/* * Copyright 2015 Google Inc. * * Use of this source code is governed by a BSD-style license that can be * found in the LICENSE file. */ #include "src/gpu/vk/GrVkCaps.h" #include #include "include/gpu/GrBackendSurface.h" #include "include/gpu/vk/GrVkBackendContext.h" #include "include/gpu/vk/GrVkExtensions.h" #include "src/core/SkCompressedDataUtils.h" #include "src/gpu/GrBackendUtils.h" #include "src/gpu/GrProgramDesc.h" #include "src/gpu/GrRenderTarget.h" #include "src/gpu/GrRenderTargetProxy.h" #include "src/gpu/GrShaderCaps.h" #include "src/gpu/GrStencilSettings.h" #include "src/gpu/GrUtil.h" #include "src/gpu/SkGr.h" #include "src/gpu/vk/GrVkGpu.h" #include "src/gpu/vk/GrVkImage.h" #include "src/gpu/vk/GrVkInterface.h" #include "src/gpu/vk/GrVkRenderTarget.h" #include "src/gpu/vk/GrVkTexture.h" #include "src/gpu/vk/GrVkUniformHandler.h" #include "src/gpu/vk/GrVkUtil.h" #ifdef SK_BUILD_FOR_ANDROID #include #endif GrVkCaps::GrVkCaps(const GrContextOptions& contextOptions, const GrVkInterface* vkInterface, VkPhysicalDevice physDev, const VkPhysicalDeviceFeatures2& features, uint32_t instanceVersion, uint32_t physicalDeviceVersion, const GrVkExtensions& extensions, GrProtected isProtected) : INHERITED(contextOptions) { /************************************************************************** * GrCaps fields **************************************************************************/ fMipmapSupport = true; // always available in Vulkan fNPOTTextureTileSupport = true; // always available in Vulkan fReuseScratchTextures = true; //TODO: figure this out fGpuTracingSupport = false; //TODO: figure this out fOversizedStencilSupport = false; //TODO: figure this out fDrawInstancedSupport = true; fSemaphoreSupport = true; // always available in Vulkan fFenceSyncSupport = true; // always available in Vulkan fCrossContextTextureSupport = true; fHalfFloatVertexAttributeSupport = true; // We always copy in/out of a transfer buffer so it's trivial to support row bytes. fReadPixelsRowBytesSupport = true; fWritePixelsRowBytesSupport = true; fTransferFromBufferToTextureSupport = true; fTransferFromSurfaceToBufferSupport = true; fMaxRenderTargetSize = 4096; // minimum required by spec fMaxTextureSize = 4096; // minimum required by spec fDynamicStateArrayGeometryProcessorTextureSupport = true; fTextureBarrierSupport = true; fShaderCaps = std::make_unique(); this->init(contextOptions, vkInterface, physDev, features, physicalDeviceVersion, extensions, isProtected); } namespace { /** * This comes from section 37.1.6 of the Vulkan spec. Format is * (|)__. */ enum class FormatCompatibilityClass { k8_1_1, k16_2_1, k24_3_1, k32_4_1, k64_8_1, kBC1_RGB_8_16_1, kBC1_RGBA_8_16, kETC2_RGB_8_16, }; } // anonymous namespace static FormatCompatibilityClass format_compatibility_class(VkFormat format) { switch (format) { case VK_FORMAT_B8G8R8A8_UNORM: case VK_FORMAT_R8G8B8A8_UNORM: case VK_FORMAT_A2B10G10R10_UNORM_PACK32: case VK_FORMAT_A2R10G10B10_UNORM_PACK32: case VK_FORMAT_R8G8B8A8_SRGB: case VK_FORMAT_R16G16_UNORM: case VK_FORMAT_R16G16_SFLOAT: return FormatCompatibilityClass::k32_4_1; case VK_FORMAT_R8_UNORM: return FormatCompatibilityClass::k8_1_1; case VK_FORMAT_R5G6B5_UNORM_PACK16: case VK_FORMAT_R16_SFLOAT: case VK_FORMAT_R8G8_UNORM: case VK_FORMAT_B4G4R4A4_UNORM_PACK16: case VK_FORMAT_R4G4B4A4_UNORM_PACK16: case VK_FORMAT_R16_UNORM: return FormatCompatibilityClass::k16_2_1; case VK_FORMAT_R16G16B16A16_SFLOAT: case VK_FORMAT_R16G16B16A16_UNORM: return FormatCompatibilityClass::k64_8_1; case VK_FORMAT_R8G8B8_UNORM: return FormatCompatibilityClass::k24_3_1; case VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK: return FormatCompatibilityClass::kETC2_RGB_8_16; case VK_FORMAT_BC1_RGB_UNORM_BLOCK: return FormatCompatibilityClass::kBC1_RGB_8_16_1; case VK_FORMAT_BC1_RGBA_UNORM_BLOCK: return FormatCompatibilityClass::kBC1_RGBA_8_16; default: SK_ABORT("Unsupported VkFormat"); } } bool GrVkCaps::canCopyImage(VkFormat dstFormat, int dstSampleCnt, bool dstHasYcbcr, VkFormat srcFormat, int srcSampleCnt, bool srcHasYcbcr) const { if ((dstSampleCnt > 1 || srcSampleCnt > 1) && dstSampleCnt != srcSampleCnt) { return false; } if (dstHasYcbcr || srcHasYcbcr) { return false; } // We require that all Vulkan GrSurfaces have been created with transfer_dst and transfer_src // as image usage flags. return format_compatibility_class(srcFormat) == format_compatibility_class(dstFormat); } bool GrVkCaps::canCopyAsBlit(VkFormat dstFormat, int dstSampleCnt, bool dstIsLinear, bool dstHasYcbcr, VkFormat srcFormat, int srcSampleCnt, bool srcIsLinear, bool srcHasYcbcr) const { // We require that all vulkan GrSurfaces have been created with transfer_dst and transfer_src // as image usage flags. if (!this->formatCanBeDstofBlit(dstFormat, dstIsLinear) || !this->formatCanBeSrcofBlit(srcFormat, srcIsLinear)) { return false; } // We cannot blit images that are multisampled. Will need to figure out if we can blit the // resolved msaa though. if (dstSampleCnt > 1 || srcSampleCnt > 1) { return false; } if (dstHasYcbcr || srcHasYcbcr) { return false; } return true; } bool GrVkCaps::canCopyAsResolve(VkFormat dstFormat, int dstSampleCnt, bool dstHasYcbcr, VkFormat srcFormat, int srcSampleCnt, bool srcHasYcbcr) const { // The src surface must be multisampled. if (srcSampleCnt <= 1) { return false; } // The dst must not be multisampled. if (dstSampleCnt > 1) { return false; } // Surfaces must have the same format. if (srcFormat != dstFormat) { return false; } if (dstHasYcbcr || srcHasYcbcr) { return false; } return true; } bool GrVkCaps::onCanCopySurface(const GrSurfaceProxy* dst, const GrSurfaceProxy* src, const SkIRect& srcRect, const SkIPoint& dstPoint) const { if (src->isProtected() == GrProtected::kYes && dst->isProtected() != GrProtected::kYes) { return false; } // TODO: Figure out a way to track if we've wrapped a linear texture in a proxy (e.g. // PromiseImage which won't get instantiated right away. Does this need a similar thing like the // tracking of external or rectangle textures in GL? For now we don't create linear textures // internally, and I don't believe anyone is wrapping them. bool srcIsLinear = false; bool dstIsLinear = false; int dstSampleCnt = 0; int srcSampleCnt = 0; if (const GrRenderTargetProxy* rtProxy = dst->asRenderTargetProxy()) { // Copying to or from render targets that wrap a secondary command buffer is not allowed // since they would require us to know the VkImage, which we don't have, as well as need us // to stop and start the VkRenderPass which we don't have access to. if (rtProxy->wrapsVkSecondaryCB()) { return false; } if (this->preferDiscardableMSAAAttachment() && dst->asTextureProxy() && rtProxy->supportsVkInputAttachment()) { dstSampleCnt = 1; } else { dstSampleCnt = rtProxy->numSamples(); } } if (const GrRenderTargetProxy* rtProxy = src->asRenderTargetProxy()) { // Copying to or from render targets that wrap a secondary command buffer is not allowed // since they would require us to know the VkImage, which we don't have, as well as need us // to stop and start the VkRenderPass which we don't have access to. if (rtProxy->wrapsVkSecondaryCB()) { return false; } if (this->preferDiscardableMSAAAttachment() && src->asTextureProxy() && rtProxy->supportsVkInputAttachment()) { srcSampleCnt = 1; } else { srcSampleCnt = rtProxy->numSamples(); } } SkASSERT((dstSampleCnt > 0) == SkToBool(dst->asRenderTargetProxy())); SkASSERT((srcSampleCnt > 0) == SkToBool(src->asRenderTargetProxy())); bool dstHasYcbcr = false; if (auto ycbcr = dst->backendFormat().getVkYcbcrConversionInfo()) { if (ycbcr->isValid()) { dstHasYcbcr = true; } } bool srcHasYcbcr = false; if (auto ycbcr = src->backendFormat().getVkYcbcrConversionInfo()) { if (ycbcr->isValid()) { srcHasYcbcr = true; } } VkFormat dstFormat, srcFormat; SkAssertResult(dst->backendFormat().asVkFormat(&dstFormat)); SkAssertResult(src->backendFormat().asVkFormat(&srcFormat)); return this->canCopyImage(dstFormat, dstSampleCnt, dstHasYcbcr, srcFormat, srcSampleCnt, srcHasYcbcr) || this->canCopyAsBlit(dstFormat, dstSampleCnt, dstIsLinear, dstHasYcbcr, srcFormat, srcSampleCnt, srcIsLinear, srcHasYcbcr) || this->canCopyAsResolve(dstFormat, dstSampleCnt, dstHasYcbcr, srcFormat, srcSampleCnt, srcHasYcbcr); } template T* get_extension_feature_struct(const VkPhysicalDeviceFeatures2& features, VkStructureType type) { // All Vulkan structs that could be part of the features chain will start with the // structure type followed by the pNext pointer. We cast to the CommonVulkanHeader // so we can get access to the pNext for the next struct. struct CommonVulkanHeader { VkStructureType sType; void* pNext; }; void* pNext = features.pNext; while (pNext) { CommonVulkanHeader* header = static_cast(pNext); if (header->sType == type) { return static_cast(pNext); } pNext = header->pNext; } return nullptr; } void GrVkCaps::init(const GrContextOptions& contextOptions, const GrVkInterface* vkInterface, VkPhysicalDevice physDev, const VkPhysicalDeviceFeatures2& features, uint32_t physicalDeviceVersion, const GrVkExtensions& extensions, GrProtected isProtected) { VkPhysicalDeviceProperties properties; GR_VK_CALL(vkInterface, GetPhysicalDeviceProperties(physDev, &properties)); VkPhysicalDeviceMemoryProperties memoryProperties; GR_VK_CALL(vkInterface, GetPhysicalDeviceMemoryProperties(physDev, &memoryProperties)); SkASSERT(physicalDeviceVersion <= properties.apiVersion); if (extensions.hasExtension(VK_KHR_SWAPCHAIN_EXTENSION_NAME, 1)) { fSupportsSwapchain = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_GET_PHYSICAL_DEVICE_PROPERTIES_2_EXTENSION_NAME, 1)) { fSupportsPhysicalDeviceProperties2 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_GET_MEMORY_REQUIREMENTS_2_EXTENSION_NAME, 1)) { fSupportsMemoryRequirements2 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_BIND_MEMORY_2_EXTENSION_NAME, 1)) { fSupportsBindMemory2 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_MAINTENANCE1_EXTENSION_NAME, 1)) { fSupportsMaintenance1 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_MAINTENANCE2_EXTENSION_NAME, 1)) { fSupportsMaintenance2 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || extensions.hasExtension(VK_KHR_MAINTENANCE3_EXTENSION_NAME, 1)) { fSupportsMaintenance3 = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || (extensions.hasExtension(VK_KHR_DEDICATED_ALLOCATION_EXTENSION_NAME, 1) && this->supportsMemoryRequirements2())) { fSupportsDedicatedAllocation = true; } if (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || (extensions.hasExtension(VK_KHR_EXTERNAL_MEMORY_CAPABILITIES_EXTENSION_NAME, 1) && this->supportsPhysicalDeviceProperties2() && extensions.hasExtension(VK_KHR_EXTERNAL_MEMORY_EXTENSION_NAME, 1) && this->supportsDedicatedAllocation())) { fSupportsExternalMemory = true; } #ifdef SK_BUILD_FOR_ANDROID // Currently Adreno devices are not supporting the QUEUE_FAMILY_FOREIGN_EXTENSION, so until they // do we don't explicitly require it here even the spec says it is required. if (extensions.hasExtension( VK_ANDROID_EXTERNAL_MEMORY_ANDROID_HARDWARE_BUFFER_EXTENSION_NAME, 2) && /* extensions.hasExtension(VK_EXT_QUEUE_FAMILY_FOREIGN_EXTENSION_NAME, 1) &&*/ this->supportsExternalMemory() && this->supportsBindMemory2()) { fSupportsAndroidHWBExternalMemory = true; fSupportsAHardwareBufferImages = true; } #endif auto ycbcrFeatures = get_extension_feature_struct( features, VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_YCBCR_CONVERSION_FEATURES); if (ycbcrFeatures && ycbcrFeatures->samplerYcbcrConversion && (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0) || (extensions.hasExtension(VK_KHR_SAMPLER_YCBCR_CONVERSION_EXTENSION_NAME, 1) && this->supportsMaintenance1() && this->supportsBindMemory2() && this->supportsMemoryRequirements2() && this->supportsPhysicalDeviceProperties2()))) { fSupportsYcbcrConversion = true; } // We always push back the default GrVkYcbcrConversionInfo so that the case of no conversion // will return a key of 0. fYcbcrInfos.push_back(GrVkYcbcrConversionInfo()); if ((isProtected == GrProtected::kYes) && (physicalDeviceVersion >= VK_MAKE_VERSION(1, 1, 0))) { fSupportsProtectedMemory = true; fAvoidUpdateBuffers = true; fShouldAlwaysUseDedicatedImageMemory = true; } if (extensions.hasExtension(VK_EXT_IMAGE_DRM_FORMAT_MODIFIER_EXTENSION_NAME, 1)) { fSupportsDRMFormatModifiers = true; } fMaxInputAttachmentDescriptors = properties.limits.maxDescriptorSetInputAttachments; // On desktop GPUs we have found that this does not provide much benefit. The perf results show // a mix of regressions, some improvements, and lots of no changes. Thus it is no worth enabling // this (especially with the rendering artifacts) on desktop. // // On Adreno devices we were expecting to see perf gains. But instead there were actually a lot // of perf regressions and only a few perf wins. This needs some follow up with qualcomm since // we do expect this to be a big win on tilers. // // On ARM devices we are seeing an average perf win of around 50%-60% across the board. if (kARM_VkVendor == properties.vendorID) { fPreferDiscardableMSAAAttachment = true; fSupportsMemorylessAttachments = true; } this->initGrCaps(vkInterface, physDev, properties, memoryProperties, features, extensions); this->initShaderCaps(properties, features); if (kQualcomm_VkVendor == properties.vendorID) { // A "clear" load for atlases runs faster on QC than a "discard" load followed by a // scissored clear. // On NVIDIA and Intel, the discard load followed by clear is faster. // TODO: Evaluate on ARM, Imagination, and ATI. fPreferFullscreenClears = true; } if (properties.vendorID == kNvidia_VkVendor || properties.vendorID == kAMD_VkVendor) { // On discrete GPUs it can be faster to read gpu only memory compared to memory that is also // mappable on the host. fGpuOnlyBuffersMorePerformant = true; // On discrete GPUs we try to use special DEVICE_LOCAL and HOST_VISIBLE memory for our // cpu write, gpu read buffers. This memory is not ideal to be kept persistently mapped. // Some discrete GPUs do not expose this special memory, however we still disable // persistently mapped buffers for all of them since most GPUs with updated drivers do // expose it. If this becomes an issue we can try to be more fine grained. fShouldPersistentlyMapCpuToGpuBuffers = false; } if (kQualcomm_VkVendor == properties.vendorID) { // On Qualcomm it looks like using vkCmdUpdateBuffer is slower than using a transfer buffer // even for small sizes. fAvoidUpdateBuffers = true; } if (kQualcomm_VkVendor == properties.vendorID) { // Adreno devices don't support push constants well fMaxPushConstantsSize = 0; } fNativeDrawIndirectSupport = features.features.drawIndirectFirstInstance; if (properties.vendorID == kQualcomm_VkVendor) { // Indirect draws seem slow on QC. Disable until we can investigate. http://skbug.com/11139 fNativeDrawIndirectSupport = false; } if (fNativeDrawIndirectSupport) { fMaxDrawIndirectDrawCount = properties.limits.maxDrawIndirectCount; SkASSERT(fMaxDrawIndirectDrawCount == 1 || features.features.multiDrawIndirect); } #ifdef SK_BUILD_FOR_UNIX if (kNvidia_VkVendor == properties.vendorID) { // On nvidia linux we see a big perf regression when not using dedicated image allocations. fShouldAlwaysUseDedicatedImageMemory = true; } #endif this->initFormatTable(vkInterface, physDev, properties); this->initStencilFormat(vkInterface, physDev); if (contextOptions.fMaxCachedVulkanSecondaryCommandBuffers >= 0) { fMaxPerPoolCachedSecondaryCommandBuffers = contextOptions.fMaxCachedVulkanSecondaryCommandBuffers; } if (!contextOptions.fDisableDriverCorrectnessWorkarounds) { this->applyDriverCorrectnessWorkarounds(properties); } this->finishInitialization(contextOptions); } void GrVkCaps::applyDriverCorrectnessWorkarounds(const VkPhysicalDeviceProperties& properties) { #if defined(SK_BUILD_FOR_WIN) if (kNvidia_VkVendor == properties.vendorID || kIntel_VkVendor == properties.vendorID) { fMustSyncCommandBuffersWithQueue = true; } #elif defined(SK_BUILD_FOR_ANDROID) if (kImagination_VkVendor == properties.vendorID) { fMustSyncCommandBuffersWithQueue = true; } #endif // Defaults to zero since all our workaround checks that use this consider things "fixed" once // above a certain api level. So this will just default to it being less which will enable // workarounds. int androidAPIVersion = 0; #if defined(SK_BUILD_FOR_ANDROID) char androidAPIVersionStr[PROP_VALUE_MAX]; int strLength = __system_property_get("ro.build.version.sdk", androidAPIVersionStr); // Defaults to zero since most checks care if it is greater than a specific value. So this will // just default to it being less. androidAPIVersion = (strLength == 0) ? 