . Value = false ;
2021-01-08 08:14:13 +00:00
SaveConfig ( ) ;
}
debugWarningDialog . Dispose ( ) ;
2019-09-02 16:03:57 +00:00
}
2021-01-08 08:14:13 +00:00
if ( ! string . IsNullOrWhiteSpace ( ConfigurationState . Instance . Graphics . ShadersDumpPath . Value ) )
2019-09-02 16:03:57 +00:00
{
2021-01-08 08:14:13 +00:00
MessageDialog shadersDumpWarningDialog = new MessageDialog ( this , DialogFlags . Modal , MessageType . Warning , ButtonsType . YesNo , null )
2020-03-30 21:39:46 +00:00
{
2021-01-08 08:14:13 +00:00
Title = "Ryujinx - Warning" ,
Text = "You have shader dumping enabled, which is designed to be used by developers only." ,
SecondaryText = "For optimal performance, it's recommended to disable shader dumping. Would you like to disable shader dumping now?"
} ;
2020-03-30 21:39:46 +00:00
2021-01-08 08:14:13 +00:00
if ( shadersDumpWarningDialog . Run ( ) = = ( int ) ResponseType . Yes )
{
ConfigurationState . Instance . Graphics . ShadersDumpPath . Value = "" ;
SaveConfig ( ) ;
2020-03-30 21:39:46 +00:00
}
2021-01-08 08:14:13 +00:00
shadersDumpWarningDialog . Dispose ( ) ;
}
}
2020-03-30 21:39:46 +00:00
2023-03-31 19:16:46 +00:00
private bool LoadApplication ( string path , bool isFirmwareTitle )
2021-01-08 08:14:13 +00:00
{
2023-03-31 19:16:46 +00:00
SystemVersion firmwareVersion = _contentManager . GetCurrentFirmwareVersion ( ) ;
if ( ! SetupValidator . CanStartApplication ( _contentManager , path , out UserError userError ) )
2021-01-08 08:14:13 +00:00
{
2023-03-31 19:16:46 +00:00
if ( SetupValidator . CanFixStartApplication ( _contentManager , path , userError , out firmwareVersion ) )
{
string message = $"Would you like to install the firmware embedded in this game? (Firmware {firmwareVersion.VersionString})" ;
2020-03-30 21:39:46 +00:00
2023-03-31 19:16:46 +00:00
ResponseType responseDialog = ( ResponseType ) GtkDialog . CreateConfirmationDialog ( "No Firmware Installed" , message ) . Run ( ) ;
2019-09-19 23:59:48 +00:00
2023-03-31 19:16:46 +00:00
if ( responseDialog ! = ResponseType . Yes | | ! SetupValidator . TryFixStartApplication ( _contentManager , path , userError , out _ ) )
{
UserErrorDialog . CreateUserErrorDialog ( userError ) ;
2021-05-04 16:19:04 +00:00
2023-03-31 19:16:46 +00:00
return false ;
}
2021-05-04 16:19:04 +00:00
2023-03-31 19:16:46 +00:00
// Tell the user that we installed a firmware for them.
2020-01-21 22:23:11 +00:00
2023-03-31 19:16:46 +00:00
firmwareVersion = _contentManager . GetCurrentFirmwareVersion ( ) ;
2019-12-21 19:52:31 +00:00
2023-03-31 19:16:46 +00:00
RefreshFirmwareLabel ( ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
message = $"No installed firmware was found but Ryujinx was able to install firmware {firmwareVersion.VersionString} from the provided game.\nThe emulator will now start." ;
2021-06-28 18:54:45 +00:00
2023-03-31 19:16:46 +00:00
GtkDialog . CreateInfoDialog ( $"Firmware {firmwareVersion.VersionString} was installed" , message ) ;
}
else
2021-06-28 18:54:45 +00:00
{
2023-03-31 19:16:46 +00:00
UserErrorDialog . CreateUserErrorDialog ( userError ) ;
2021-06-28 18:54:45 +00:00
2023-03-31 19:16:46 +00:00
return false ;
2021-06-28 18:54:45 +00:00
}
2023-03-31 19:16:46 +00:00
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
Logger . Notice . Print ( LogClass . Application , $"Using Firmware Version: {firmwareVersion?.VersionString}" ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
if ( isFirmwareTitle )
{
Logger . Info ? . Print ( LogClass . Application , "Loading as Firmware Title (NCA)." ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
return _emulationContext . LoadNca ( path ) ;
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
if ( Directory . Exists ( path ) )
{
string [ ] romFsFiles = Directory . GetFiles ( path , "*.istorage" ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
if ( romFsFiles . Length = = 0 )
{
romFsFiles = Directory . GetFiles ( path , "*.romfs" ) ;
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
if ( romFsFiles . Length > 0 )
{
Logger . Info ? . Print ( LogClass . Application , "Loading as cart with RomFS." ) ;
return _emulationContext . LoadCart ( path , romFsFiles [ 0 ] ) ;
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
Logger . Info ? . Print ( LogClass . Application , "Loading as cart WITHOUT RomFS." ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
return _emulationContext . LoadCart ( path ) ;
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
if ( File . Exists ( path ) )
{
switch ( System . IO . Path . GetExtension ( path ) . ToLowerInvariant ( ) )
{
case ".xci" :
Logger . Info ? . Print ( LogClass . Application , "Loading as XCI." ) ;
return _emulationContext . LoadXci ( path ) ;
case ".nca" :
Logger . Info ? . Print ( LogClass . Application , "Loading as NCA." ) ;
return _emulationContext . LoadNca ( path ) ;
case ".nsp" :
case ".pfs0" :
Logger . Info ? . Print ( LogClass . Application , "Loading as NSP." ) ;
return _emulationContext . LoadNsp ( path ) ;
default :
Logger . Info ? . Print ( LogClass . Application , "Loading as Homebrew." ) ;
try
{
return _emulationContext . LoadProgram ( path ) ;
}
catch ( ArgumentOutOfRangeException )
2020-09-01 09:09:42 +00:00
{
2023-03-31 19:16:46 +00:00
Logger . Error ? . Print ( LogClass . Application , "The specified file is not supported by Ryujinx." ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
return false ;
}
}
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
Logger . Warning ? . Print ( LogClass . Application , "Please specify a valid XCI/NCA/NSP/PFS0/NRO file." ) ;
2021-01-08 08:14:13 +00:00
2023-03-31 19:16:46 +00:00
return false ;
}
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
public void RunApplication ( string path , bool startFullscreen = false )
{
if ( _gameLoaded )
{
GtkDialog . CreateInfoDialog ( "A game has already been loaded" , "Please stop emulation or close the emulator before launching another game." ) ;
}
else
{
PerformanceCheck ( ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
Logger . RestartTime ( ) ;
2020-09-01 09:09:42 +00:00
2023-03-31 19:16:46 +00:00
RendererWidget = CreateRendererWidget ( ) ;
2020-05-05 11:51:04 +00:00
2023-03-31 19:16:46 +00:00
SwitchToRenderWidget ( startFullscreen ) ;
2021-06-28 18:54:45 +00:00
2023-03-31 19:16:46 +00:00
InitializeSwitchInstance ( ) ;
2019-09-02 16:03:57 +00:00
2023-03-31 19:16:46 +00:00
UpdateGraphicsConfig ( ) ;
2019-09-02 16:03:57 +00:00
2023-03-31 19:16:46 +00:00
bool isFirmwareTitle = false ;
if ( path . StartsWith ( "@SystemContent" ) )
2019-09-02 16:03:57 +00:00
{
2023-03-31 19:16:46 +00:00
path = _virtualFileSystem . SwitchPathToSystemPath ( path ) ;
isFirmwareTitle = true ;
2019-09-02 16:03:57 +00:00
}
2021-01-08 08:14:13 +00:00
2023-03-31 19:16:46 +00:00
if ( ! LoadApplication ( path , isFirmwareTitle ) )
{
2021-01-08 08:14:13 +00:00
_emulationContext . Dispose ( ) ;
2023-03-31 19:16:46 +00:00
SwitchToGameTable ( ) ;
2020-03-29 03:25:54 +00:00
return ;
2019-09-02 16:03:57 +00:00
}
2023-01-04 23:01:44 +00:00
SetupProgressUiHandlers ( ) ;
2021-01-08 08:14:13 +00:00
_currentEmulatedGamePath = path ;
2020-01-21 22:23:11 +00:00
2020-02-13 17:43:29 +00:00
_deviceExitStatus . Reset ( ) ;
2020-02-06 11:38:24 +00:00
2020-11-17 21:40:19 +00:00
Translator . IsReadyForTranslation . Reset ( ) ;
2022-12-06 22:00:25 +00:00
2023-03-31 19:16:46 +00:00
Thread windowThread = new ( CreateGameWindow )
2020-11-17 21:40:19 +00:00
{
Name = "GUI.WindowThread"
} ;
windowThread . Start ( ) ;
2021-03-18 20:40:20 +00:00
_gameLoaded = true ;
_actionMenu . Sensitive = true ;
2022-12-04 19:17:11 +00:00
UpdateMenuItem . Sensitive = false ;
2021-03-18 20:40:20 +00:00
_lastScannedAmiiboId = "" ;
2019-09-02 16:03:57 +00:00
2020-01-12 02:10:55 +00:00
_firmwareInstallFile . Sensitive = false ;
_firmwareInstallDirectory . Sensitive = false ;
