Fix vote and shuffle shader instructions on AMD GPUs (#5540)

* 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
This commit is contained in:
gdkchan 2023-08-16 21:31:07 -03:00 committed by GitHub
parent 64079c034c
commit 6ed613a6e6
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
35 changed files with 445 additions and 265 deletions

View file

@ -151,6 +151,14 @@ namespace Ryujinx.Graphics.Vulkan
SType = StructureType.PhysicalDeviceProperties2,
};
PhysicalDeviceSubgroupProperties propertiesSubgroup = new()
{
SType = StructureType.PhysicalDeviceSubgroupProperties,
PNext = properties2.PNext,
};
properties2.PNext = &propertiesSubgroup;
PhysicalDeviceBlendOperationAdvancedPropertiesEXT propertiesBlendOperationAdvanced = new()
{
SType = StructureType.PhysicalDeviceBlendOperationAdvancedPropertiesExt,
@ -164,18 +172,6 @@ namespace Ryujinx.Graphics.Vulkan
properties2.PNext = &propertiesBlendOperationAdvanced;
}
PhysicalDeviceSubgroupSizeControlPropertiesEXT propertiesSubgroupSizeControl = new()
{
SType = StructureType.PhysicalDeviceSubgroupSizeControlPropertiesExt,
};
bool supportsSubgroupSizeControl = _physicalDevice.IsDeviceExtensionPresent("VK_EXT_subgroup_size_control");
if (supportsSubgroupSizeControl)
{
properties2.PNext = &propertiesSubgroupSizeControl;
}
bool supportsTransformFeedback = _physicalDevice.IsDeviceExtensionPresent(ExtTransformFeedback.ExtensionName);
PhysicalDeviceTransformFeedbackPropertiesEXT propertiesTransformFeedback = new()
@ -315,7 +311,6 @@ namespace Ryujinx.Graphics.Vulkan
_physicalDevice.IsDeviceExtensionPresent(KhrDrawIndirectCount.ExtensionName),
_physicalDevice.IsDeviceExtensionPresent("VK_EXT_fragment_shader_interlock"),
_physicalDevice.IsDeviceExtensionPresent("VK_NV_geometry_shader_passthrough"),
supportsSubgroupSizeControl,
features2.Features.ShaderFloat64,
featuresShaderInt8.ShaderInt8,
_physicalDevice.IsDeviceExtensionPresent("VK_EXT_shader_stencil_export"),
@ -335,9 +330,7 @@ namespace Ryujinx.Graphics.Vulkan
_physicalDevice.IsDeviceExtensionPresent("VK_NV_viewport_array2"),
_physicalDevice.IsDeviceExtensionPresent(ExtExternalMemoryHost.ExtensionName),
supportsDepthClipControl && featuresDepthClipControl.DepthClipControl,
propertiesSubgroupSizeControl.MinSubgroupSize,
propertiesSubgroupSizeControl.MaxSubgroupSize,
propertiesSubgroupSizeControl.RequiredSubgroupSizeStages,
propertiesSubgroup.SubgroupSize,
supportedSampleCounts,
portabilityFlags,
vertexBufferAlignment,
@ -623,6 +616,7 @@ namespace Ryujinx.Graphics.Vulkan
maximumImagesPerStage: Constants.MaxImagesPerStage,
maximumComputeSharedMemorySize: (int)limits.MaxComputeSharedMemorySize,
maximumSupportedAnisotropy: (int)limits.MaxSamplerAnisotropy,
shaderSubgroupSize: (int)Capabilities.SubgroupSize,
storageBufferOffsetAlignment: (int)limits.MinStorageBufferOffsetAlignment,
gatherBiasPrecision: IsIntelWindows || IsAmdWindows ? (int)Capabilities.SubTexelPrecisionBits : 0);
}