Metal sounded like a good idea to get in the emulator but frankly I
underestimated just how experimental and not ready it was.
From my write up in the Discord:
```
As is, Metal supports only a few games.
The games it does support freeze on first use of not playing them via Vulkan, because shader translation is broken.
So you need to use a dirty hack to not delete all your shaders.
Not to mention it breaks many games via MoltenVK because of changes to the shared GPU code.
Merging Metal seemed like a great idea, because of the few games it does support.
But I don't think it's worth it. Many of the games it breaks via MoltenVK *don't work via Metal*.
Which effectively makes current Ryubing worse for Mac users than Ryujinx 1.1.1403.
I think what I'm gonna do is revert Metal, and reopen it as a PR. That way, you can still take advantage of the Metal backend as is, but without making other games worse with no solution.
```
For what it's worth, the shader translation part could at least be
"fixed" by always applying a 30ms delay for shader translation to Metal.
That being said, that solution sucks ass.
The MoltenVK regressions are even worse.
I hope this is not a let down to the Mac users. I hope you realize I'm
reverting this because you're actively getting a worse experience with
it in the emulator.
This is not a continuation of the Metal backend; this is simply bringing
the branch up to date and merging it as-is behind an experiment.
---------
Co-authored-by: Isaac Marovitz <isaacryu@icloud.com>
Co-authored-by: Samuliak <samuliak77@gmail.com>
Co-authored-by: SamoZ256 <96914946+SamoZ256@users.noreply.github.com>
Co-authored-by: Isaac Marovitz <42140194+IsaacMarovitz@users.noreply.github.com>
Co-authored-by: riperiperi <rhy3756547@hotmail.com>
Co-authored-by: Gabriel A <gab.dark.100@gmail.com>
* Report base and extra sets from the backend
* Pass texture set index everywhere
* Key textures using set and binding (rather than just binding)
* Start using extra sets for array textures
* Shader cache version bump
* Separate new commands, some PR feedback
* Introduce new manual descriptor set reservation method that prevents it from being used by something else while owned by an array
* Move bind extra sets logic to new method
* Should only use separate array is MaximumExtraSets is not zero
* Format whitespace
* Add support for large sampler arrays on Vulkan
* Shader cache version bump
* Format whitespace
* Move DescriptorSetManager to PipelineLayoutCacheEntry to allow different pool sizes per layout
* Handle array textures with different types on the same buffer
* Somewhat better caching system
* Avoid useless buffer data modification checks
* Move redundant bindings update checking to the backend
* Fix an issue where texture arrays would get the same bindings across stages on Vulkan
* Backport some fixes from part 2
* Fix typo
* PR feedback
* Format whitespace
* Add some missing XML docs
* GPU: Add fallback when textureGatherOffsets is not supported.
This PR adds a fallback for GPUs or APIs that don't support an equivalent to the method `textureGatherOffsets`, where each of the 4 gathered texels has an individual offset. This is done by reusing the existing code to handle non-const offsets for texture instructions, though it has also been corrected as there were a few implementation issues.
MoltenVK reports support for this capability, and it didn't error when we initially released the MacOS build, but that has since changed. MVK still reports support, but spirv-cross has been fixed in a way that it _attempts_ to use this capability, but the metal compiler errors since it doesn't exist.
Some other fixes:
- textureGatherOffsets emulation has been changed significantly. It now uses 4 texture sample instructions (not gather), calculates a base texel (i=0 j=0) and adds the offsets onto it before converting into a tex coord. The final result is offset into a texel center, so it shouldn't be subject to interpolation, though this isn't perfect and could have some error with floating point formats with linear sampling. It is subject to texture wrap mode as it should be, which is why texelFetch was not used.
- Maybe gather should be used here with component `w` (i=0, j=0), though this multiplies number of texels fetched by 4... The way it was doing this before _was_ wrong_, but doing it right would avoid issues with texel center precision.
- textureGatherOffset (singular) now performs textureGather with the offset applied to the coords, rather than the slower fallback where each texel is fetched individually.
* Increment shader cache version, remove unused arg
* Use base texture size for gather coord offset.
Implicit LOD for gather is not supported.
* Use 4 texture gathers for offsets emulation
Avoids issues with interpolation at cost of performance
(not sure how bad this is)
* Address Feedback
* Implement vertex and geometry shader conversion to compute
* Call InitializeReservedCounts for compute too
* PR feedback
* Set clip distance mask for geometry and tessellation shaders too
* Transform feedback emulation only for vertex
* Move shuffle handling out of the backend to a transform pass
* Handle subgroup sizes higher than 32
* Stop using the subgroup size control extension
* Make GenerateShuffleFunction static
* Shader cache version bump
* Move some properties out of ShaderConfig
* Stop using ShaderConfig on backends
* Replace ShaderConfig usages on Translator and passes
* Move remaining properties out of ShaderConfig and delete ShaderConfig
* Remove ResourceManager property from TranslatorContext
* Move Rewriter passes to separate transform pass files
* Fix TransformPasses.RunPass on cases where a node is removed
* Move remaining ClipDistancePrimitivesWritten and UsedFeatures updates to decode stage
* Reduce excessive parameter passing a bit by using structs more
* Remove binding parameter from ShaderProperties methods since it is redundant
* Replace decoder instruction checks with switch statement
* Put GLSL on the same plan as SPIR-V for input/output declaration
* Stop mutating TranslatorContext state when Translate is called
* Pass most of the graphics state using a struct instead of individual query methods
* Auto-format
* Auto-format
* Add backend logging interface
* Auto-format
* Remove unnecessary use of interpolated strings
* Remove more modifications of AttributeUsage after decode
* PR feedback
* gl_Layer is not supported on compute