2023-03-31 19:16:46 +00:00
DiscordIntegrationModule . SwitchToPlayingState ( _emulationContext . Processes . ActiveApplication . ProgramIdText ,
_emulationContext . Processes . ActiveApplication . ApplicationControlProperties . Title [ ( int ) _emulationContext . System . State . DesiredTitleLanguage ] . NameString . ToString ( ) ) ;
2019-09-02 16:03:57 +00:00
2023-03-31 19:16:46 +00:00
_applicationLibrary . LoadAndSaveMetaData ( _emulationContext . Processes . ActiveApplication . ProgramIdText , appMetadata = >
2019-09-02 16:03:57 +00:00
{
2020-01-12 03:01:04 +00:00
appMetadata . LastPlayed = DateTime . UtcNow . ToString ( ) ;
} ) ;
2019-09-02 16:03:57 +00:00
}
}
2021-05-04 16:19:04 +00:00
private RendererWidgetBase CreateRendererWidget ( )
2019-09-02 16:03:57 +00:00
{
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
if ( ConfigurationState . Instance . Graphics . GraphicsBackend = = GraphicsBackend . Vulkan )
2020-08-18 01:49:37 +00:00
{
2021-05-04 16:19:04 +00:00
return new VKRenderer ( InputManager , ConfigurationState . Instance . Logger . GraphicsDebugLevel ) ;
2020-08-18 01:49:37 +00:00
}
2021-05-04 16:19:04 +00:00
else
{
return new GlRenderer ( InputManager , ConfigurationState . Instance . Logger . GraphicsDebugLevel ) ;
}
}
2020-08-18 01:49:37 +00:00
2021-07-06 18:04:21 +00:00
private void SwitchToRenderWidget ( bool startFullscreen = false )
2021-05-04 16:19:04 +00:00
{
_viewBox . Remove ( _gameTableWindow ) ;
RendererWidget . Expand = true ;
_viewBox . Child = RendererWidget ;
2021-01-24 23:02:00 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . ShowAll ( ) ;
EditFooterForGameRenderer ( ) ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
if ( Window . State . HasFlag ( Gdk . WindowState . Fullscreen ) )
{
ToggleExtraWidgets ( false ) ;
}
2021-07-06 18:04:21 +00:00
else if ( startFullscreen | | ConfigurationState . Instance . Ui . StartFullscreen . Value )
2021-05-04 16:19:04 +00:00
{
FullScreen_Toggled ( null , null ) ;
}
}
private void SwitchToGameTable ( )
{
if ( Window . State . HasFlag ( Gdk . WindowState . Fullscreen ) )
2019-09-02 16:03:57 +00:00
{
2021-05-04 16:19:04 +00:00
ToggleExtraWidgets ( true ) ;
}
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . Exit ( ) ;
2020-07-23 13:12:19 +00:00
2021-05-04 16:19:04 +00:00
if ( RendererWidget . Window ! = Window & & RendererWidget . Window ! = null )
{
RendererWidget . Window . Dispose ( ) ;
}
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . Dispose ( ) ;
2020-02-12 00:56:19 +00:00
2021-12-04 23:02:30 +00:00
if ( OperatingSystem . IsWindows ( ) )
{
_windowsMultimediaTimerResolution ? . Dispose ( ) ;
_windowsMultimediaTimerResolution = null ;
}
2021-05-04 16:19:04 +00:00
DisplaySleep . Restore ( ) ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
_viewBox . Remove ( RendererWidget ) ;
_viewBox . Add ( _gameTableWindow ) ;
2020-06-16 18:28:02 +00:00
2021-05-04 16:19:04 +00:00
_gameTableWindow . Expand = true ;
2020-02-13 17:43:29 +00:00
2021-05-04 16:19:04 +00:00
Window . Title = $"Ryujinx {Program.Version}" ;
2020-07-23 13:12:19 +00:00
2021-05-04 16:19:04 +00:00
_emulationContext = null ;
_gameLoaded = false ;
RendererWidget = null ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
DiscordIntegrationModule . SwitchToMainMenu ( ) ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
RecreateFooterForMenu ( ) ;
2021-01-08 08:14:13 +00:00
2021-05-04 16:19:04 +00:00
UpdateColumns ( ) ;
UpdateGameTable ( ) ;
2020-02-13 17:43:29 +00:00
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
RefreshFirmwareLabel ( ) ;
HandleRelaunch ( ) ;
2021-05-04 16:19:04 +00:00
}
private void CreateGameWindow ( )
{
2021-12-04 23:02:30 +00:00
if ( OperatingSystem . IsWindows ( ) )
2021-05-04 16:19:04 +00:00
{
_windowsMultimediaTimerResolution = new WindowsMultimediaTimerResolution ( 1 ) ;
}
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
DisplaySleep . Prevent ( ) ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . Initialize ( _emulationContext ) ;
2020-02-13 17:43:29 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . WaitEvent . WaitOne ( ) ;
2020-02-13 17:43:29 +00:00
2021-05-04 16:19:04 +00:00
RendererWidget . Start ( ) ;
2020-02-12 00:56:19 +00:00
2021-05-04 16:19:04 +00:00
_emulationContext . Dispose ( ) ;
_deviceExitStatus . Set ( ) ;
2020-01-21 22:23:11 +00:00
2021-05-04 16:19:04 +00:00
// NOTE: Everything that is here will not be executed when you close the UI.
Application . Invoke ( delegate
{
SwitchToGameTable ( ) ;
2020-01-21 22:23:11 +00:00
} ) ;
2020-02-13 17:43:29 +00:00
}
private void RecreateFooterForMenu ( )
{
2020-03-07 02:40:06 +00:00
_listStatusBox . Show ( ) ;
_statusBar . Hide ( ) ;
2020-02-13 17:43:29 +00:00
}
2020-02-12 00:56:19 +00:00
2021-01-08 08:14:13 +00:00
private void EditFooterForGameRenderer ( )
2020-02-13 17:43:29 +00:00
{
2020-03-07 02:40:06 +00:00
_listStatusBox . Hide ( ) ;
_statusBar . Show ( ) ;
2020-02-12 00:56:19 +00:00
}
public void ToggleExtraWidgets ( bool show )
{
2021-05-04 16:19:04 +00:00
if ( RendererWidget ! = null )
2020-02-12 00:56:19 +00:00
{
if ( show )
{
_menuBar . ShowAll ( ) ;
2020-03-07 02:40:06 +00:00
_footerBox . Show ( ) ;
_statusBar . Show ( ) ;
2020-02-12 00:56:19 +00:00
}
else
{
_menuBar . Hide ( ) ;
_footerBox . Hide ( ) ;
}
}
2020-01-21 22:23:11 +00:00
}
2021-01-08 08:14:13 +00:00
private void UpdateGameMetadata ( string titleId )
2020-01-21 22:23:11 +00:00
{
if ( _gameLoaded )
{
2021-01-08 08:14:13 +00:00
_applicationLibrary . LoadAndSaveMetaData ( titleId , appMetadata = >
2020-01-21 22:23:11 +00:00
{
DateTime lastPlayedDateTime = DateTime . Parse ( appMetadata . LastPlayed ) ;
double sessionTimePlayed = DateTime . UtcNow . Subtract ( lastPlayedDateTime ) . TotalSeconds ;
2019-09-02 16:03:57 +00:00
2020-01-21 22:23:11 +00:00
appMetadata . TimePlayed + = Math . Round ( sessionTimePlayed , MidpointRounding . AwayFromZero ) ;
} ) ;
2019-09-02 16:03:57 +00:00
}
}
2021-01-08 08:14:13 +00:00
public void UpdateGraphicsConfig ( )
2020-07-07 02:41:07 +00:00
{
2020-12-16 02:19:07 +00:00
int resScale = ConfigurationState . Instance . Graphics . ResScale ;
2020-07-07 02:41:07 +00:00
float resScaleCustom = ConfigurationState . Instance . Graphics . ResScaleCustom ;
2020-12-16 02:19:07 +00:00
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
Graphics . Gpu . GraphicsConfig . ResScale = ( resScale = = - 1 ) ? resScaleCustom : resScale ;
Graphics . Gpu . GraphicsConfig . MaxAnisotropy = ConfigurationState . Instance . Graphics . MaxAnisotropy ;
Graphics . Gpu . GraphicsConfig . ShadersDumpPath = ConfigurationState . Instance . Graphics . ShadersDumpPath ;
Graphics . Gpu . GraphicsConfig . EnableShaderCache = ConfigurationState . Instance . Graphics . EnableShaderCache ;
Graphics . Gpu . GraphicsConfig . EnableTextureRecompression = ConfigurationState . Instance . Graphics . EnableTextureRecompression ;
2022-11-16 17:53:04 +00:00
Graphics . Gpu . GraphicsConfig . EnableMacroHLE = ConfigurationState . Instance . Graphics . EnableMacroHLE ;
2020-07-07 02:41:07 +00:00
}
2021-01-08 08:14:13 +00:00
public void SaveConfig ( )
2020-05-03 02:00:53 +00:00
{
2020-05-03 10:08:21 +00:00
ConfigurationState . Instance . ToFileFormat ( ) . SaveConfig ( Program . ConfigurationPath ) ;
2020-05-03 02:00:53 +00:00
}
2021-01-08 08:14:13 +00:00
private void End ( )
2019-09-02 16:03:57 +00:00
{
2019-11-29 04:32:51 +00:00
if ( _ending )
{
return ;
}
_ending = true ;
2021-01-08 08:14:13 +00:00
if ( _emulationContext ! = null )
2019-09-02 16:03:57 +00:00
{
2023-03-31 19:16:46 +00:00
UpdateGameMetadata ( _emulationContext . Processes . ActiveApplication . ProgramIdText ) ;
2020-02-06 11:38:24 +00:00
2021-05-04 16:19:04 +00:00
if ( RendererWidget ! = null )
2020-02-06 11:38:24 +00:00
{
2021-01-08 08:14:13 +00:00
// We tell the widget that we are exiting.