0 : atoi(androidAPIVersionStr); #endif // Protected memory features have problems in Android P and earlier. if (fSupportsProtectedMemory && (kQualcomm_VkVendor == properties.vendorID)) { if (androidAPIVersion <= 28) { fSupportsProtectedMemory = false; } } // On Mali galaxy s7 we see lots of rendering issues when we suballocate VkImages. if (kARM_VkVendor == properties.vendorID && androidAPIVersion <= 28) { fShouldAlwaysUseDedicatedImageMemory = true; } // On Mali galaxy s7 and s9 we see lots of rendering issues with image filters dropping out when // using only primary command buffers. We also see issues on the P30 running android 28. if (kARM_VkVendor == properties.vendorID && androidAPIVersion <= 28) { fPreferPrimaryOverSecondaryCommandBuffers = false; // If we are using secondary command buffers our code isn't setup to insert barriers into // the secondary cb so we need to disable support for them. fTextureBarrierSupport = false; fBlendEquationSupport = kBasic_BlendEquationSupport; } // We've seen numerous driver bugs on qualcomm devices running on android P (api 28) or earlier // when trying to using discardable msaa attachments and loading from resolve. So we disable the // feature for those devices. if (properties.vendorID == kQualcomm_VkVendor && androidAPIVersion <= 28) { fPreferDiscardableMSAAAttachment = false; fSupportsDiscardableMSAAForDMSAA = false; } // On Mali G series GPUs, applying transfer functions in the fragment shader with half-floats // produces answers that are much less accurate than expected/required. This forces full floats // for some intermediate values to get acceptable results. if (kARM_VkVendor == properties.vendorID) { fShaderCaps->fColorSpaceMathNeedsFloat = true; } // On various devices, when calling vkCmdClearAttachments on a primary command buffer, it // corrupts the bound buffers on the command buffer. As a workaround we invalidate our knowledge // of bound buffers so that we will rebind them on the next draw. if (kQualcomm_VkVendor == properties.vendorID || kAMD_VkVendor == properties.vendorID) { fMustInvalidatePrimaryCmdBufferStateAfterClearAttachments = true; } // On Qualcomm and Arm the gpu resolves an area larger than the render pass bounds when using // discardable msaa attachments. This causes the resolve to resolve uninitialized data from the // msaa image into the resolve image. if (kQualcomm_VkVendor == properties.vendorID || kARM_VkVendor == properties.vendorID) { fMustLoadFullImageWithDiscardableMSAA = true; } #ifdef SK_BUILD_FOR_UNIX if (kIntel_VkVendor == properties.vendorID) { // At least on our linux Debug Intel HD405 bot we are seeing issues doing read pixels with // non-conherent memory. It seems like the device is not properly honoring the // vkInvalidateMappedMemoryRanges calls correctly. Other linux intel devices seem to work // okay. However, since I'm not sure how to target a specific intel devices or driver // version I am going to stop all intel linux from using non-coherent memory. Currently we // are not shipping anything on these platforms and the only real thing that will regress is // read backs. If we find later we do care about this performance we can come back to figure // out how to do a more narrow workaround. fMustUseCoherentHostVisibleMemory = true; } #endif //////////////////////////////////////////////////////////////////////////// // GrCaps workarounds //////////////////////////////////////////////////////////////////////////// #ifdef SK_BUILD_FOR_ANDROID // MSAA CCPR was slow on Android. http://skbug.com/9676 fDriverDisableMSAAClipAtlas = true; #endif if (kARM_VkVendor == properties.vendorID) { fAvoidWritePixelsFastPath = true; // bugs.skia.org/8064 } // AMD advertises support for MAX_UINT vertex input attributes, but in reality only supports 32. if (kAMD_VkVendor == properties.vendorID) { fMaxVertexAttributes = std::min(fMaxVertexAttributes, 32); } // Adreno devices fail when trying to read the dest using an input attachment and texture // barriers. if (kQualcomm_VkVendor == properties.vendorID) { fTextureBarrierSupport = false; } // On ARM indirect draws are broken on Android 9 and earlier. This was tested on a P30 and // Mate 20x running android 9. if (properties.vendorID == kARM_VkVendor && androidAPIVersion <= 28) { fNativeDrawIndirectSupport = false; } //////////////////////////////////////////////////////////////////////////// // GrShaderCaps workarounds //////////////////////////////////////////////////////////////////////////// if (kImagination_VkVendor == properties.vendorID) { fShaderCaps->fAtan2ImplementedAsAtanYOverX = true; } } void GrVkCaps::initGrCaps(const GrVkInterface* vkInterface, VkPhysicalDevice physDev, const VkPhysicalDeviceProperties& properties, const VkPhysicalDeviceMemoryProperties& memoryProperties, const VkPhysicalDeviceFeatures2& features, const GrVkExtensions& extensions) { // So GPUs, like AMD, are reporting MAX_INT support vertex attributes. In general, there is no // need for us ever to support that amount, and it makes tests which tests all the vertex // attribs timeout looping over that many. For now, we'll cap this at 64 max and can raise it if // we ever find that need. static const uint32_t kMaxVertexAttributes = 64; fMaxVertexAttributes = std::min(properties.limits.maxVertexInputAttributes, kMaxVertexAttributes); // GrCaps::fSampleLocationsSupport refers to the ability to *query* the sample locations (not // program them). For now we just set this to true if the device uses standard locations, and // return the standard locations back when queried. if (properties.limits.standardSampleLocations) { fSampleLocationsSupport = true; } if (extensions.hasExtension(VK_EXT_CONSERVATIVE_RASTERIZATION_EXTENSION_NAME, 1)) { fConservativeRasterSupport = true; } fWireframeSupport = true; // We could actually query and get a max size for each config, however maxImageDimension2D will // give the minimum max size across all configs. So for simplicity we will use that for now. fMaxRenderTargetSize = std::min(properties.limits.maxImageDimension2D, (uint32_t)INT_MAX); fMaxTextureSize = std::min(properties.limits.maxImageDimension2D, (uint32_t)INT_MAX); if (fDriverBugWorkarounds.max_texture_size_limit_4096) { fMaxTextureSize = std::min(fMaxTextureSize, 4096); } // TODO: check if RT's larger than 4k incur a performance cost on ARM. fMaxPreferredRenderTargetSize = fMaxRenderTargetSize; fMaxPushConstantsSize = std::min(properties.limits.maxPushConstantsSize, (uint32_t)INT_MAX); // Assuming since we will always map in the end to upload the data we might as well just map // from the get go. There is no hard data to suggest this is faster or slower. fBufferMapThreshold = 0; fMapBufferFlags = kCanMap_MapFlag | kSubset_MapFlag | kAsyncRead_MapFlag; fOversizedStencilSupport = true; if (extensions.hasExtension(VK_EXT_BLEND_OPERATION_ADVANCED_EXTENSION_NAME, 2) && this->supportsPhysicalDeviceProperties2()) { VkPhysicalDeviceBlendOperationAdvancedPropertiesEXT blendProps; blendProps.