WIP: In-shader vertex loading support

ExternalDependencies.xcodeproj/project.pbxproj

@@ -80,6 +80,10 @@
 /* End PBXAggregateTarget section */
 
 /* Begin PBXBuildFile section */
+ 164DCE072AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */; };
+ 164DCE082AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */; };
+ 164DCE092AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */; };
+ 164DCE0A2AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */; };
  2FEA0CDB2490322B00EEF3AD /* spirv_cfg.hpp in Headers */ = {isa = PBXBuildFile; fileRef = A976290921CC60BC00B52A68 /* spirv_cfg.hpp */; };
  2FEA0CDC2490322B00EEF3AD /* spirv_cross_parsed_ir.hpp in Headers */ = {isa = PBXBuildFile; fileRef = A976290821CC60BC00B52A68 /* spirv_cross_parsed_ir.hpp */; };
  2FEA0CDD2490322B00EEF3AD /* spirv_common.hpp in Headers */ = {isa = PBXBuildFile; fileRef = A976290721CC60BC00B52A68 /* spirv_common.hpp */; };
@@ -2367,6 +2371,7 @@
 /* End PBXContainerItemProxy section */
 
 /* Begin PBXFileReference section */
+ 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = spirv_msl_vertex_loader.cpp; sourceTree = "<group>"; };
  2FEA0B4A2490321700EEF3AD /* libglslang.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libglslang.a; sourceTree = BUILT_PRODUCTS_DIR; };
  2FEA0CD82490322100EEF3AD /* libSPIRVTools.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libSPIRVTools.a; sourceTree = BUILT_PRODUCTS_DIR; };
  2FEA0CEF2490322B00EEF3AD /* libSPIRVCross.a */ = {isa = PBXFileReference; explicitFileType = archive.ar; includeInIndex = 0; path = libSPIRVCross.a; sourceTree = BUILT_PRODUCTS_DIR; };
@@ -3019,6 +3024,7 @@
  A976290321CC60BC00B52A68 /* spirv_cross.hpp */,
  A976290621CC60BC00B52A68 /* spirv_glsl.cpp */,
  A976290A21CC60BC00B52A68 /* spirv_glsl.hpp */,
+ 164DCE062AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp */,
  A976290D21CC60BC00B52A68 /* spirv_msl.cpp */,
  A976290221CC60BC00B52A68 /* spirv_msl.hpp */,
  A976290421CC60BC00B52A68 /* spirv_parser.cpp */,
@@ -5813,6 +5819,7 @@
  2FEA0CE72490322B00EEF3AD /* spirv_cross.cpp in Sources */,
  2FEA0CE82490322B00EEF3AD /* spirv_reflect.cpp in Sources */,
  2FEA0CE92490322B00EEF3AD /* spirv_glsl.cpp in Sources */,
+ 164DCE082AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */,
  2FEA0CEA2490322B00EEF3AD /* spirv_cross_parsed_ir.cpp in Sources */,
  );
  runOnlyForDeploymentPostprocessing = 0;
@@ -6385,6 +6392,7 @@
  A976291621CC60BC00B52A68 /* spirv_cross.cpp in Sources */,
  450A4F68221C5A95007203D7 /* spirv_reflect.cpp in Sources */,
  A976291821CC60BC00B52A68 /* spirv_glsl.cpp in Sources */,
+ 164DCE0A2AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */,
  A976292821CC60BC00B52A68 /* spirv_cross_parsed_ir.cpp in Sources */,
  );
  runOnlyForDeploymentPostprocessing = 0;
@@ -6399,6 +6407,7 @@
  A976291521CC60BC00B52A68 /* spirv_cross.cpp in Sources */,
  450A4F67221C5A95007203D7 /* spirv_reflect.cpp in Sources */,
  A976291721CC60BC00B52A68 /* spirv_glsl.cpp in Sources */,
+ 164DCE072AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */,
  A976292721CC60BC00B52A68 /* spirv_cross_parsed_ir.cpp in Sources */,
  );
  runOnlyForDeploymentPostprocessing = 0;
@@ -6413,6 +6422,7 @@
  DCFD7C9B2A45BA7D007BBBF7 /* spirv_cross.cpp in Sources */,
  DCFD7C9C2A45BA7D007BBBF7 /* spirv_reflect.cpp in Sources */,
  DCFD7C9D2A45BA7D007BBBF7 /* spirv_glsl.cpp in Sources */,
+ 164DCE092AABA74B00E4B88F /* spirv_msl_vertex_loader.cpp in Sources */,
  DCFD7C9E2A45BA7D007BBBF7 /* spirv_cross_parsed_ir.cpp in Sources */,
  );
  runOnlyForDeploymentPostprocessing = 0;