2021-05-04 16:19:04 +00:00
RendererWidget . Exit ( ) ;
2020-02-13 17:43:29 +00:00
// Wait for the other thread to dispose the HLE context before exiting.
_deviceExitStatus . WaitOne ( ) ;
2021-05-04 16:19:04 +00:00
RendererWidget . Dispose ( ) ;
2020-02-06 11:38:24 +00:00
}
2019-09-02 16:03:57 +00:00
}
2020-01-21 22:23:11 +00:00
Dispose ( ) ;
2021-01-08 08:14:13 +00:00
Program . Exit ( ) ;
2020-01-21 22:23:11 +00:00
Application . Quit ( ) ;
}
2021-01-08 08:14:13 +00:00
//
// Events
//
2019-12-22 02:49:51 +00:00
private void Application_Added ( object sender , ApplicationAddedEventArgs args )
2019-11-29 04:32:51 +00:00
{
Application . Invoke ( delegate
{
_tableStore . AppendValues (
2019-12-22 02:49:51 +00:00
args . AppData . Favorite ,
new Gdk . Pixbuf ( args . AppData . Icon , 75 , 75 ) ,
$"{args.AppData.TitleName}\n{args.AppData.TitleId.ToUpper()}" ,
args . AppData . Developer ,
args . AppData . Version ,
args . AppData . TimePlayed ,
args . AppData . LastPlayed ,
args . AppData . FileExtension ,
args . AppData . FileSize ,
2020-03-25 17:09:38 +00:00
args . AppData . Path ,
args . AppData . ControlHolder ) ;
2020-01-31 18:21:46 +00:00
} ) ;
}
2019-12-22 02:49:51 +00:00
2020-01-31 18:21:46 +00:00
private void ApplicationCount_Updated ( object sender , ApplicationCountUpdatedEventArgs args )
{
Application . Invoke ( delegate
{
2019-12-22 02:49:51 +00:00
_progressLabel . Text = $"{args.NumAppsLoaded}/{args.NumAppsFound} Games Loaded" ;
2020-01-31 18:21:46 +00:00
float barValue = 0 ;
if ( args . NumAppsFound ! = 0 )
{
barValue = ( float ) args . NumAppsLoaded / args . NumAppsFound ;
}
2021-03-03 00:39:36 +00:00
_progressBar . Fraction = barValue ;
2020-06-26 10:30:16 +00:00
2021-01-08 08:14:13 +00:00
// Reset the vertical scrollbar to the top when titles finish loading
if ( args . NumAppsLoaded = = args . NumAppsFound )
2020-06-26 10:30:16 +00:00
{
_gameTableWindow . Vadjustment . Value = 0 ;
}
2019-11-29 04:32:51 +00:00
} ) ;
}
2020-03-07 02:40:06 +00:00
private void Update_StatusBar ( object sender , StatusUpdatedEventArgs args )
{
Application . Invoke ( delegate
{
2021-12-23 16:33:56 +00:00
_gameStatus . Text = args . GameStatus ;
_fifoStatus . Text = args . FifoStatus ;
_gpuName . Text = args . GpuName ;
_dockedMode . Text = args . DockedMode ;
_aspectRatio . Text = args . AspectRatio ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
_gpuBackend . Text = args . GpuBackend ;
2021-12-23 16:33:56 +00:00
_volumeStatus . Text = GetVolumeLabelText ( args . Volume ) ;
2020-03-07 02:40:06 +00:00
if ( args . VSyncEnabled )
{
_vSyncStatus . Attributes = new Pango . AttrList ( ) ;
_vSyncStatus . Attributes . Insert ( new Pango . AttrForeground ( 11822 , 60138 , 51657 ) ) ;
}
else
{
_vSyncStatus . Attributes = new Pango . AttrList ( ) ;
_vSyncStatus . Attributes . Insert ( new Pango . AttrForeground ( ushort . MaxValue , 17733 , 21588 ) ) ;
}
} ) ;
}
2019-11-29 04:32:51 +00:00
private void FavToggle_Toggled ( object sender , ToggledArgs args )
{
_tableStore . GetIter ( out TreeIter treeIter , new TreePath ( args . Path ) ) ;
2021-01-08 08:14:13 +00:00
string titleId = _tableStore . GetValue ( treeIter , 2 ) . ToString ( ) . Split ( "\n" ) [ 1 ] . ToLower ( ) ;
bool newToggleValue = ! ( bool ) _tableStore . GetValue ( treeIter , 0 ) ;
2019-10-13 06:02:07 +00:00
2020-01-12 03:01:04 +00:00
_tableStore . SetValue ( treeIter , 0 , newToggleValue ) ;
2019-11-29 04:32:51 +00:00
2021-01-08 08:14:13 +00:00
_applicationLibrary . LoadAndSaveMetaData ( titleId , appMetadata = >
2019-11-29 04:32:51 +00:00
{
2020-01-12 03:01:04 +00:00
appMetadata . Favorite = newToggleValue ;
} ) ;
2019-11-29 04:32:51 +00:00
}
2020-06-26 10:30:16 +00:00
private void Column_Clicked ( object sender , EventArgs args )
{
TreeViewColumn column = ( TreeViewColumn ) sender ;
ConfigurationState . Instance . Ui . ColumnSort . SortColumnId . Value = column . SortColumnId ;
ConfigurationState . Instance . Ui . ColumnSort . SortAscending . Value = column . SortOrder = = SortType . Ascending ;
SaveConfig ( ) ;
}
2019-11-29 04:32:51 +00:00
private void Row_Activated ( object sender , RowActivatedArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-22 02:49:51 +00:00
_gameTableSelection . GetSelected ( out TreeIter treeIter ) ;
2021-01-08 08:14:13 +00:00
2019-11-29 04:32:51 +00:00
string path = ( string ) _tableStore . GetValue ( treeIter , 9 ) ;
2019-09-02 16:03:57 +00:00
2023-03-31 19:16:46 +00:00
RunApplication ( path ) ;
2019-09-02 16:03:57 +00:00
}
2020-11-19 00:34:28 +00:00
private void VSyncStatus_Clicked ( object sender , ButtonReleaseEventArgs args )
{
_emulationContext . EnableDeviceVsync = ! _emulationContext . EnableDeviceVsync ;
2021-03-18 23:09:33 +00:00
Logger . Info ? . Print ( LogClass . Application , $"VSync toggled to: {_emulationContext.EnableDeviceVsync}" ) ;
2020-11-19 00:34:28 +00:00
}
private void DockedMode_Clicked ( object sender , ButtonReleaseEventArgs args )
{
ConfigurationState . Instance . System . EnableDockedMode . Value = ! ConfigurationState . Instance . System . EnableDockedMode . Value ;
}
2021-12-23 16:33:56 +00:00
private string GetVolumeLabelText ( float volume )
{
string icon = volume = = 0 ? "🔇" : "🔊" ;
return $"{icon} {(int)(volume * 100)}%" ;
}
private void VolumeStatus_Clicked ( object sender , ButtonReleaseEventArgs args )
{
if ( _emulationContext ! = null )
{
if ( _emulationContext . IsAudioMuted ( ) )
{
_emulationContext . SetVolume ( ConfigurationState . Instance . System . AudioVolume ) ;
}
else
{
_emulationContext . SetVolume ( 0 ) ;
}
}
}
2020-12-16 02:19:07 +00:00
private void AspectRatio_Clicked ( object sender , ButtonReleaseEventArgs args )
{
AspectRatio aspectRatio = ConfigurationState . Instance . Graphics . AspectRatio . Value ;
2022-02-13 13:50:07 +00:00
ConfigurationState . Instance . Graphics . AspectRatio . Value = ( ( int ) aspectRatio + 1 ) > Enum . GetNames < AspectRatio > ( ) . Length - 1 ? AspectRatio . Fixed4x3 : aspectRatio + 1 ;
2020-12-16 02:19:07 +00:00
}
2019-12-22 02:49:51 +00:00
private void Row_Clicked ( object sender , ButtonReleaseEventArgs args )
{
2021-01-08 08:14:13 +00:00
if ( args . Event . Button ! = 3 /* Right Click */ )
{
return ;
}
2019-12-22 02:49:51 +00:00
_gameTableSelection . GetSelected ( out TreeIter treeIter ) ;
2021-01-08 08:14:13 +00:00
if ( treeIter . UserData = = IntPtr . Zero )
{
return ;
}
string titleFilePath = _tableStore . GetValue ( treeIter , 9 ) . ToString ( ) ;
string titleName = _tableStore . GetValue ( treeIter , 2 ) . ToString ( ) . Split ( "\n" ) [ 0 ] ;
string titleId = _tableStore . GetValue ( treeIter , 2 ) . ToString ( ) . Split ( "\n" ) [ 1 ] . ToLower ( ) ;
2019-12-22 02:49:51 +00:00
2020-03-25 17:09:38 +00:00
BlitStruct < ApplicationControlProperty > controlData = ( BlitStruct < ApplicationControlProperty > ) _tableStore . GetValue ( treeIter , 10 ) ;
2021-08-12 21:56:24 +00:00