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_BLEND_OPERATION_ADVANCED_PROPERTIES_EXT; blendProps.pNext = nullptr; VkPhysicalDeviceProperties2 props; props.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROPERTIES_2; props.pNext = &blendProps; GR_VK_CALL(vkInterface, GetPhysicalDeviceProperties2(physDev, &props)); if (blendProps.advancedBlendAllOperations == VK_TRUE) { fShaderCaps->fAdvBlendEqInteraction = GrShaderCaps::kAutomatic_AdvBlendEqInteraction; auto blendFeatures = get_extension_feature_struct( features, VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_BLEND_OPERATION_ADVANCED_FEATURES_EXT); if (blendFeatures && blendFeatures->advancedBlendCoherentOperations == VK_TRUE) { fBlendEquationSupport = kAdvancedCoherent_BlendEquationSupport; } else { fBlendEquationSupport = kAdvanced_BlendEquationSupport; } } } if (kARM_VkVendor == properties.vendorID) { fShouldCollapseSrcOverToSrcWhenAble = true; } // We're seeing vkCmdClearAttachments take a lot of cpu time when clearing the color attachment. // We really should only be getting in there for partial clears. So instead we will do all // partial clears as draws. if (kQualcomm_VkVendor == properties.vendorID) { fPerformPartialClearsAsDraws = true; } } void GrVkCaps::initShaderCaps(const VkPhysicalDeviceProperties& properties, const VkPhysicalDeviceFeatures2& features) { GrShaderCaps* shaderCaps = fShaderCaps.get(); shaderCaps->fVersionDeclString = "#version 330\n"; // Vulkan is based off ES 3.0 so the following should all be supported shaderCaps->fUsesPrecisionModifiers = true; shaderCaps->fFlatInterpolationSupport = true; // Flat interpolation appears to be slow on Qualcomm GPUs. This was tested in GL and is assumed // to be true with Vulkan as well. shaderCaps->fPreferFlatInterpolation = kQualcomm_VkVendor != properties.vendorID; shaderCaps->fSampleMaskSupport = true; shaderCaps->fShaderDerivativeSupport = true; // ARM GPUs calculate `matrix * vector` in SPIR-V at full precision, even when the inputs are // RelaxedPrecision. Rewriting the multiply as a sum of vector*scalar fixes this. (skia:11769) shaderCaps->fRewriteMatrixVectorMultiply = (kARM_VkVendor == properties.vendorID); shaderCaps->fDualSourceBlendingSupport = features.features.dualSrcBlend; shaderCaps->fIntegerSupport = true; shaderCaps->fNonsquareMatrixSupport = true; shaderCaps->fInverseHyperbolicSupport = true; shaderCaps->fVertexIDSupport = true; shaderCaps->fInfinitySupport = true; shaderCaps->fNonconstantArrayIndexSupport = true; shaderCaps->fBitManipulationSupport = true; // Assume the minimum precisions mandated by the SPIR-V spec. shaderCaps->fFloatIs32Bits = true; shaderCaps->fHalfIs32Bits = false; shaderCaps->fMaxFragmentSamplers = std::min( std::min(properties.limits.maxPerStageDescriptorSampledImages, properties.limits.maxPerStageDescriptorSamplers), (uint32_t)INT_MAX); } bool stencil_format_supported(const GrVkInterface* interface, VkPhysicalDevice physDev, VkFormat format) { VkFormatProperties props; memset(&props, 0, sizeof(VkFormatProperties)); GR_VK_CALL(interface, GetPhysicalDeviceFormatProperties(physDev, format, &props)); return SkToBool(VK_FORMAT_FEATURE_DEPTH_STENCIL_ATTACHMENT_BIT & props.optimalTilingFeatures); } void GrVkCaps::initStencilFormat(const GrVkInterface* interface, VkPhysicalDevice physDev) { if (stencil_format_supported(interface, physDev, VK_FORMAT_S8_UINT)) { fPreferredStencilFormat = VK_FORMAT_S8_UINT; } else if (stencil_format_supported(interface, physDev, VK_FORMAT_D24_UNORM_S8_UINT)) { fPreferredStencilFormat = VK_FORMAT_D24_UNORM_S8_UINT; } else { SkASSERT(stencil_format_supported(interface, physDev, VK_FORMAT_D32_SFLOAT_S8_UINT)); fPreferredStencilFormat = VK_FORMAT_D32_SFLOAT_S8_UINT; } } static bool format_is_srgb(VkFormat format) { SkASSERT(GrVkFormatIsSupported(format)); switch (format) { case VK_FORMAT_R8G8B8A8_SRGB: return true; default: return false; } } // These are all the valid VkFormats that we support in Skia. They are roughly ordered from most // frequently used to least to improve look up times in arrays. static constexpr VkFormat kVkFormats[] = { VK_FORMAT_R8G8B8A8_UNORM, VK_FORMAT_R8_UNORM, VK_FORMAT_B8G8R8A8_UNORM, VK_FORMAT_R5G6B5_UNORM_PACK16, VK_FORMAT_R16G16B16A16_SFLOAT, VK_FORMAT_R16_SFLOAT, VK_FORMAT_R8G8B8_UNORM, VK_FORMAT_R8G8_UNORM, VK_FORMAT_A2B10G10R10_UNORM_PACK32, VK_FORMAT_A2R10G10B10_UNORM_PACK32, VK_FORMAT_B4G4R4A4_UNORM_PACK16, VK_FORMAT_R4G4B4A4_UNORM_PACK16, VK_FORMAT_R8G8B8A8_SRGB, VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK, VK_FORMAT_BC1_RGB_UNORM_BLOCK, VK_FORMAT_BC1_RGBA_UNORM_BLOCK, VK_FORMAT_R16_UNORM, VK_FORMAT_R16G16_UNORM, VK_FORMAT_G8_B8_R8_3PLANE_420_UNORM, VK_FORMAT_G8_B8R8_2PLANE_420_UNORM, VK_FORMAT_R16G16B16A16_UNORM, VK_FORMAT_R16G16_SFLOAT, }; void GrVkCaps::setColorType(GrColorType colorType, std::initializer_list formats) { #ifdef SK_DEBUG for (size_t i = 0; i < kNumVkFormats; ++i) { const auto& formatInfo = fFormatTable[i]; for (int j = 0; j < formatInfo.fColorTypeInfoCount; ++j) { const auto& ctInfo = formatInfo.fColorTypeInfos[j]; if (ctInfo.fColorType == colorType && !SkToBool(ctInfo.fFlags & ColorTypeInfo::kWrappedOnly_Flag)) { bool found = false; for (auto it = formats.begin(); it != formats.end(); ++it) { if (kVkFormats[i] == *it) { found = true; } } SkASSERT(found); } } } #endif int idx = static_cast(colorType); for (auto it = formats.begin(); it != formats.end(); ++it) { const auto& info = this->getFormatInfo(*it); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { if (info.fColorTypeInfos[i].fColorType == colorType) { fColorTypeToFormatTable[idx] = *it; return; } } } } const GrVkCaps::FormatInfo& GrVkCaps::getFormatInfo(VkFormat format) const { GrVkCaps* nonConstThis = const_cast(this); return nonConstThis->getFormatInfo(format); } GrVkCaps::FormatInfo& GrVkCaps::getFormatInfo(VkFormat format) { static_assert(SK_ARRAY_COUNT(kVkFormats) == GrVkCaps::kNumVkFormats, "Size of VkFormats array must match static value in header"); for (size_t i = 0; i < SK_ARRAY_COUNT(kVkFormats); ++i) { if (kVkFormats[i] == format) { return fFormatTable[i]; } } static FormatInfo kInvalidFormat; return kInvalidFormat; } void GrVkCaps::initFormatTable(const GrVkInterface* interface, VkPhysicalDevice physDev, const VkPhysicalDeviceProperties& properties) { static_assert(SK_ARRAY_COUNT(kVkFormats) == GrVkCaps::kNumVkFormats, "Size of VkFormats array must match static value in header"); std::fill_n(fColorTypeToFormatTable, kGrColorTypeCnt, VK_FORMAT_UNDEFINED); // Go through all the formats and init their support surface and data GrColorTypes. // Format: VK_FORMAT_R8G8B8A8_UNORM { constexpr VkFormat format = VK_FORMAT_R8G8B8A8_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 2; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R8G8B8A8_UNORM, Surface: kRGBA_8888 { constexpr GrColorType ct = GrColorType::kRGBA_8888; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } // Format: VK_FORMAT_R8G8B8A8_UNORM, Surface: kRGB_888x { constexpr GrColorType ct = GrColorType::kRGB_888x; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag; ctInfo.fReadSwizzle = GrSwizzle::RGB1(); } } } // Format: VK_FORMAT_R8_UNORM { constexpr VkFormat format = VK_FORMAT_R8_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 2; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R8_UNORM, Surface: kAlpha_8 { constexpr GrColorType ct = GrColorType::kAlpha_8; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; ctInfo.fReadSwizzle = GrSwizzle("000r"); ctInfo.fWriteSwizzle = GrSwizzle("a000"); } // Format: VK_FORMAT_R8_UNORM, Surface: kGray_8 { constexpr GrColorType ct = GrColorType::kGray_8; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag; ctInfo.fReadSwizzle = GrSwizzle("rrr1"); } } } // Format: VK_FORMAT_B8G8R8A8_UNORM { constexpr VkFormat format = VK_FORMAT_B8G8R8A8_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_B8G8R8A8_UNORM, Surface: kBGRA_8888 { constexpr GrColorType ct = GrColorType::kBGRA_8888; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R5G6B5_UNORM_PACK16 { constexpr VkFormat format = VK_FORMAT_R5G6B5_UNORM_PACK16; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R5G6B5_UNORM_PACK16, Surface: kBGR_565 { constexpr GrColorType ct = GrColorType::kBGR_565; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R16G16B16A16_SFLOAT { constexpr VkFormat format = VK_FORMAT_R16G16B16A16_SFLOAT; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 2; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16G16B16A16_SFLOAT, Surface: GrColorType::kRGBA_F16 { constexpr GrColorType ct = GrColorType::kRGBA_F16; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } // Format: VK_FORMAT_R16G16B16A16_SFLOAT, Surface: GrColorType::kRGBA_F16_Clamped { constexpr GrColorType ct = GrColorType::kRGBA_F16_Clamped; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R16_SFLOAT { constexpr VkFormat format = VK_FORMAT_R16_SFLOAT; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16_SFLOAT, Surface: kAlpha_F16 { constexpr GrColorType ct = GrColorType::kAlpha_F16; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; ctInfo.fReadSwizzle = GrSwizzle("000r"); ctInfo.fWriteSwizzle = GrSwizzle("a000"); } } } // Format: VK_FORMAT_R8G8B8_UNORM { constexpr VkFormat format = VK_FORMAT_R8G8B8_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R8G8B8_UNORM, Surface: kRGB_888x { constexpr GrColorType ct = GrColorType::kRGB_888x; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; // The Vulkan format is 3 bpp so we must convert to/from that when transferring. ctInfo.fTransferColorType = GrColorType::kRGB_888; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R8G8_UNORM { constexpr VkFormat format = VK_FORMAT_R8G8_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R8G8_UNORM, Surface: kRG_88 { constexpr GrColorType ct = GrColorType::kRG_88; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_A2B10G10R10_UNORM_PACK32 { constexpr VkFormat format = VK_FORMAT_A2B10G10R10_UNORM_PACK32; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_A2B10G10R10_UNORM_PACK32, Surface: kRGBA_1010102 { constexpr GrColorType ct = GrColorType::kRGBA_1010102; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_A2R10G10B10_UNORM_PACK32 { constexpr VkFormat format = VK_FORMAT_A2R10G10B10_UNORM_PACK32; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_A2R10G10B10_UNORM_PACK32, Surface: kBGRA_1010102 { constexpr GrColorType ct = GrColorType::kBGRA_1010102; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_B4G4R4A4_UNORM_PACK16 { constexpr VkFormat format = VK_FORMAT_B4G4R4A4_UNORM_PACK16; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_B4G4R4A4_UNORM_PACK16, Surface: kABGR_4444 { constexpr GrColorType ct = GrColorType::kABGR_4444; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; ctInfo.fReadSwizzle = GrSwizzle::BGRA(); ctInfo.fWriteSwizzle = GrSwizzle::BGRA(); } } } // Format: VK_FORMAT_R4G4B4A4_UNORM_PACK16 { constexpr VkFormat format = VK_FORMAT_R4G4B4A4_UNORM_PACK16; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R4G4B4A4_UNORM_PACK16, Surface: kABGR_4444 { constexpr GrColorType ct = GrColorType::kABGR_4444; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R8G8B8A8_SRGB { constexpr VkFormat format = VK_FORMAT_R8G8B8A8_SRGB; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R8G8B8A8_SRGB, Surface: kRGBA_8888_SRGB { constexpr GrColorType ct = GrColorType::kRGBA_8888_SRGB; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R16_UNORM { constexpr VkFormat format = VK_FORMAT_R16_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16_UNORM, Surface: kAlpha_16 { constexpr GrColorType ct = GrColorType::kAlpha_16; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; ctInfo.fReadSwizzle = GrSwizzle("000r"); ctInfo.fWriteSwizzle = GrSwizzle("a000"); } } } // Format: VK_FORMAT_R16G16_UNORM { constexpr VkFormat format = VK_FORMAT_R16G16_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16G16_UNORM, Surface: kRG_1616 { constexpr GrColorType ct = GrColorType::kRG_1616; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R16G16B16A16_UNORM { constexpr VkFormat format = VK_FORMAT_R16G16B16A16_UNORM; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16G16B16A16_UNORM, Surface: kRGBA_16161616 { constexpr GrColorType ct = GrColorType::kRGBA_16161616; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_R16G16_SFLOAT { constexpr VkFormat format = VK_FORMAT_R16G16_SFLOAT; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_R16G16_SFLOAT, Surface: kRG_F16 { constexpr GrColorType ct = GrColorType::kRG_F16; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kRenderable_Flag; } } } // Format: VK_FORMAT_G8_B8_R8_3PLANE_420_UNORM { constexpr VkFormat format = VK_FORMAT_G8_B8_R8_3PLANE_420_UNORM; auto& info = this->getFormatInfo(format); if (fSupportsYcbcrConversion) { info.init(interface, physDev, properties, format); } if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_G8_B8_R8_3PLANE_420_UNORM, Surface: kRGB_888x { constexpr GrColorType ct = GrColorType::kRGB_888x; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kWrappedOnly_Flag; } } } // Format: VK_FORMAT_G8_B8R8_2PLANE_420_UNORM { constexpr VkFormat format = VK_FORMAT_G8_B8R8_2PLANE_420_UNORM; auto& info = this->getFormatInfo(format); if (fSupportsYcbcrConversion) { info.init(interface, physDev, properties, format); } if (SkToBool(info.fOptimalFlags & FormatInfo::kTexturable_Flag)) { info.fColorTypeInfoCount = 1; info.fColorTypeInfos = std::make_unique(info.fColorTypeInfoCount); int ctIdx = 0; // Format: VK_FORMAT_G8_B8R8_2PLANE_420_UNORM, Surface: kRGB_888x { constexpr GrColorType ct = GrColorType::kRGB_888x; auto& ctInfo = info.fColorTypeInfos[ctIdx++]; ctInfo.fColorType = ct; ctInfo.fTransferColorType = ct; ctInfo.fFlags = ColorTypeInfo::kUploadData_Flag | ColorTypeInfo::kWrappedOnly_Flag; } } } // Format: VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK { constexpr VkFormat format = VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); // Setting this to texel block size // No supported GrColorTypes. } // Format: VK_FORMAT_BC1_RGB_UNORM_BLOCK { constexpr VkFormat format = VK_FORMAT_BC1_RGB_UNORM_BLOCK; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); // Setting this to texel block size // No supported GrColorTypes. } // Format: VK_FORMAT_BC1_RGBA_UNORM_BLOCK { constexpr VkFormat format = VK_FORMAT_BC1_RGBA_UNORM_BLOCK; auto& info = this->getFormatInfo(format); info.init(interface, physDev, properties, format); // Setting this to texel block size // No supported GrColorTypes. } //////////////////////////////////////////////////////////////////////////// // Map GrColorTypes (used for creating GrSurfaces) to VkFormats. The order in which the formats // are passed into the setColorType function indicates the priority in selecting which format // we use for a given GrcolorType. this->setColorType(GrColorType::kAlpha_8, { VK_FORMAT_R8_UNORM }); this->setColorType(GrColorType::kBGR_565, { VK_FORMAT_R5G6B5_UNORM_PACK16 }); this->setColorType(GrColorType::kABGR_4444, { VK_FORMAT_R4G4B4A4_UNORM_PACK16, VK_FORMAT_B4G4R4A4_UNORM_PACK16 }); this->setColorType(GrColorType::kRGBA_8888, { VK_FORMAT_R8G8B8A8_UNORM }); this->setColorType(GrColorType::kRGBA_8888_SRGB, { VK_FORMAT_R8G8B8A8_SRGB }); this->setColorType(GrColorType::kRGB_888x, { VK_FORMAT_R8G8B8_UNORM, VK_FORMAT_R8G8B8A8_UNORM }); this->setColorType(GrColorType::kRG_88, { VK_FORMAT_R8G8_UNORM }); this->setColorType(GrColorType::kBGRA_8888, { VK_FORMAT_B8G8R8A8_UNORM }); this->setColorType(GrColorType::kRGBA_1010102, { VK_FORMAT_A2B10G10R10_UNORM_PACK32 }); this->setColorType(GrColorType::kBGRA_1010102, { VK_FORMAT_A2R10G10B10_UNORM_PACK32 }); this->setColorType(GrColorType::kGray_8, { VK_FORMAT_R8_UNORM }); this->setColorType(GrColorType::kAlpha_F16, { VK_FORMAT_R16_SFLOAT }); this->setColorType(GrColorType::kRGBA_F16, { VK_FORMAT_R16G16B16A16_SFLOAT }); this->setColorType(GrColorType::kRGBA_F16_Clamped, { VK_FORMAT_R16G16B16A16_SFLOAT }); this->setColorType(GrColorType::kAlpha_16, { VK_FORMAT_R16_UNORM }); this->setColorType(GrColorType::kRG_1616, { VK_FORMAT_R16G16_UNORM }); this->setColorType(GrColorType::kRGBA_16161616, { VK_FORMAT_R16G16B16A16_UNORM }); this->setColorType(GrColorType::kRG_F16, { VK_FORMAT_R16G16_SFLOAT }); } void GrVkCaps::FormatInfo::InitFormatFlags(VkFormatFeatureFlags vkFlags, uint16_t* flags) { if (SkToBool(VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT & vkFlags) && SkToBool(VK_FORMAT_FEATURE_SAMPLED_IMAGE_FILTER_LINEAR_BIT & vkFlags)) { *flags = *flags | kTexturable_Flag; // Ganesh assumes that all renderable surfaces are also texturable if (SkToBool(VK_FORMAT_FEATURE_COLOR_ATTACHMENT_BLEND_BIT & vkFlags)) { *flags = *flags | kRenderable_Flag; } } // TODO: For Vk w/ VK_KHR_maintenance1 extension support, check // VK_FORMAT_FEATURE_TRANSFER_[SRC|DST]_BIT_KHR explicitly to set copy flags // Can do similar check for VK_KHR_sampler_ycbcr_conversion added bits if (SkToBool(VK_FORMAT_FEATURE_BLIT_SRC_BIT & vkFlags)) { *flags = *flags | kBlitSrc_Flag; } if (SkToBool(VK_FORMAT_FEATURE_BLIT_DST_BIT & vkFlags)) { *flags = *flags | kBlitDst_Flag; } } void GrVkCaps::FormatInfo::initSampleCounts(const GrVkInterface* interface, VkPhysicalDevice physDev, const VkPhysicalDeviceProperties& physProps, VkFormat format) { VkImageUsageFlags usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_SAMPLED_BIT | VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT; VkImageFormatProperties properties; GR_VK_CALL(interface, GetPhysicalDeviceImageFormatProperties(physDev, format, VK_IMAGE_TYPE_2D, VK_IMAGE_TILING_OPTIMAL, usage, 0, // createFlags &properties)); VkSampleCountFlags flags = properties.sampleCounts; if (flags & VK_SAMPLE_COUNT_1_BIT) { fColorSampleCounts.push_back(1); } if (kImagination_VkVendor == physProps.vendorID) { // MSAA does not work on imagination return; } if (kIntel_VkVendor == physProps.vendorID) { // MSAA doesn't work well on Intel GPUs chromium:527565, chromium:983926 return; } if (flags & VK_SAMPLE_COUNT_2_BIT) { fColorSampleCounts.push_back(2); } if (flags & VK_SAMPLE_COUNT_4_BIT) { fColorSampleCounts.push_back(4); } if (flags & VK_SAMPLE_COUNT_8_BIT) { fColorSampleCounts.push_back(8); } if (flags & VK_SAMPLE_COUNT_16_BIT) { fColorSampleCounts.push_back(16); } // Standard sample locations are not defined for more than 16 samples, and we don't need more // than 16. Omit 32 and 64. } void GrVkCaps::FormatInfo::init(const GrVkInterface* interface, VkPhysicalDevice physDev, const VkPhysicalDeviceProperties& properties, VkFormat format) { VkFormatProperties props; memset(&props, 0, sizeof(VkFormatProperties)); GR_VK_CALL(interface, GetPhysicalDeviceFormatProperties(physDev, format, &props)); InitFormatFlags(props.linearTilingFeatures, &fLinearFlags); InitFormatFlags(props.optimalTilingFeatures, &fOptimalFlags); if (fOptimalFlags & kRenderable_Flag) { this->initSampleCounts(interface, physDev, properties, format); } } // For many checks in caps, we need to know whether the GrBackendFormat is external or not. If it is // external the VkFormat will be VK_NULL_HANDLE which is not handled by our various format // capability checks. static bool backend_format_is_external(const GrBackendFormat& format) { const GrVkYcbcrConversionInfo* ycbcrInfo = format.getVkYcbcrConversionInfo(); SkASSERT(ycbcrInfo); // All external formats have a valid ycbcrInfo used for sampling and a non zero external format. if (ycbcrInfo->isValid() && ycbcrInfo->fExternalFormat != 0) { #ifdef SK_DEBUG VkFormat vkFormat; SkAssertResult(format.asVkFormat(&vkFormat)); SkASSERT(vkFormat == VK_FORMAT_UNDEFINED); #endif return true; } return false; } bool GrVkCaps::isFormatSRGB(const GrBackendFormat& format) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return false; } if (backend_format_is_external(format)) { return false; } return format_is_srgb(vkFormat); } bool GrVkCaps::isFormatTexturable(const GrBackendFormat& format, GrTextureType) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return false; } if (backend_format_is_external(format)) { // We can always texture from an external format (assuming we have the ycbcr conversion // info which we require to be passed in). return true; } return this->isVkFormatTexturable(vkFormat); } bool GrVkCaps::isVkFormatTexturable(VkFormat format) const { const FormatInfo& info = this->getFormatInfo(format); return SkToBool(FormatInfo::kTexturable_Flag & info.fOptimalFlags); } bool GrVkCaps::isFormatAsColorTypeRenderable(GrColorType ct, const GrBackendFormat& format, int sampleCount) const { if (!this->isFormatRenderable(format, sampleCount)) { return false; } VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return false; } const auto& info = this->getFormatInfo(vkFormat); if (!SkToBool(info.colorTypeFlags(ct) & ColorTypeInfo::kRenderable_Flag)) { return false; } return true; } bool GrVkCaps::isFormatRenderable(const GrBackendFormat& format, int sampleCount) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return false; } return this->isFormatRenderable(vkFormat, sampleCount); } bool GrVkCaps::isFormatRenderable(VkFormat format, int sampleCount) const { return sampleCount <= this->maxRenderTargetSampleCount(format); } int GrVkCaps::getRenderTargetSampleCount(int requestedCount, const GrBackendFormat& format) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return 0; } return this->getRenderTargetSampleCount(requestedCount, vkFormat); } int GrVkCaps::getRenderTargetSampleCount(int requestedCount, VkFormat format) const { requestedCount = std::max(1, requestedCount); const FormatInfo& info = this->getFormatInfo(format); int count = info.fColorSampleCounts.count(); if (!count) { return 0; } if (1 == requestedCount) { SkASSERT(info.fColorSampleCounts.count() && info.fColorSampleCounts[0] == 1); return 1; } for (int i = 0; i < count; ++i) { if (info.fColorSampleCounts[i] >= requestedCount) { return info.fColorSampleCounts[i]; } } return 0; } int GrVkCaps::maxRenderTargetSampleCount(const GrBackendFormat& format) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return 0; } return this->maxRenderTargetSampleCount(vkFormat); } int GrVkCaps::maxRenderTargetSampleCount(VkFormat format) const { const FormatInfo& info = this->getFormatInfo(format); const auto& table = info.fColorSampleCounts; if (!table.count()) { return 0; } return table[table.count() - 1]; } static inline size_t align_to_4(size_t v) { switch (v & 0b11) { // v is already a multiple of 4. case 0: return v; // v is a multiple of 2 but not 4. case 2: return 2 * v; // v is not a multiple of 2. default: return 4 * v; } } GrCaps::SupportedWrite GrVkCaps::supportedWritePixelsColorType(GrColorType surfaceColorType, const GrBackendFormat& surfaceFormat, GrColorType srcColorType) const { VkFormat vkFormat; if (!surfaceFormat.asVkFormat(&vkFormat)) { return {GrColorType::kUnknown, 0}; } // We don't support the ability to upload to external formats or formats that require a ycbcr // sampler. In general these