ExternalRevisions/README.md

@@ -133,6 +133,7 @@ if you encounter any building errors, you may need to re-add the *SPIRV-Cross* l
  spirv_cross.hpp
  spirv_glsl.cpp
  spirv_glsl.hpp
+ spirv_msl_vertex_loader.cpp
  spirv_msl.cpp
  spirv_msl.hpp
  spirv_parser.cpp

ExternalRevisions/SPIRV-Cross_repo_revision

@@ -1 +1 @@
-bccaa94db814af33d8ef05c153e7c34d8bd4d685
+324b9393f2a40d4883d280669886e701f12e22a6

MoltenVK/MoltenVK/API/mvk_config.h

@@ -961,6 +961,17 @@ typedef struct {
  */
  float timestampPeriodLowPassAlpha;
 
+ /**
+ * Force MoltenVK to always use shader code to load vertices (rather than translate to Metal's vertex descriptors).
+ * (Default is to only use shader vertex loaders in situations that Metal's descriptors don't support.)
+ *
+ * The initial value or this parameter is set by the
+ * MVK_CONFIG_FORCE_SHADER_VERTEX_LOADER
+ * runtime environment variable or MoltenVK compile-time build setting.
+ * If neither is set, this setting is disabled by default.
+ */
+ VkBool32 forceShaderVertexLoader;
+
 } MVKConfiguration;
 
 

MoltenVK/MoltenVK/API/mvk_private_api.h

@@ -149,6 +149,7 @@ typedef struct {
  VkBool32 programmableSamplePositions; /**< If true, programmable MSAA sample positions are supported. */
  VkBool32 shaderBarycentricCoordinates; /**< If true, fragment shader barycentric coordinates are supported. */
  MTLArgumentBuffersTier argumentBuffersTier; /**< The argument buffer tier available on this device, as a Metal enumeration. */
+ VkBool32 pixelTypeLoads; /**< Supports loading from types like `device rgb9e5<float3>*` in MSL */
  VkBool32 needsSampleDrefLodArrayWorkaround; /**< If true, sampling from arrayed depth images with explicit LoD is broken and needs a workaround. */
  VkDeviceSize hostMemoryPageSize; /**< The size of a page of host memory on this platform. */
 } MVKPhysicalDeviceMetalFeatures;