_ = new GameTableContextMenu ( this , _virtualFileSystem , _accountManager , _libHacHorizonManager . RyujinxClient , titleFilePath , titleName , titleId , controlData ) ;
2019-12-22 02:49:51 +00:00
}
2019-11-29 04:32:51 +00:00
private void Load_Application_File ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-09-14 21:52:08 +00:00
using ( FileChooserNative fileChooser = new FileChooserNative ( "Choose the file to open" , this , FileChooserAction . Open , "Open" , "Cancel" ) )
2019-09-02 16:03:57 +00:00
{
2021-09-14 21:52:08 +00:00
FileFilter filter = new FileFilter ( )
{
Name = "Switch Executables"
} ;
filter . AddPattern ( "*.xci" ) ;
filter . AddPattern ( "*.nsp" ) ;
filter . AddPattern ( "*.pfs0" ) ;
filter . AddPattern ( "*.nca" ) ;
filter . AddPattern ( "*.nro" ) ;
filter . AddPattern ( "*.nso" ) ;
fileChooser . AddFilter ( filter ) ;
2021-01-08 08:14:13 +00:00
if ( fileChooser . Run ( ) = = ( int ) ResponseType . Accept )
{
2023-03-31 19:16:46 +00:00
RunApplication ( fileChooser . Filename ) ;
2021-01-08 08:14:13 +00:00
}
2019-09-02 16:03:57 +00:00
}
}
2019-11-29 04:32:51 +00:00
private void Load_Application_Folder ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-09-14 21:52:08 +00:00
using ( FileChooserNative fileChooser = new FileChooserNative ( "Choose the folder to open" , this , FileChooserAction . SelectFolder , "Open" , "Cancel" ) )
2019-09-02 16:03:57 +00:00
{
2021-01-08 08:14:13 +00:00
if ( fileChooser . Run ( ) = = ( int ) ResponseType . Accept )
{
2023-03-31 19:16:46 +00:00
RunApplication ( fileChooser . Filename ) ;
2021-01-08 08:14:13 +00:00
}
2019-09-02 16:03:57 +00:00
}
}
2021-06-28 18:54:45 +00:00
private void FileMenu_StateChanged ( object o , StateChangedArgs args )
{
_appletMenu . Sensitive = _emulationContext = = null & & _contentManager . GetCurrentFirmwareVersion ( ) ! = null & & _contentManager . GetCurrentFirmwareVersion ( ) . Major > 3 ;
_loadApplicationFile . Sensitive = _emulationContext = = null ;
_loadApplicationFolder . Sensitive = _emulationContext = = null ;
}
private void Load_Mii_Edit_Applet ( object sender , EventArgs args )
{
2022-03-22 19:46:16 +00:00
string contentPath = _contentManager . GetInstalledContentPath ( 0x0100000000001009 , StorageId . BuiltInSystem , NcaContentType . Program ) ;
2021-06-28 18:54:45 +00:00
2023-03-31 19:16:46 +00:00
RunApplication ( contentPath ) ;
2021-06-28 18:54:45 +00:00
}
2019-11-29 04:32:51 +00:00
private void Open_Ryu_Folder ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-01-08 08:14:13 +00:00
OpenHelper . OpenFolder ( AppDataManager . BaseDirPath ) ;
2019-09-02 16:03:57 +00:00
}
2020-09-20 04:31:05 +00:00
private void OpenLogsFolder_Pressed ( object sender , EventArgs args )
{
2023-01-07 08:06:13 +00:00
string logPath = System . IO . Path . Combine ( ReleaseInformation . GetBaseApplicationDirectory ( ) , "Logs" ) ;
2020-09-20 04:31:05 +00:00
2021-01-08 08:14:13 +00:00
new DirectoryInfo ( logPath ) . Create ( ) ;
2020-09-20 04:31:05 +00:00
2021-01-08 08:14:13 +00:00
OpenHelper . OpenFolder ( logPath ) ;
2020-09-20 04:31:05 +00:00
}
2019-11-29 04:32:51 +00:00
private void Exit_Pressed ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-01-14 22:30:52 +00:00
if ( ! _gameLoaded | | ! ConfigurationState . Instance . ShowConfirmExit | | GtkDialog . CreateExitDialog ( ) )
2020-10-09 23:06:48 +00:00
{
2021-01-08 08:14:13 +00:00
End ( ) ;
2020-10-09 23:06:48 +00:00
}
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void Window_Close ( object sender , DeleteEventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-01-14 22:30:52 +00:00
if ( ! _gameLoaded | | ! ConfigurationState . Instance . ShowConfirmExit | | GtkDialog . CreateExitDialog ( ) )
2020-10-09 23:06:48 +00:00
{
2021-01-08 08:14:13 +00:00
End ( ) ;
2020-10-09 23:06:48 +00:00
}
else
{
args . RetVal = true ;
}
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void StopEmulation_Pressed ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-09-11 19:16:48 +00:00
if ( _emulationContext ! = null )
{
2023-03-31 19:16:46 +00:00
UpdateGameMetadata ( _emulationContext . Processes . ActiveApplication . ProgramIdText ) ;
2021-09-11 19:16:48 +00:00
}
2021-09-11 20:08:25 +00:00
_pauseEmulation . Sensitive = false ;
2021-09-18 12:31:44 +00:00
_resumeEmulation . Sensitive = false ;
2022-12-04 19:17:11 +00:00
UpdateMenuItem . Sensitive = true ;
2021-05-04 16:19:04 +00:00
RendererWidget ? . Exit ( ) ;
2019-09-02 16:03:57 +00:00
}
2021-09-11 20:08:25 +00:00
private void PauseEmulation_Pressed ( object sender , EventArgs args )
{
2021-09-18 12:31:44 +00:00
_pauseEmulation . Sensitive = false ;
_resumeEmulation . Sensitive = true ;
2021-09-11 20:08:25 +00:00
_emulationContext . System . TogglePauseEmulation ( true ) ;
}
private void ResumeEmulation_Pressed ( object sender , EventArgs args )
{
2021-09-18 12:31:44 +00:00
_pauseEmulation . Sensitive = true ;
_resumeEmulation . Sensitive = false ;
2021-09-11 20:08:25 +00:00
_emulationContext . System . TogglePauseEmulation ( false ) ;
}
public void ActivatePauseMenu ( )
{
_pauseEmulation . Sensitive = true ;
2021-09-18 12:31:44 +00:00
_resumeEmulation . Sensitive = false ;
2021-09-11 20:08:25 +00:00
}
public void TogglePause ( )
{
2021-09-18 12:31:44 +00:00
_pauseEmulation . Sensitive ^ = true ;
_resumeEmulation . Sensitive ^ = true ;
_emulationContext . System . TogglePauseEmulation ( _resumeEmulation . Sensitive ) ;
2021-09-11 20:08:25 +00:00
}
2020-01-12 02:10:55 +00:00
private void Installer_File_Pressed ( object o , EventArgs args )
{
2021-09-14 21:52:08 +00:00
FileChooserNative fileChooser = new FileChooserNative ( "Choose the firmware file to open" , this , FileChooserAction . Open , "Open" , "Cancel" ) ;
FileFilter filter = new FileFilter
{
Name = "Switch Firmware Files"
} ;
filter . AddPattern ( "*.zip" ) ;
filter . AddPattern ( "*.xci" ) ;
2020-01-12 02:10:55 +00:00
2021-09-14 21:52:08 +00:00
fileChooser . AddFilter ( filter ) ;
2020-01-12 02:10:55 +00:00
HandleInstallerDialog ( fileChooser ) ;
}
private void Installer_Directory_Pressed ( object o , EventArgs args )
{
2021-09-14 21:52:08 +00:00
FileChooserNative directoryChooser = new FileChooserNative ( "Choose the firmware directory to open" , this , FileChooserAction . SelectFolder , "Open" , "Cancel" ) ;
2020-01-12 02:10:55 +00:00
HandleInstallerDialog ( directoryChooser ) ;
}
2021-09-14 21:52:08 +00:00
private void HandleInstallerDialog ( FileChooserNative fileChooser )
2020-01-12 02:10:55 +00:00
{
if ( fileChooser . Run ( ) = = ( int ) ResponseType . Accept )
{
try
{
string filename = fileChooser . Filename ;
fileChooser . Dispose ( ) ;
2020-05-03 02:00:53 +00:00
SystemVersion firmwareVersion = _contentManager . VerifyFirmwarePackage ( filename ) ;
2020-01-12 02:10:55 +00:00
2021-08-12 21:56:24 +00:00
if ( firmwareVersion is null )
2020-01-12 02:10:55 +00:00
{
2021-01-08 08:14:13 +00:00
GtkDialog . CreateErrorDialog ( $"A valid system firmware was not found in {filename}." ) ;