types of formats are only used for sampling in a shader. if (backend_format_is_external(surfaceFormat) || GrVkFormatNeedsYcbcrSampler(vkFormat)) { return {GrColorType::kUnknown, 0}; } // The VkBufferImageCopy bufferOffset field must be both a multiple of 4 and of a single texel. size_t offsetAlignment = align_to_4(GrVkFormatBytesPerBlock(vkFormat)); const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { const auto& ctInfo = info.fColorTypeInfos[i]; if (ctInfo.fColorType == surfaceColorType) { return {ctInfo.fTransferColorType, offsetAlignment}; } } return {GrColorType::kUnknown, 0}; } GrCaps::SurfaceReadPixelsSupport GrVkCaps::surfaceSupportsReadPixels( const GrSurface* surface) const { if (surface->isProtected()) { return SurfaceReadPixelsSupport::kUnsupported; } if (auto tex = static_cast(surface->asTexture())) { auto texImage = tex->textureImage(); if (!texImage) { return SurfaceReadPixelsSupport::kUnsupported; } // We can't directly read from a VkImage that has a ycbcr sampler. if (texImage->ycbcrConversionInfo().isValid()) { return SurfaceReadPixelsSupport::kCopyToTexture2D; } // We can't directly read from a compressed format if (GrVkFormatIsCompressed(texImage->imageFormat())) { return SurfaceReadPixelsSupport::kCopyToTexture2D; } return SurfaceReadPixelsSupport::kSupported; } else if (auto rt = surface->asRenderTarget()) { if (rt->numSamples() > 1) { return SurfaceReadPixelsSupport::kCopyToTexture2D; } return SurfaceReadPixelsSupport::kSupported; } return SurfaceReadPixelsSupport::kUnsupported; } GrColorType GrVkCaps::transferColorType(VkFormat vkFormat, GrColorType surfaceColorType) const { const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { if (info.fColorTypeInfos[i].fColorType == surfaceColorType) { return info.fColorTypeInfos[i].fTransferColorType; } } return GrColorType::kUnknown; } bool GrVkCaps::onSurfaceSupportsWritePixels(const GrSurface* surface) const { if (auto rt = surface->asRenderTarget()) { return rt->numSamples() <= 1 && SkToBool(surface->asTexture()); } // We can't write to a texture that has a ycbcr sampler. if (auto tex = static_cast(surface->asTexture())) { auto texImage = tex->textureImage(); if (!texImage) { return false; } // We can't directly read from a VkImage that has a ycbcr sampler. if (texImage->ycbcrConversionInfo().isValid()) { return false; } } return true; } bool GrVkCaps::onAreColorTypeAndFormatCompatible(GrColorType ct, const GrBackendFormat& format) const { VkFormat vkFormat; if (!format.asVkFormat(&vkFormat)) { return false; } const GrVkYcbcrConversionInfo* ycbcrInfo = format.getVkYcbcrConversionInfo(); SkASSERT(ycbcrInfo); if (ycbcrInfo->isValid() && !GrVkFormatNeedsYcbcrSampler(vkFormat)) { // Format may be undefined for external images, which are required to have YCbCr conversion. if (VK_FORMAT_UNDEFINED == vkFormat && ycbcrInfo->fExternalFormat != 0) { return true; } return false; } const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { if (info.fColorTypeInfos[i].fColorType == ct) { return true; } } return false; } GrBackendFormat GrVkCaps::onGetDefaultBackendFormat(GrColorType ct) const { VkFormat format = this->getFormatFromColorType(ct); if (format == VK_FORMAT_UNDEFINED) { return {}; } return GrBackendFormat::MakeVk(format); } bool GrVkCaps::onSupportsDynamicMSAA(const GrRenderTargetProxy* rtProxy) const { // We must be able to use the rtProxy as an input attachment to load into the discardable msaa // attachment. Also the rtProxy should have a sample count of 1 so that it can be used as a // resolve attachment. return this->supportsDiscardableMSAAForDMSAA() && rtProxy->supportsVkInputAttachment() && rtProxy->numSamples() == 1; } bool GrVkCaps::renderTargetSupportsDiscardableMSAA(const GrVkRenderTarget* rt) const { return rt->resolveAttachment() && rt->resolveAttachment()->supportsInputAttachmentUsage() && ((rt->numSamples() > 1 && this->preferDiscardableMSAAAttachment()) || (rt->numSamples() == 1 && this->supportsDiscardableMSAAForDMSAA())); } bool GrVkCaps::programInfoWillUseDiscardableMSAA(const GrProgramInfo& programInfo) const { return programInfo.targetHasVkResolveAttachmentWithInput() && programInfo.numSamples() > 1 && ((programInfo.targetsNumSamples() > 1 && this->preferDiscardableMSAAAttachment()) || (programInfo.targetsNumSamples() == 1 && this->supportsDiscardableMSAAForDMSAA())); } GrBackendFormat GrVkCaps::getBackendFormatFromCompressionType( SkImage::CompressionType compressionType) const { switch (compressionType) { case SkImage::CompressionType::kNone: return {}; case SkImage::CompressionType::kETC2_RGB8_UNORM: if (this->isVkFormatTexturable(VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK)) { return GrBackendFormat::MakeVk(VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK); } return {}; case SkImage::CompressionType::kBC1_RGB8_UNORM: if (this->isVkFormatTexturable(VK_FORMAT_BC1_RGB_UNORM_BLOCK)) { return GrBackendFormat::MakeVk(VK_FORMAT_BC1_RGB_UNORM_BLOCK); } return {}; case SkImage::CompressionType::kBC1_RGBA8_UNORM: if (this->isVkFormatTexturable(VK_FORMAT_BC1_RGBA_UNORM_BLOCK)) { return GrBackendFormat::MakeVk(VK_FORMAT_BC1_RGBA_UNORM_BLOCK); } return {}; } SkUNREACHABLE; } GrSwizzle GrVkCaps::onGetReadSwizzle(const GrBackendFormat& format, GrColorType colorType) const { VkFormat vkFormat; SkAssertResult(format.asVkFormat(&vkFormat)); const auto* ycbcrInfo = format.getVkYcbcrConversionInfo(); SkASSERT(ycbcrInfo); if (ycbcrInfo->isValid() && ycbcrInfo->fExternalFormat != 0) { // We allow these to work with any color type and never swizzle. See // onAreColorTypeAndFormatCompatible. return GrSwizzle{"rgba"}; } const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { const auto& ctInfo = info.fColorTypeInfos[i]; if (ctInfo.fColorType == colorType) { return ctInfo.fReadSwizzle; } } SkDEBUGFAILF("Illegal color type (%d) and format (%d) combination.", (int)colorType, (int)vkFormat); return {}; } GrSwizzle GrVkCaps::getWriteSwizzle(const GrBackendFormat& format, GrColorType colorType) const { VkFormat vkFormat; SkAssertResult(format.asVkFormat(&vkFormat)); const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { const auto& ctInfo = info.fColorTypeInfos[i]; if (ctInfo.fColorType == colorType) { return ctInfo.fWriteSwizzle; } } SkDEBUGFAILF("Illegal color type (%d) and format (%d) combination.", (int)colorType, (int)vkFormat); return {}; } GrDstSampleFlags GrVkCaps::onGetDstSampleFlagsForProxy(const GrRenderTargetProxy* rt) const { bool isMSAAWithResolve = rt->numSamples() > 1 && rt->asTextureProxy(); // TODO: Currently if we have an msaa rt with a resolve, the supportsVkInputAttachment call // references whether the resolve is supported as an input attachment. We need to add a check to // allow checking the color attachment (msaa or not) supports input attachment specifically. if (!isMSAAWithResolve && rt->supportsVkInputAttachment()) { return GrDstSampleFlags::kRequiresTextureBarrier | GrDstSampleFlags::kAsInputAttachment; } return GrDstSampleFlags::kNone; } uint64_t GrVkCaps::computeFormatKey(const GrBackendFormat& format) const { VkFormat vkFormat; SkAssertResult(format.asVkFormat(&vkFormat)); #ifdef SK_DEBUG // We should never be trying to compute a key for an external format const GrVkYcbcrConversionInfo* ycbcrInfo = format.getVkYcbcrConversionInfo(); SkASSERT(ycbcrInfo); SkASSERT(!ycbcrInfo->isValid() || ycbcrInfo->fExternalFormat == 0); #endif // A VkFormat has a size of 64 bits. return (uint64_t)vkFormat; } GrCaps::SupportedRead GrVkCaps::onSupportedReadPixelsColorType( GrColorType srcColorType, const GrBackendFormat& srcBackendFormat, GrColorType dstColorType) const { VkFormat vkFormat; if (!srcBackendFormat.asVkFormat(&vkFormat)) { return {GrColorType::kUnknown, 0}; } if (GrVkFormatNeedsYcbcrSampler(vkFormat)) { return {GrColorType::kUnknown, 0}; } SkImage::CompressionType compression = GrBackendFormatToCompressionType(srcBackendFormat); if (compression != SkImage::CompressionType::kNone) { return { SkCompressionTypeIsOpaque(compression) ? GrColorType::kRGB_888x : GrColorType::kRGBA_8888, 0 }; } // The VkBufferImageCopy bufferOffset field must be both a multiple of 4 and of a single texel. size_t offsetAlignment = align_to_4(GrVkFormatBytesPerBlock(vkFormat)); const auto& info = this->getFormatInfo(vkFormat); for (int i = 0; i < info.fColorTypeInfoCount; ++i) { const auto& ctInfo = info.fColorTypeInfos[i]; if (ctInfo.fColorType == srcColorType) { return {ctInfo.fTransferColorType, offsetAlignment}; } } return {GrColorType::kUnknown, 0}; } int GrVkCaps::getFragmentUniformBinding() const { return GrVkUniformHandler::kUniformBinding; } int GrVkCaps::getFragmentUniformSet() const { return GrVkUniformHandler::kUniformBufferDescSet; } void GrVkCaps::addExtraSamplerKey(GrProcessorKeyBuilder* b, GrSamplerState samplerState, const GrBackendFormat& format) const { const GrVkYcbcrConversionInfo* ycbcrInfo = format.getVkYcbcrConversionInfo(); if (!ycbcrInfo) { return; } GrVkSampler::Key key = GrVkSampler::GenerateKey(samplerState, *ycbcrInfo); constexpr size_t numInts = (sizeof(key) + 3) / 4; uint32_t tmp[numInts]; memcpy(tmp, &key, sizeof(key)); for (size_t i = 0; i < numInts; ++i) { b->add32(tmp[i]); } } /** * For Vulkan we want to cache the entire VkPipeline for reuse of draws. The Desc here holds all * the information needed to differentiate one pipeline from another. * * The GrProgramDesc contains all the information need to create the actual shaders for the * pipeline. * * For Vulkan we need to add to the GrProgramDesc to include the rest of the state on the * pipline. This includes stencil settings, blending information, render pass format, draw face * information, and primitive type. Note that some state is set dynamically on the pipeline for * each draw and thus is not included in this descriptor. This includes the viewport, scissor, * and blend constant. */ GrProgramDesc GrVkCaps::makeDesc(GrRenderTarget* rt, const GrProgramInfo& programInfo, ProgramDescOverrideFlags overrideFlags) const { GrProgramDesc desc; GrProgramDesc::Build(&desc, programInfo, *this); GrProcessorKeyBuilder b(desc.key()); // This will become part of the sheared off key used to persistently cache // the SPIRV code. It needs to be added right after the base key so that, // when the base-key is sheared off, the shearing code can include it in the // reduced key (c.f. the +4s in the SkData::MakeWithCopy calls in // GrVkPipelineStateBuilder.cpp). b.add32(GrVkGpu::kShader_PersistentCacheKeyType); GrVkRenderPass::SelfDependencyFlags selfDepFlags = GrVkRenderPass::SelfDependencyFlags::kNone; if (programInfo.renderPassBarriers() & GrXferBarrierFlags::kBlend) { selfDepFlags |= GrVkRenderPass::SelfDependencyFlags::kForNonCoherentAdvBlend; } if (programInfo.renderPassBarriers() & GrXferBarrierFlags::kTexture) { selfDepFlags |= GrVkRenderPass::SelfDependencyFlags::kForInputAttachment; } bool needsResolve = this->programInfoWillUseDiscardableMSAA(programInfo); bool forceLoadFromResolve = overrideFlags & GrCaps::ProgramDescOverrideFlags::kVulkanHasResolveLoadSubpass; SkASSERT(!forceLoadFromResolve || needsResolve); GrVkRenderPass::LoadFromResolve loadFromResolve = GrVkRenderPass::LoadFromResolve::kNo; if (needsResolve && (programInfo.colorLoadOp() == GrLoadOp::kLoad || forceLoadFromResolve)) { loadFromResolve = GrVkRenderPass::LoadFromResolve::kLoad; } if (rt) { GrVkRenderTarget* vkRT = (GrVkRenderTarget*) rt; SkASSERT(!needsResolve || (vkRT->resolveAttachment() && vkRT->resolveAttachment()->supportsInputAttachmentUsage())); bool needsStencil = programInfo.needsStencil() || programInfo.isStencilEnabled(); // TODO: support failure in getSimpleRenderPass auto rp = vkRT->getSimpleRenderPass(needsResolve, needsStencil, selfDepFlags, loadFromResolve); SkASSERT(rp); rp->genKey(&b); #ifdef SK_DEBUG if (!rp->isExternal()) { // This is to ensure ReconstructAttachmentsDescriptor keeps matching // getSimpleRenderPass' result GrVkRenderPass::AttachmentsDescriptor attachmentsDescriptor; GrVkRenderPass::AttachmentFlags attachmentFlags; GrVkRenderTarget::ReconstructAttachmentsDescriptor(*this, programInfo, &attachmentsDescriptor, &attachmentFlags); SkASSERT(rp->isCompatible(attachmentsDescriptor, attachmentFlags, selfDepFlags, loadFromResolve)); } #endif } else { GrVkRenderPass::AttachmentsDescriptor attachmentsDescriptor; GrVkRenderPass::AttachmentFlags attachmentFlags; GrVkRenderTarget::ReconstructAttachmentsDescriptor(*this, programInfo, &attachmentsDescriptor, &attachmentFlags); // kExternal_AttachmentFlag is only set for wrapped secondary command buffers - which // will always go through the above 'rt' path (i.e., we can always pass 0 as the final // parameter to GenKey). GrVkRenderPass::GenKey(&b, attachmentFlags, attachmentsDescriptor, selfDepFlags, loadFromResolve, 0); } GrStencilSettings stencil = programInfo.nonGLStencilSettings(); stencil.genKey(&b, true); programInfo.pipeline().genKey(&b, *this); b.add32(programInfo.numSamples()); // Vulkan requires the full primitive type as part of its key b.add32(programInfo.primitiveTypeKey()); b.flush(); return desc; } GrInternalSurfaceFlags GrVkCaps::getExtraSurfaceFlagsForDeferredRT() const { // We always create vulkan RT with the input attachment flag; return GrInternalSurfaceFlags::kVkRTSupportsInputAttachment; } VkShaderStageFlags GrVkCaps::getPushConstantStageFlags() const { VkShaderStageFlags stageFlags = VK_SHADER_STAGE_VERTEX_BIT | VK_SHADER_STAGE_FRAGMENT_BIT; return stageFlags; } #if GR_TEST_UTILS std::vector GrVkCaps::getTestingCombinations() const { std::vector combos = { { GrColorType::kAlpha_8, GrBackendFormat::MakeVk(VK_FORMAT_R8_UNORM) }, { GrColorType::kBGR_565, GrBackendFormat::MakeVk(VK_FORMAT_R5G6B5_UNORM_PACK16) }, { GrColorType::kABGR_4444, GrBackendFormat::MakeVk(VK_FORMAT_R4G4B4A4_UNORM_PACK16)}, { GrColorType::kABGR_4444, GrBackendFormat::MakeVk(VK_FORMAT_B4G4R4A4_UNORM_PACK16)}, { GrColorType::kRGBA_8888, GrBackendFormat::MakeVk(VK_FORMAT_R8G8B8A8_UNORM) }, { GrColorType::kRGBA_8888_SRGB, GrBackendFormat::MakeVk(VK_FORMAT_R8G8B8A8_SRGB) }, { GrColorType::kRGB_888x, GrBackendFormat::MakeVk(VK_FORMAT_R8G8B8A8_UNORM) }, { GrColorType::kRGB_888x, GrBackendFormat::MakeVk(VK_FORMAT_R8G8B8_UNORM) }, { GrColorType::kRG_88, GrBackendFormat::MakeVk(VK_FORMAT_R8G8_UNORM) }, { GrColorType::kBGRA_8888, GrBackendFormat::MakeVk(VK_FORMAT_B8G8R8A8_UNORM) }, { GrColorType::kRGBA_1010102, GrBackendFormat::MakeVk(VK_FORMAT_A2B10G10R10_UNORM_PACK32)}, { GrColorType::kBGRA_1010102, GrBackendFormat::MakeVk(VK_FORMAT_A2R10G10B10_UNORM_PACK32)}, { GrColorType::kGray_8, GrBackendFormat::MakeVk(VK_FORMAT_R8_UNORM) }, { GrColorType::kAlpha_F16, GrBackendFormat::MakeVk(VK_FORMAT_R16_SFLOAT) }, { GrColorType::kRGBA_F16, GrBackendFormat::MakeVk(VK_FORMAT_R16G16B16A16_SFLOAT) }, { GrColorType::kRGBA_F16_Clamped, GrBackendFormat::MakeVk(VK_FORMAT_R16G16B16A16_SFLOAT) }, { GrColorType::kAlpha_16, GrBackendFormat::MakeVk(VK_FORMAT_R16_UNORM) }, { GrColorType::kRG_1616, GrBackendFormat::MakeVk(VK_FORMAT_R16G16_UNORM) }, { GrColorType::kRGBA_16161616, GrBackendFormat::MakeVk(VK_FORMAT_R16G16B16A16_UNORM) }, { GrColorType::kRG_F16, GrBackendFormat::MakeVk(VK_FORMAT_R16G16_SFLOAT) }, // These two compressed formats both have an effective colorType of kRGB_888x { GrColorType::kRGB_888x, GrBackendFormat::MakeVk(VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK)}, { GrColorType::kRGB_888x, GrBackendFormat::MakeVk(VK_FORMAT_BC1_RGB_UNORM_BLOCK) }, { GrColorType::kRGBA_8888, GrBackendFormat::MakeVk(VK_FORMAT_BC1_RGBA_UNORM_BLOCK) }, }; return combos; } #endif