MoltenVK/MoltenVK/Commands/MVKCmdDraw.mm

@@ -191,9 +191,6 @@
  atIndex: pipeline->getOutputBufferIndex().stages[kMVKShaderStageVertex]];
  }
  [mtlTessCtlEncoder setStageInRegion: MTLRegionMake2D(_firstVertex, _firstInstance, _vertexCount, _instanceCount)];
- // If there are vertex bindings with a zero vertex divisor, I need to offset them by
- // _firstInstance * stride, since that is the expected behaviour for a divisor of 0.
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _firstInstance);
  id<MTLComputePipelineState> vtxState = pipeline->getTessVertexStageState();
  if (cmdEncoder->getDevice()->_pMetalFeatures->nonUniformThreadgroups) {
 #if MVK_MACOS_OR_IOS
@@ -294,7 +291,6 @@
  MVKRenderSubpass* subpass = cmdEncoder->getSubpass();
  uint32_t viewCount = subpass->isMultiview() ? subpass->getViewCountInMetalPass(cmdEncoder->getMultiviewPassIndex()) : 1;
  uint32_t instanceCount = _instanceCount * viewCount;
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _firstInstance);
  if (cmdEncoder->_pDeviceMetalFeatures->baseVertexInstanceDrawing) {
  [cmdEncoder->_mtlRenderEncoder drawPrimitives: cmdEncoder->_mtlPrimitiveType
  vertexStart: _firstVertex
@@ -422,9 +418,6 @@
  offset: idxBuffOffset
  atIndex: pipeline->getIndirectParamsIndex().stages[kMVKShaderStageVertex]];
  [mtlTessCtlEncoder setStageInRegion: MTLRegionMake2D(_vertexOffset, _firstInstance, _indexCount, _instanceCount)];
- // If there are vertex bindings with a zero vertex divisor, I need to offset them by
- // _firstInstance * stride, since that is the expected behaviour for a divisor of 0.
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _firstInstance);
  id<MTLComputePipelineState> vtxState = ibb.mtlIndexType == MTLIndexTypeUInt16 ? pipeline->getTessVertexStageIndex16State() : pipeline->getTessVertexStageIndex32State();
  if (cmdEncoder->getDevice()->_pMetalFeatures->nonUniformThreadgroups) {
 #if MVK_MACOS_OR_IOS
@@ -528,7 +521,6 @@
  MVKRenderSubpass* subpass = cmdEncoder->getSubpass();
  uint32_t viewCount = subpass->isMultiview() ? subpass->getViewCountInMetalPass(cmdEncoder->getMultiviewPassIndex()) : 1;
  uint32_t instanceCount = _instanceCount * viewCount;
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _firstInstance);
  if (cmdEncoder->_pDeviceMetalFeatures->baseVertexInstanceDrawing) {
  [cmdEncoder->_mtlRenderEncoder drawIndexedPrimitives: cmdEncoder->_mtlPrimitiveType
  indexCount: _indexCount
@@ -1230,9 +1222,6 @@
  indirectBufferOffset: mtlTempIndBuffOfst];
  mtlTempIndBuffOfst += sizeof(MTLStageInRegionIndirectArguments);
  }
- // If this is a synthetic command that originated in a direct call, and there are vertex bindings with a zero vertex
- // divisor, I need to offset them by _firstInstance * stride, since that is the expected behaviour for a divisor of 0.
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _directCmdFirstInstance);
  [mtlTessCtlEncoder dispatchThreadgroupsWithIndirectBuffer: mtlIndBuff
  indirectBufferOffset: mtlTempIndBuffOfst
  threadsPerThreadgroup: MTLSizeMake(vtxThreadExecWidth, 1, 1)];
@@ -1311,7 +1300,6 @@
  cmdEncoder->_graphicsResourcesState.beginMetalRenderPass();
  cmdEncoder->getPushConstants(VK_SHADER_STAGE_VERTEX_BIT)->beginMetalRenderPass();
  } else {
- cmdEncoder->_graphicsResourcesState.offsetZeroDivisorVertexBuffers(stage, pipeline, _directCmdFirstInstance);
  [cmdEncoder->_mtlRenderEncoder drawIndexedPrimitives: cmdEncoder->_mtlPrimitiveType
  indexType: (MTLIndexType)ibb.mtlIndexType
  indexBuffer: ibb.mtlBuffer

MoltenVK/MoltenVK/Commands/MVKCommandEncoderState.h

@@ -556,9 +556,6 @@ class MVKGraphicsResourcesCommandEncoderState : public MVKResourcesCommandEncode
  MTLResourceUsage mtlUsage,
  MTLRenderStages mtlStages) override;
 
- /** Offset all buffers for vertex attribute bindings with zero divisors by the given number of strides. */
- void offsetZeroDivisorVertexBuffers(MVKGraphicsStage stage, MVKGraphicsPipeline* pipeline, uint32_t firstInstance);
-
  /**
  * Marks the buffer binding using the index as having been overridden,
  * such as by push constants or internal rendering in some transfers.

MoltenVK/MoltenVK/Commands/MVKCommandEncoderState.mm

@@ -733,30 +733,6 @@
  encodeBinding<MVKMTLSamplerStateBinding>(shaderStage.samplerStateBindings, shaderStage.areSamplerStateBindingsDirty, bindSampler);
 }
 