2020-01-12 02:10:55 +00:00
return ;
}
2021-08-12 21:56:24 +00:00
string dialogTitle = $"Install Firmware {firmwareVersion.VersionString}" ;
2020-05-03 02:00:53 +00:00
SystemVersion currentVersion = _contentManager . GetCurrentFirmwareVersion ( ) ;
2020-01-12 02:10:55 +00:00
string dialogMessage = $"System version {firmwareVersion.VersionString} will be installed." ;
if ( currentVersion ! = null )
{
2021-01-08 08:14:13 +00:00
dialogMessage + = $"\n\nThis will replace the current system version {currentVersion.VersionString}. " ;
2020-01-12 02:10:55 +00:00
}
2021-01-08 08:14:13 +00:00
dialogMessage + = "\n\nDo you want to continue?" ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
ResponseType responseInstallDialog = ( ResponseType ) GtkDialog . CreateConfirmationDialog ( dialogTitle , dialogMessage ) . Run ( ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
MessageDialog waitingDialog = GtkDialog . CreateWaitingDialog ( dialogTitle , "Installing firmware..." ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
if ( responseInstallDialog = = ResponseType . Yes )
2020-01-12 02:10:55 +00:00
{
2020-08-03 23:32:53 +00:00
Logger . Info ? . Print ( LogClass . Application , $"Installing firmware {firmwareVersion.VersionString}" ) ;
2021-04-04 12:06:59 +00:00
2020-01-12 02:10:55 +00:00
Thread thread = new Thread ( ( ) = >
{
2021-01-08 08:14:13 +00:00
Application . Invoke ( delegate
2020-01-12 02:10:55 +00:00
{
2021-01-08 08:14:13 +00:00
waitingDialog . Run ( ) ;
} ) ;
2020-01-12 02:10:55 +00:00
try
{
2020-01-21 22:23:11 +00:00
_contentManager . InstallFirmware ( filename ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
Application . Invoke ( delegate
2020-01-12 02:10:55 +00:00
{
2021-01-08 08:14:13 +00:00
waitingDialog . Dispose ( ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
string message = $"System version {firmwareVersion.VersionString} successfully installed." ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
GtkDialog . CreateInfoDialog ( dialogTitle , message ) ;
Logger . Info ? . Print ( LogClass . Application , message ) ;
2021-06-28 18:54:45 +00:00
// Purge Applet Cache.
DirectoryInfo miiEditorCacheFolder = new DirectoryInfo ( System . IO . Path . Combine ( AppDataManager . GamesDirPath , "0100000000001009" , "cache" ) ) ;
if ( miiEditorCacheFolder . Exists )
{
miiEditorCacheFolder . Delete ( true ) ;
}
2021-01-08 08:14:13 +00:00
} ) ;
2020-01-12 02:10:55 +00:00
}
catch ( Exception ex )
{
2021-01-08 08:14:13 +00:00
Application . Invoke ( delegate
2020-01-12 02:10:55 +00:00
{
2021-01-08 08:14:13 +00:00
waitingDialog . Dispose ( ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
GtkDialog . CreateErrorDialog ( ex . Message ) ;
} ) ;
2020-01-12 02:10:55 +00:00
}
finally
{
RefreshFirmwareLabel ( ) ;
}
} ) ;
2020-01-13 00:21:54 +00:00
thread . Name = "GUI.FirmwareInstallerThread" ;
2020-01-12 02:10:55 +00:00
thread . Start ( ) ;
}
}
2022-01-12 11:22:19 +00:00
catch ( MissingKeyException ex )
2021-01-26 17:45:07 +00:00
{
Logger . Error ? . Print ( LogClass . Application , ex . ToString ( ) ) ;
2021-08-12 20:48:15 +00:00
UserErrorDialog . CreateUserErrorDialog ( UserError . FirmwareParsingFailed ) ;
2021-01-26 17:45:07 +00:00
}
2020-01-12 02:10:55 +00:00
catch ( Exception ex )
{
2021-01-08 08:14:13 +00:00
GtkDialog . CreateErrorDialog ( ex . Message ) ;
}
}
else
{
fileChooser . Dispose ( ) ;
}
}
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
private void RefreshFirmwareLabel ( )
{
SystemVersion currentFirmware = _contentManager . GetCurrentFirmwareVersion ( ) ;
2020-01-12 02:10:55 +00:00
2021-01-08 08:14:13 +00:00
Application . Invoke ( delegate
{
_firmwareVersionLabel . Text = currentFirmware ! = null ? currentFirmware . VersionString : "0.0.0" ;
} ) ;
}
2020-01-12 02:10:55 +00:00
2023-01-22 17:39:00 +00:00
private void InstallFileTypes_Pressed ( object sender , EventArgs e )
{
if ( FileAssociationHelper . Install ( ) )
{
GtkDialog . CreateInfoDialog ( "Install file types" , "File types successfully installed!" ) ;
}
else
{
GtkDialog . CreateErrorDialog ( "Failed to install file types." ) ;
}
}
private void UninstallFileTypes_Pressed ( object sender , EventArgs e )
{
if ( FileAssociationHelper . Uninstall ( ) )
{
GtkDialog . CreateInfoDialog ( "Uninstall file types" , "File types successfully uninstalled!" ) ;
}
else
{
GtkDialog . CreateErrorDialog ( "Failed to uninstall file types." ) ;
}
}
2021-01-08 08:14:13 +00:00
private void HandleRelaunch ( )
{
if ( _userChannelPersistence . PreviousIndex ! = - 1 & & _userChannelPersistence . ShouldRestart )
{
_userChannelPersistence . ShouldRestart = false ;
2020-01-12 02:10:55 +00:00
2023-03-31 19:16:46 +00:00
RunApplication ( _currentEmulatedGamePath ) ;
2020-01-12 02:10:55 +00:00
}
else
{
2021-01-08 08:14:13 +00:00
// otherwise, clear state.
_userChannelPersistence = new UserChannelPersistence ( ) ;
_currentEmulatedGamePath = null ;
Vulkan backend (#2518)
* WIP Vulkan implementation
* No need to initialize attributes on the SPIR-V backend anymore
* Allow multithreading shaderc and vkCreateShaderModule
You'll only really see the benefit here with threaded-gal or parallel shader cache compile.
Fix shaderc multithreaded changes
Thread safety for shaderc Options constructor
Dunno how they managed to make a constructor not thread safe, but you do you. May avoid some freezes.
* Support multiple levels/layers for blit.
Fixes MK8D when scaled, maybe a few other games. AMD software "safe" blit not supported right now.
* TextureStorage should hold a ref of the foreign storage, otherwise it might be freed while in use
* New depth-stencil blit method for AMD
* Workaround for AMD driver bug
* Fix some tessellation related issues (still doesn't work?)
* Submit command buffer before Texture GetData. (UE4 fix)
* DrawTexture support
* Fix BGRA on OpenGL backend
* Fix rebase build break
* Support format aliasing on SetImage
* Fix uniform buffers being lost when bindings are out of order
* Fix storage buffers being lost when bindings are out of order
(also avoid allocations when changing bindings)
* Use current command buffer for unscaled copy (perf)
Avoids flushing commands and renting a command buffer when fulfilling copy dependencies and when games do unscaled copies.
* Update to .net6
* Update Silk.NET to version 2.10.1
Somehow, massive performance boost. Seems like their vtable for looking up vulkan methods was really slow before.
* Fix PrimitivesGenerated query, disable Transform Feedback queries for now
Lets Splatoon 2 work on nvidia. (mostly)
* Update counter queue to be similar to the OGL one
Fixes softlocks when games had to flush counters.
* Don't throw when ending conditional rendering for now
This should be re-enabled when conditional rendering is enabled on nvidia etc.
* Update findMSB/findLSB to match master's instruction enum
* Fix triangle overlay on SMO, Captain Toad, maybe others?
* Don't make Intel Mesa pay for Intel Windows bugs
* Fix samplers with MinFilter Linear or Nearest (fixes New Super Mario Bros U Deluxe black borders)
* Update Spv.Generator
* Add alpha test emulation on shader (but no shader specialisation yet...)
* Fix R4G4B4A4Unorm texture format permutation
* Validation layers should be enabled for any log level other than None
* Add barriers around vkCmdCopyImage
Write->Read barrier for src image (we want to wait for a write to read it)
Write->Read barrier for dst image (we want to wait for the copy to complete before use)
* Be a bit more careful with texture access flags, since it can be used for anything
* Device local mapping for all buffers
May avoid issues with drivers with NVIDIA on linux/older gpus on windows when using large buffers (?)
Also some performance things and fixes issues with opengl games loading textures weird.
* Cleanup, disable device local buffers for now.
* Add single queue support
Multiqueue seems to be a bit more responsive on NVIDIA. Should fix texture flush on intel. AMD has been forced to single queue for an experiment.
* Fix some validation errors around extended dynamic state
* Remove Intel bug workaround, it was fixed on the latest driver
* Use circular queue for checking consumption on command buffers
Speeds up games that spam command buffers a little. Avoids checking multiple command buffers if multiple are active at once.
* Use SupportBufferUpdater, add single layer flush
* Fix counter queue leak when game decides to use host conditional rendering
* Force device local storage for textures (fixes linux performance)
* Port #3019
* Insert barriers around vkCmdBlitImage (may fix some amd flicker)
* Fix transform feedback on Intel, gl_Position feedback and clears to inexistent depth buffers
* Don't pause transform feedback for multi draw
* Fix draw outside of render pass and missing capability
* Workaround for wrong last attribute on AMD (affects FFVII, STRIKERS1945, probably more)
* Better workaround for AMD vertex buffer size alignment issue
* More instructions + fixes on SPIR-V backend
* Allow custom aspect ratio on Vulkan
* Correct GTK UI status bar positions
* SPIR-V: Functions must always end with a return
* SPIR-V: Fix ImageQuerySizeLod
* SPIR-V: Set DepthReplacing execution mode when FragDepth is modified
* SPIR-V: Implement LoopContinue IR instruction
* SPIR-V: Geometry shader support
* SPIR-V: Use correct binding number on storage buffers array
* Reduce allocations for Spir-v serialization
Passes BinaryWriter instead of the stream to Write and WriteOperand
- Removes creation of BinaryWriter for each instruction
- Removes allocations for literal string
* Some optimizations to Spv.Generator
- Dictionary for lookups of type declarations, constants, extinst
- LiteralInteger internal data format -> ushort
- Deterministic HashCode implementation to avoid spirv result not being the same between runs
- Inline operand list instead of List<T>, falls back to array if many operands. (large performance boost)
TODO: improve instruction allocation, structured program creator, ssa?
* Pool Spv.Generator resources, cache delegates, spv opts
- Pools for Instructions and LiteralIntegers. Can be passed in when creating the generator module.
- NewInstruction is called instead of new Instruction()
- Ryujinx SpirvGenerator passes in some pools that are static. The idea is for these to be shared between threads eventually.
- Estimate code size when creating the output MemoryStream
- LiteralInteger pools using ThreadStatic pools that are initialized before and after creation... not sure of a better way since the way these are created is via implicit cast.
Also, cache delegates for Spv.Generator for functions that are passed around to GenerateBinary etc, since passing the function raw creates a delegate on each call.
TODO: update python spv cs generator to make the coregrammar with NewInstruction and the `params` overloads.
* LocalDefMap for Ssa Rewriter
Rather than allocating a large array of all registers for each block in the shader, allocate one array of all registers and clear it between blocks. Reduces allocations in the shader translator.
* SPIR-V: Transform feedback support
* SPIR-V: Fragment shader interlock support (and image coherency)
* SPIR-V: Add early fragment tests support
* SPIR-V: Implement SwizzleAdd, add missing Triangles ExecutionMode for geometry shaders, remove SamplerType field from TextureMeta
* Don't pass depth clip state right now (fix decals)
Explicitly disabling it is incorrect. OpenGL currently automatically disables based on depth clamp, which is the behaviour if this state is omitted.
* Multisampling support
* Multisampling: Use resolve if src samples count > dst samples count
* Multisampling: We can only resolve for unscaled copies
* SPIR-V: Only add FSI exec mode if used.
* SPIR-V: Use ConstantComposite for Texture Offset Vector
Fixes a bunch of freezes with SPIR-V on AMD hardware, and validation errors. Note: Obviously assumes input offsets are constant, which they currently are.
* SPIR-V: Don't OpReturn if we already OpExit'ed
Fixes spir-v parse failure and stack smashing in RADV (obviously you still need bolist)
* SPIR-V: Only use input attribute type for input attributes
Output vertex attributes should always be of type float.
* Multithreaded Pipeline Compilation
* Address some feedback
* Make this 32
* Update topology with GpuAccessorState
* Cleanup for merge (note: disables spir-v)
* Make more robust to shader compilation failure
- Don't freeze when GLSL compilation fails
- Background SPIR-V pipeline compile failure results in skipped draws, similar to GLSL compilation failure.
* Fix Multisampling
* Only update fragment scale count if a vertex texture needs a scale.
Fixes a performance regression introduced by texture scaling in the vertex stage where support buffer updates would be very frequent, even at 1x, if any textures were used on the vertex stage.
This check doesn't exactly look cheap (a flag in the shader stage would probably be preferred), but it is much cheaper than uploading scales in both vulkan and opengl, so it will do for now.
* Use a bitmap to do granular tracking for buffer uploads.
This path is only taken if the much faster check of "is the buffer rented at all" is triggered, so it doesn't actually end up costing too much, and the time saved by not ending render passes (and on gpu for not waiting on barriers) is probably helpful.
Avoids ending render passes to update buffer data (not all the time)
- 140-180 to 35-45 in SMO metro kingdom (these updates are in the UI)
- Very variable 60-150(!) to 16-25 in mario kart 8 (these updates are in the UI)
As well as allowing more data to be preloaded persistently, this will also allow more data to be loaded in the preload buffer, which should be faster as it doesn't need to insert barriers between draws. (and on tbdr, does not need to flush and reload tile memory)
Improves performance in GPU limited scenarios. Should notably improve performance on TBDR gpus. Still a lot more to do here.
* Copy query results after RP ends, rather than ending to copy
We need to end the render pass to get the data (submit command buffer) anyways...
Reduces render passes created in games that use queries.
* Rework Query stuff a bit to avoid render pass end
Tries to reset returned queries in background when possible, rather than ending the render pass.
Still ends render pass when resetting a counter after draws, but maybe that can be solved too. (by just pulling an empty object off the pool?)
* Remove unnecessary lines
Was for testing
* Fix validation error for query reset
Need to think of a better way to do this.
* SPIR-V: Fix SwizzleAdd and some validation errors
* SPIR-V: Implement attribute indexing and StoreAttribute
* SPIR-V: Fix TextureSize for MS and Buffer sampler types
* Fix relaunch issues
* SPIR-V: Implement LogicalExclusiveOr
* SPIR-V: Constant buffer indexing support
* Ignore unsupported attributes rather than throwing (matches current GLSL behaviour)
* SPIR-V: Implement tessellation support
* SPIR-V: Geometry shader passthrough support
* SPIR-V: Implement StoreShader8/16 and StoreStorage8/16
* SPIR-V: Resolution scale support and fix TextureSample multisample with LOD bug
* SPIR-V: Fix field index for scale count
* SPIR-V: Fix another case of wrong field index
* SPIRV/GLSL: More scaling related fixes
* SPIR-V: Fix ImageLoad CompositeExtract component type
* SPIR-V: Workaround for Intel FrontFacing bug
* Enable SPIR-V backend by default
* Allow null samplers (samplers are not required when only using texelFetch to access the texture)
* Fix some validation errors related to texel block view usage flag and invalid image barrier base level
* Use explicit subgroup size if we can (might fix some block flickering on AMD)
* Take componentMask and scissor into account when clearing framebuffer attachments
* Add missing barriers around CmdFillBuffer (fixes Monster Hunter Rise flickering on NVIDIA)
* Use ClampToEdge for Clamp sampler address mode on Vulkan (fixes Hollow Knight)
Clamp is unsupported on Vulkan, but ClampToEdge behaves almost the same. ClampToBorder on the other hand (which was being used before) is pretty different
* Shader specialization for new Vulkan required state (fixes remaining alpha test issues, vertex stretching on AMD on Crash Bandicoot, etc)
* Check if the subgroup size is supported before passing a explicit size
* Only enable ShaderFloat64 if the GPU supports it
* We don't need to recompile shaders if alpha test state changed but alpha test is disabled
* Enable shader cache on Vulkan and implement MultiplyHighS32/U32 on SPIR-V (missed those before)
* Fix pipeline state saving before it is updated.
This should fix a few warnings and potential stutters due to bad pipeline states being saved in the cache. You may need to clear your guest cache.