-void MVKGraphicsResourcesCommandEncoderState::offsetZeroDivisorVertexBuffers(MVKGraphicsStage stage,
- MVKGraphicsPipeline* pipeline,
- uint32_t firstInstance) {
- auto& shaderStage = _shaderStageResourceBindings[kMVKShaderStageVertex];
- for (auto& binding : pipeline->getZeroDivisorVertexBindings()) {
- uint32_t mtlBuffIdx = pipeline->getMetalBufferIndexForVertexAttributeBinding(binding.first);
- auto iter = std::find_if(shaderStage.bufferBindings.begin(), shaderStage.bufferBindings.end(), [mtlBuffIdx](const MVKMTLBufferBinding& b) { return b.index == mtlBuffIdx; });
- if (!iter) { continue; }
- switch (stage) {
- case kMVKGraphicsStageVertex:
- [_cmdEncoder->getMTLComputeEncoder(kMVKCommandUseTessellationVertexTessCtl) setBufferOffset: iter->offset + firstInstance * binding.second
- atIndex: mtlBuffIdx];
- break;
- case kMVKGraphicsStageRasterization:
- [_cmdEncoder->_mtlRenderEncoder setVertexBufferOffset: iter->offset + firstInstance * binding.second
- atIndex: mtlBuffIdx];
- break;
- default:
- assert(false); // If we hit this, something went wrong.
- break;
- }
- }
-}
-
 void MVKGraphicsResourcesCommandEncoderState::endMetalRenderPass() {
  MVKResourcesCommandEncoderState::endMetalRenderPass();
  _renderUsageStages.clear();
@@ -831,16 +807,6 @@
  offset: b.offset
  atIndex: b.index];
  }
-
- // Add any translated vertex bindings for this binding
- auto xltdVtxBindings = pipeline->getTranslatedVertexBindings();
- for (auto& xltdBind : xltdVtxBindings) {
- if (b.index == pipeline->getMetalBufferIndexForVertexAttributeBinding(xltdBind.binding)) {
- [cmdEncoder->_mtlRenderEncoder setVertexBuffer: b.mtlBuffer
- offset: b.offset + xltdBind.translationOffset
- atIndex: pipeline->getMetalBufferIndexForVertexAttributeBinding(xltdBind.translationBinding)];
- }
- }
  }
  } else {
  b.isDirty = true; // We haven't written it out, so leave dirty until next time.

MoltenVK/MoltenVK/GPUObjects/MVKDevice.h

@@ -131,6 +131,9 @@ class MVKPhysicalDevice : public MVKDispatchableVulkanAPIObject {
  /** Populates the specified structure with the features of this device. */
  void getFeatures(VkPhysicalDeviceFeatures2* features);
 
+ /** Returns a structure with the features of this device */
+ const VkPhysicalDeviceFeatures& getFeatures() const { return _features; }
+
  /** Populates the specified structure with the properties of this device. */
  void getProperties(VkPhysicalDeviceProperties* properties);
 

MoltenVK/MoltenVK/GPUObjects/MVKDevice.mm

@@ -717,7 +717,7 @@
  }
  case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PORTABILITY_SUBSET_PROPERTIES_KHR: {
  auto* portabilityProps = (VkPhysicalDevicePortabilitySubsetPropertiesKHR*)next;
- portabilityProps->minVertexInputBindingStrideAlignment = (uint32_t)_metalFeatures.vertexStrideAlignment;
+ portabilityProps->minVertexInputBindingStrideAlignment = 1; // Shader vertex loader has no restrictions
  break;
  }
  case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTERNAL_MEMORY_HOST_PROPERTIES_EXT: {
@@ -1642,6 +1642,7 @@
  initMetalFeatures(); // Call second.
  initFeatures(); // Call third.
  initLimits(); // Call fourth.
+ _pixelFormats.modifyCapabilitiesFromPhysicalDevice();
  initExtensions();
  initMemoryProperties();
  initExternalMemoryProperties();
@@ -1861,6 +1862,7 @@
  if (supportsMTLFeatureSet(iOS_GPUFamily4_v1)) {
  _metalFeatures.postDepthCoverage = true;
  _metalFeatures.nonUniformThreadgroups = true;
+ _metalFeatures.pixelTypeLoads = true;
  }
 
  if (supportsMTLFeatureSet(iOS_GPUFamily5_v1)) {
@@ -2025,6 +2027,9 @@
  _metalFeatures.renderLinearTextures = true;
  _metalFeatures.tileBasedDeferredRendering = true;
 
+ if (supportsMTLGPUFamily(Apple4)) {
+ _metalFeatures.pixelTypeLoads = true;
+ }
 #if MVK_XCODE_12
  if (supportsMTLGPUFamily(Apple6)) {
  _metalFeatures.astcHDRTextures = true;

MoltenVK/MoltenVK/GPUObjects/MVKPipeline.h

@@ -292,12 +292,6 @@ class MVKGraphicsPipeline : public MVKPipeline {
  /** Returns the Metal vertex buffer index to use for the specified vertex attribute binding number. */
  uint32_t getMetalBufferIndexForVertexAttributeBinding(uint32_t binding) { return _device->getMetalBufferIndexForVertexAttributeBinding(binding); }
 