* Allow null samplers on OpenGL backend
* _unit0Sampler should be set only for binding 0
* Remove unused PipelineConverter format variable (was causing IOR)
* Raise textures limit to 64 on Vulkan
* No need to pack the shader binaries if shader cache is disabled
* Fix backbuffer not being cleared and scissor not being re-enabled on OpenGL
* Do not clear unbound framebuffer color attachments
* Geometry shader passthrough emulation
* Consolidate UpdateDepthMode and GetDepthMode implementation
* Fix A1B5G5R5 texture format and support R4G4 on Vulkan
* Add barrier before use of some modified images
* Report 32 bit query result on AMD windows (smo issue)
* Add texture recompression support (disabled for now)
It recompresses ASTC textures into BC7, which might reduce VRAM usage significantly on games that uses ASTC textures
* Do not report R4G4 format as supported on Vulkan
It was causing mario head to become white on Super Mario 64 (???)
* Improvements to -1 to 1 depth mode.
- Transformation is only applied on the last stage in the vertex pipeline.
- Should fix some issues with geometry and tessellation (hopefully)
- Reading back FragCoord Z on fragment will transform back to -1 to 1.
* Geometry Shader index count from ThreadsPerInputPrimitive
Generally fixes SPIR-V emitting too many triangles, may change games in OpenGL
* Remove gl_FragDepth scaling
This is always 0-1; the other two issues were causing the problems. Fixes regression with Xenoblade.
* Add Gl StencilOp enum values to Vulkan
* Update guest cache to v1.1 (due to specialization state changes)
This will explode your shader cache from earlier vulkan build, but it must be done. :pensive:
* Vulkan/SPIR-V support for viewport inverse
* Fix typo
* Don't create query pools for unsupported query types
* Return of the Vector Indexing Bug
One day, everyone will get this right.
* Check for transform feedback query support
Sometimes transform feedback is supported without the query type.
* Fix gl_FragCoord.z transformation
FragCoord.z is always in 0-1, even when the real depth range is -1 to 1. Turns out the only bug was geo and tess stage outputs.
Fixes Pokemon Sword/Shield, possibly others.
* Fix Avalonia Rebase
Vulkan is currently not available on Avalonia, but the build does work and you can use opengl.
* Fix headless build
* Add support for BC6 and BC7 decompression, decompress all BC formats if they are not supported by the host
* Fix BCn 4/5 conversion, GetTextureTarget
BCn 4/5 could generate invalid data when a line's size in bytes was not divisible by 4, which both backends expect.
GetTextureTarget was not creating a view with the replacement format.
* Fix dependency
* Fix inverse viewport transform vector type on SPIR-V
* Do not require null descriptors support
* If MultiViewport is not supported, do not try to set more than one viewport/scissor
* Bounds check on bitmap add.
* Flush queries on attachment change rather than program change
Occlusion queries are usually used in a depth only pass so the attachments changing is a better indication of the query block ending.
Write mask changes are also considered since some games do depth only pass by setting 0 write mask on all the colour targets.
* Add support for avalonia (#6)
* add avalonia support
* only lock around skia flush
* addressed review
* cleanup
* add fallback size if avalonia attempts to render but the window size is 0. read desktop scale after enabling dpi check
* fix getting window handle on linux. skip render is size is 0
* Combine non-buffer with buffer image descriptor sets
* Support multisample texture copy with automatic resolve on Vulkan
* Remove old CompileShader methods from the Vulkan backend
* Add minimal pipeline layouts that only contains used bindings
They are used by helper shaders, the intention is avoiding needing to recompile the shaders (from GLSL to SPIR-V) if the bindings changes on the translated guest shaders
* Pre-compile helper shader as SPIR-V, and some fixes
* Remove pre-compiled shaderc binary for Windows as its no longer needed by default
* Workaround RADV crash
Enabling the descriptor indexing extension, even if it is not used, forces the radv driver to use "bolist".
* Use RobustBufferAccess on NVIDIA gpus
Avoids the SMO waterfall triangle on older NVIDIA gpus.
* Implement GPU selector and expose texture recompression on the UI and config
* Fix and enable background compute shader compilation
Also disables warnings from shader cache pipeline misses.
* Fix error due to missing subpass dependency when Attachment Write -> Shader Read barriers are added
* If S8D24 is not supported, use D32FS8
* Ensure all fences are destroyed on dispose
* Pre-allocate arrays up front on DescriptorSetUpdater, allows the removal of some checks
* Add missing clear layer parameter after rebase
* Use selected gpu from config for avalonia (#7)
* use configured device
* address review
* Fix D32S8 copy workaround (AMD)
Fixes water in Pokemon Legends Arceus on AMD GPUs. Possibly fixes other things.
* Use push descriptors for uniform buffer updates (disabled for now)
* Push descriptor support check, buffer redundancy checks
Should make push descriptors faster, needs more testing though.
* Increase light command buffer pool to 2 command buffers, throw rather than returning invalid cbs
* Adjust bindings array sizes
* Force submit command buffers if memory in use by its resources is high
* Add workaround for AMD GCN cubemap view sins
`ImageCreateCubeCompatibleBit` seems to generally break 2D array textures with mipmaps... even if they are eventually aliased as a cubemap with mipmaps. Forcing a copy here works around the issue.
This could be used in future if enabling this bit reduces performance on certain GPUs. (mobile class is generally a worry)
Currently also enabled on Linux as I don't know if they managed to dodge this bug (someone please tell me). Not enabled on Vega at the moment, but easy to add if the issue is there.
* Add mobile, non-RX variants to the GCN regex.
Also make sure that the 3 digit ones only include numbers starting with 7 or 8.
* Increase image limit per stage from 8 to 16
Xenoblade Chronicles 2 was hiting the limit of 8
* Minor code cleanup
* Fix NRE caused by SupportBufferUpdater calling pipeline ClearBuffer
* Add gpu selector to Avalonia (#8)
* Add gpu selector to avalonia settings
* show backend label on window
* some fixes
* address review
* Minor changes to the Avalonia UI
* Update graphics window UI and locales. (#9)
* Update xaml and update locales
* locale updates
Did my best here but likely needs to be checked by native speakers, especially the use of ampersands in greek, russian and turkish?
* Fix locales with more (?) correct translations.
* add separator to render widget
* fix spanish and portuguese
* Add new IdList, replaces buffer list that could not remove elements and had unbounded growth
* Don't crash the settings window if Vulkan is not supported
* Fix Actions menu not being clickable on GTK UI after relaunch
* Rename VulkanGraphicsDevice to VulkanRenderer and Renderer to OpenGLRenderer
* Fix IdList and make it not thread safe
* Revert useless OpenGL format table changes
* Fix headless project build
* List throws ArgumentOutOfRangeException
* SPIR-V: Fix tessellation
* Increase shader cache version due to tessellation fix
* Reduce number of Sync objects created (improves perf in some specific titles)
* Fix vulkan validation errors for NPOT compressed upload and GCN workaround.
* Add timestamp to the shader cache and force rebuild if host cache is outdated
* Prefer Mail box present mode for popups (#11)
* Prefer Mail box present mode
* fix debug
* switch present mode when vsync is toggled
* only disable vsync on the main window
* SPIR-V: Fix geometry shader input load with transform feedback
* BC7 Encoder: Prefer more precision on alpha rather than RGB when alpha is 0
* Fix Avalonia build
* Address initial PR feedback
* Only set transform feedback outputs on last vertex stage
* Address riperiperi PR feedback
* Remove outdated comment
* Remove unused constructor
* Only throw for negative results
* Throw for QueueSubmit and other errors
No point in delaying the inevitable
* Transform feedback decorations inside gl_PerVertex struct breaks the NVIDIA compiler
* Fix some resolution scale issues
* No need for two UpdateScale calls
* Fix comments on SPIR-V generator project
* Try to fix shader local memory size
On DOOM, a shader is using local memory, but both Low and High size are 0, CRS size is 1536, it seems to store on that region?