- /** Returns the collection of translated vertex bindings. */
- MVKArrayRef<MVKTranslatedVertexBinding> getTranslatedVertexBindings() { return _translatedVertexBindings.contents(); }
-
- /** Returns the collection of instance-rate vertex bindings whose divisor is zero, along with their strides. */
- MVKArrayRef<MVKZeroDivisorVertexBinding> getZeroDivisorVertexBindings() { return _zeroDivisorVertexBindings.contents(); }
-
  /** Returns the MTLArgumentEncoder for the descriptor set. */
  MVKMTLArgumentEncoder& getMTLArgumentEncoder(uint32_t descSetIndex, MVKShaderStage stage) override { return _mtlArgumentEncoders[descSetIndex].stages[stage]; }
 
@@ -336,6 +330,7 @@ class MVKGraphicsPipeline : public MVKPipeline {
  bool addTessCtlShaderToPipeline(MTLComputePipelineDescriptor* plDesc, const VkGraphicsPipelineCreateInfo* pCreateInfo, SPIRVToMSLConversionConfiguration& shaderConfig, SPIRVShaderOutputs& prevOutput, SPIRVShaderInputs& nextInputs, const VkPipelineShaderStageCreateInfo* pTessCtlSS, VkPipelineCreationFeedback* pTessCtlFB);
  bool addTessEvalShaderToPipeline(MTLRenderPipelineDescriptor* plDesc, const VkGraphicsPipelineCreateInfo* pCreateInfo, SPIRVToMSLConversionConfiguration& shaderConfig, SPIRVShaderOutputs& prevOutput, const VkPipelineShaderStageCreateInfo* pTessEvalSS, VkPipelineCreationFeedback* pTessEvalFB, const VkPipelineShaderStageCreateInfo*& pFragmentSS);
  bool addFragmentShaderToPipeline(MTLRenderPipelineDescriptor* plDesc, const VkGraphicsPipelineCreateInfo* pCreateInfo, SPIRVToMSLConversionConfiguration& shaderConfig, SPIRVShaderOutputs& prevOutput, const VkPipelineShaderStageCreateInfo* pFragmentSS, VkPipelineCreationFeedback* pFragmentFB);
+ bool canVertexInputUseMetalDescriptor(const VkPipelineVertexInputStateCreateInfo* pVI);
  template<class T>
  bool addVertexInputToPipeline(T* inputDesc, const VkPipelineVertexInputStateCreateInfo* pVI, const SPIRVToMSLConversionConfiguration& shaderConfig);
  void adjustVertexInputForMultiview(MTLVertexDescriptor* inputDesc, const VkPipelineVertexInputStateCreateInfo* pVI, uint32_t viewCount, uint32_t oldViewCount = 1);
@@ -344,7 +339,6 @@ class MVKGraphicsPipeline : public MVKPipeline {
  bool isRenderingPoints(const VkGraphicsPipelineCreateInfo* pCreateInfo);
  bool isRasterizationDisabled(const VkGraphicsPipelineCreateInfo* pCreateInfo);
  bool verifyImplicitBuffer(bool needsBuffer, MVKShaderImplicitRezBinding& index, MVKShaderStage stage, const char* name);
- uint32_t getTranslatedVertexBinding(uint32_t binding, uint32_t translationOffset, uint32_t maxBinding);
  uint32_t getImplicitBufferIndex(MVKShaderStage stage, uint32_t bufferIndexOffset);
  MVKMTLFunction getMTLFunction(SPIRVToMSLConversionConfiguration& shaderConfig,
  const VkPipelineShaderStageCreateInfo* pShaderStage,
@@ -361,8 +355,6 @@ class MVKGraphicsPipeline : public MVKPipeline {
  MVKSmallVector<VkRect2D, kMVKMaxViewportScissorCount> _scissors;
  MVKSmallVector<VkDynamicState> _dynamicState;
  MVKSmallVector<MTLSamplePosition> _customSamplePositions;
- MVKSmallVector<MVKTranslatedVertexBinding> _translatedVertexBindings;
- MVKSmallVector<MVKZeroDivisorVertexBinding> _zeroDivisorVertexBindings;
  MVKSmallVector<MVKStagedMTLArgumentEncoders> _mtlArgumentEncoders;
  MVKSmallVector<MVKStagedDescriptorBindingUse> _descriptorBindingUse;
  MVKSmallVector<MVKShaderStage> _stagesUsingPhysicalStorageBufferAddressesCapability;