* Remove RectangleF that is now unused
* Fix ImageGather with multiple offsets
Needs ImageGatherExtended capability, and must use `ConstantComposite` instead of `CompositeConstruct`
* Address PR feedback from jD in all projects except Avalonia
* Address most of jD PR feedback on Avalonia
* Remove unsafe
* Fix VulkanSkiaGpu
* move present mode request out of Create Swapchain method
* split more parts of create swapchain
* addressed reviews
* addressed review
* Address second batch of jD PR feedback
* Fix buffer <-> image copy row length and height alignment
AlignUp helper does not support NPOT alignment, and ASTC textures can have NPOT block sizes
* Better fix for NPOT alignment issue
* Use switch expressions on Vulkan EnumConversion
Thanks jD
* Fix Avalonia build
* Add Vulkan selection prompt on startup
* Grammar fixes on Vulkan prompt message
* Add missing Vulkan migration flag
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Emmanuel Hansen <emmausssss@gmail.com>
Co-authored-by: MutantAura <44103205+MutantAura@users.noreply.github.com>
2022-07-31 21:26:06 +00:00
_actionMenu . Sensitive = false ;
_firmwareInstallFile . Sensitive = true ;
_firmwareInstallDirectory . Sensitive = true ;
2020-01-12 02:10:55 +00:00
}
}
2020-12-01 22:02:27 +00:00
private void FullScreen_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-01-08 08:14:13 +00:00
if ( ! Window . State . HasFlag ( Gdk . WindowState . Fullscreen ) )
2019-09-02 16:03:57 +00:00
{
Fullscreen ( ) ;
2020-02-12 00:56:19 +00:00
ToggleExtraWidgets ( false ) ;
2019-09-02 16:03:57 +00:00
}
else
{
Unfullscreen ( ) ;
2020-02-12 00:56:19 +00:00
ToggleExtraWidgets ( true ) ;
2019-09-02 16:03:57 +00:00
}
}
2020-12-01 22:02:27 +00:00
private void StartFullScreen_Toggled ( object sender , EventArgs args )
{
ConfigurationState . Instance . Ui . StartFullscreen . Value = _startFullScreen . Active ;
SaveConfig ( ) ;
}
2022-03-15 01:35:41 +00:00
private void ShowConsole_Toggled ( object sender , EventArgs args )
{
ConfigurationState . Instance . Ui . ShowConsole . Value = _showConsole . Active ;
SaveConfig ( ) ;
}
2021-04-23 20:26:31 +00:00
private void OptionMenu_StateChanged ( object o , StateChangedArgs args )
{
_manageUserProfiles . Sensitive = _emulationContext = = null ;
}
2019-11-29 04:32:51 +00:00
private void Settings_Pressed ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-02-20 00:34:41 +00:00
SettingsWindow settingsWindow = new SettingsWindow ( this , _virtualFileSystem , _contentManager ) ;
settingsWindow . SetSizeRequest ( ( int ) ( settingsWindow . DefaultWidth * Program . WindowScaleFactor ) , ( int ) ( settingsWindow . DefaultHeight * Program . WindowScaleFactor ) ) ;
settingsWindow . Show ( ) ;
2019-09-02 16:03:57 +00:00
}
2021-07-24 18:48:00 +00:00
private void HideUi_Pressed ( object sender , EventArgs args )
{
ToggleExtraWidgets ( false ) ;
}
2022-01-03 08:39:43 +00:00
private void ManageCheats_Pressed ( object sender , EventArgs args )
{
2023-03-31 19:16:46 +00:00
var window = new CheatWindow ( _virtualFileSystem ,
_emulationContext . Processes . ActiveApplication . ProgramId ,
_emulationContext . Processes . ActiveApplication . ApplicationControlProperties . Title [ ( int ) _emulationContext . System . State . DesiredTitleLanguage ] . NameString . ToString ( ) ) ;
2022-01-03 08:39:43 +00:00
window . Destroyed + = CheatWindow_Destroyed ;
window . Show ( ) ;
}
private void CheatWindow_Destroyed ( object sender , EventArgs e )
{
_emulationContext . EnableCheats ( ) ;
( sender as CheatWindow ) . Destroyed - = CheatWindow_Destroyed ;
}
2021-04-23 20:26:31 +00:00
private void ManageUserProfiles_Pressed ( object sender , EventArgs args )
{
UserProfilesManagerWindow userProfilesManagerWindow = new UserProfilesManagerWindow ( _accountManager , _contentManager , _virtualFileSystem ) ;
userProfilesManagerWindow . SetSizeRequest ( ( int ) ( userProfilesManagerWindow . DefaultWidth * Program . WindowScaleFactor ) , ( int ) ( userProfilesManagerWindow . DefaultHeight * Program . WindowScaleFactor ) ) ;
userProfilesManagerWindow . Show ( ) ;
}
2020-12-16 00:41:42 +00:00
private void Simulate_WakeUp_Message_Pressed ( object sender , EventArgs args )
{
2021-01-08 08:14:13 +00:00
if ( _emulationContext ! = null )
{
_emulationContext . System . SimulateWakeUpMessage ( ) ;
}
2020-12-16 00:41:42 +00:00
}
2021-03-18 20:40:20 +00:00
private void ActionMenu_StateChanged ( object o , StateChangedArgs args )
{
2021-06-28 20:09:43 +00:00
_scanAmiibo . Sensitive = _emulationContext ! = null & & _emulationContext . System . SearchingForAmiibo ( out int _ ) ;
_takeScreenshot . Sensitive = _emulationContext ! = null ;
2021-03-18 20:40:20 +00:00
}
private void Scan_Amiibo ( object sender , EventArgs args )
{
if ( _emulationContext . System . SearchingForAmiibo ( out int deviceId ) )
{
AmiiboWindow amiiboWindow = new AmiiboWindow
{
LastScannedAmiiboShowAll = _lastScannedAmiiboShowAll ,
LastScannedAmiiboId = _lastScannedAmiiboId ,
DeviceId = deviceId ,
2023-03-31 19:16:46 +00:00
TitleId = _emulationContext . Processes . ActiveApplication . ProgramIdText . ToUpper ( )
2021-03-18 20:40:20 +00:00
} ;
amiiboWindow . DeleteEvent + = AmiiboWindow_DeleteEvent ;
amiiboWindow . Show ( ) ;
}
else
{
GtkDialog . CreateInfoDialog ( $"Amiibo" , "The game is currently not ready to receive Amiibo scan data. Ensure that you have an Amiibo-compatible game open and ready to receive Amiibo scan data." ) ;
}
}
2021-06-28 20:09:43 +00:00
private void Take_Screenshot ( object sender , EventArgs args )
{
if ( _emulationContext ! = null & & RendererWidget ! = null )
{
RendererWidget . ScreenshotRequested = true ;
}
}
2021-03-18 20:40:20 +00:00
private void AmiiboWindow_DeleteEvent ( object sender , DeleteEventArgs args )
{
if ( ( ( AmiiboWindow ) sender ) . AmiiboId ! = "" & & ( ( AmiiboWindow ) sender ) . Response = = ResponseType . Ok )
{
_lastScannedAmiiboId = ( ( AmiiboWindow ) sender ) . AmiiboId ;
_lastScannedAmiiboShowAll = ( ( AmiiboWindow ) sender ) . LastScannedAmiiboShowAll ;
_emulationContext . System . ScanAmiibo ( ( ( AmiiboWindow ) sender ) . DeviceId , ( ( AmiiboWindow ) sender ) . AmiiboId , ( ( AmiiboWindow ) sender ) . UseRandomUuid ) ;
}
}
2019-11-29 04:32:51 +00:00
private void Update_Pressed ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2020-09-29 20:05:25 +00:00
if ( Updater . CanUpdate ( true ) )
2019-09-02 16:03:57 +00:00
{
2021-02-23 15:19:02 +00:00
Updater . BeginParse ( this , true ) . ContinueWith ( task = >
{
2021-03-18 22:44:39 +00:00
Logger . Error ? . Print ( LogClass . Application , $"Updater error: {task.Exception}" ) ;
2021-02-23 15:19:02 +00:00
} , TaskContinuationOptions . OnlyOnFaulted ) ;
2019-09-02 16:03:57 +00:00
}
}
2019-11-29 04:32:51 +00:00
private void About_Pressed ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2021-02-20 00:34:41 +00:00
AboutWindow aboutWindow = new AboutWindow ( ) ;
aboutWindow . SetSizeRequest ( ( int ) ( aboutWindow . DefaultWidth * Program . WindowScaleFactor ) , ( int ) ( aboutWindow . DefaultHeight * Program . WindowScaleFactor ) ) ;
aboutWindow . Show ( ) ;
2019-11-29 04:32:51 +00:00
}
private void Fav_Toggled ( object sender , EventArgs args )
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . FavColumn . Value = _favToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void Icon_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . IconColumn . Value = _iconToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2021-02-09 09:24:37 +00:00
private void App_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . AppColumn . Value = _appToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void Developer_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . DevColumn . Value = _developerToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void Version_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . VersionColumn . Value = _versionToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void TimePlayed_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . TimePlayedColumn . Value = _timePlayedToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void LastPlayed_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . LastPlayedColumn . Value = _lastPlayedToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void FileExt_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . FileExtColumn . Value = _fileExtToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void FileSize_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . FileSizeColumn . Value = _fileSizeToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
2019-09-02 16:03:57 +00:00
}
2019-11-29 04:32:51 +00:00
private void Path_Toggled ( object sender , EventArgs args )
2019-09-02 16:03:57 +00:00
{
2019-12-21 19:52:31 +00:00
ConfigurationState . Instance . Ui . GuiColumns . PathColumn . Value = _pathToggle . Active ;
2019-11-29 04:32:51 +00:00
2019-12-21 19:52:31 +00:00
SaveConfig ( ) ;
2019-11-29 04:32:51 +00:00
UpdateColumns ( ) ;
}
private void RefreshList_Pressed ( object sender , ButtonReleaseEventArgs args )
{
UpdateGameTable ( ) ;
}
2019-09-02 16:03:57 +00:00
}
2022-11-12 23:36:36 +00:00
}
