GraphiteDawn: cache single texture bind group
Bug: b/260368758
Change-Id: Iafc8c340b37feb8cb184f9da9878f9190b3213bc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/780956
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Quyen Le <lehoangquyen@chromium.org>
diff --git a/src/gpu/graphite/dawn/DawnCommandBuffer.cpp b/src/gpu/graphite/dawn/DawnCommandBuffer.cpp
index f6cb2ba..4d6c69e 100644
--- a/src/gpu/graphite/dawn/DawnCommandBuffer.cpp
+++ b/src/gpu/graphite/dawn/DawnCommandBuffer.cpp
@@ -284,6 +284,8 @@
return false;
}
+ this->trackResource(msaaLoadTexture);
+
// Creating intermediate render pass (copy from resolve texture -> MSAA load texture)
RenderPassDesc intermediateRenderPassDesc = {};
intermediateRenderPassDesc.fColorAttachment.fLoadOp = LoadOp::kDiscard;
@@ -529,37 +531,49 @@
SkASSERT(fActiveRenderPassEncoder);
SkASSERT(fActiveGraphicsPipeline);
- // TODO: optimize for single texture.
- std::vector<wgpu::BindGroupEntry> entries(2 * command.fNumTexSamplers);
+ wgpu::BindGroup bindGroup;
+ if (command.fNumTexSamplers == 1) {
+ // Optimize for single texture.
+ SkASSERT(fActiveGraphicsPipeline->numTexturesAndSamplers() == 2);
- for (int i = 0; i < command.fNumTexSamplers; ++i) {
const auto* texture =
- static_cast<const DawnTexture*>(drawPass.getTexture(command.fTextureIndices[i]));
+ static_cast<const DawnTexture*>(drawPass.getTexture(command.fTextureIndices[0]));
const auto* sampler =
- static_cast<const DawnSampler*>(drawPass.getSampler(command.fSamplerIndices[i]));
- auto& wgpuTextureView = texture->sampleTextureView();
- auto& wgpuSampler = sampler->dawnSampler();
+ static_cast<const DawnSampler*>(drawPass.getSampler(command.fSamplerIndices[0]));
- // Assuming shader generator assigns binding slot to sampler then texture,
- // then the next sampler and texture, and so on, we need to use
- // 2 * i as base binding index of the sampler and texture.
- // TODO: https://b.corp.google.com/issues/259457090:
- // Better configurable way of assigning samplers and textures' bindings.
- entries[2 * i].binding = 2 * i;
- entries[2 * i].sampler = wgpuSampler;
+ bindGroup = fResourceProvider->findOrCreateSingleTextureSamplerBindGroup(sampler, texture);
+ } else {
+ std::vector<wgpu::BindGroupEntry> entries(2 * command.fNumTexSamplers);
- entries[2 * i + 1].binding = 2 * i + 1;
- entries[2 * i + 1].textureView = wgpuTextureView;
+ for (int i = 0; i < command.fNumTexSamplers; ++i) {
+ const auto* texture = static_cast<const DawnTexture*>(
+ drawPass.getTexture(command.fTextureIndices[i]));
+ const auto* sampler = static_cast<const DawnSampler*>(
+ drawPass.getSampler(command.fSamplerIndices[i]));
+ auto& wgpuTextureView = texture->sampleTextureView();
+ auto& wgpuSampler = sampler->dawnSampler();
+
+ // Assuming shader generator assigns binding slot to sampler then texture,
+ // then the next sampler and texture, and so on, we need to use
+ // 2 * i as base binding index of the sampler and texture.
+ // TODO: https://b.corp.google.com/issues/259457090:
+ // Better configurable way of assigning samplers and textures' bindings.
+ entries[2 * i].binding = 2 * i;
+ entries[2 * i].sampler = wgpuSampler;
+
+ entries[2 * i + 1].binding = 2 * i + 1;
+ entries[2 * i + 1].textureView = wgpuTextureView;
+ }
+
+ wgpu::BindGroupDescriptor desc;
+ const auto& groupLayouts = fActiveGraphicsPipeline->dawnGroupLayouts();
+ desc.layout = groupLayouts[DawnGraphicsPipeline::kTextureBindGroupIndex];
+ desc.entryCount = entries.size();
+ desc.entries = entries.data();
+
+ bindGroup = fSharedContext->device().CreateBindGroup(&desc);
}
- wgpu::BindGroupDescriptor desc;
- const auto& groupLayouts = fActiveGraphicsPipeline->dawnGroupLayouts();
- desc.layout = groupLayouts[DawnGraphicsPipeline::kTextureBindGroupIndex];
- desc.entryCount = entries.size();
- desc.entries = entries.data();
-
- auto bindGroup = fSharedContext->device().CreateBindGroup(&desc);
-
fActiveRenderPassEncoder.SetBindGroup(DawnGraphicsPipeline::kTextureBindGroupIndex, bindGroup);
}
diff --git a/src/gpu/graphite/dawn/DawnGraphicsPipeline.cpp b/src/gpu/graphite/dawn/DawnGraphicsPipeline.cpp
index 9bf00b4..d2f219f 100644
--- a/src/gpu/graphite/dawn/DawnGraphicsPipeline.cpp
+++ b/src/gpu/graphite/dawn/DawnGraphicsPipeline.cpp
@@ -386,27 +386,33 @@
bool hasFragmentSamplers = hasFragmentSkSL && numTexturesAndSamplers > 0;
if (hasFragmentSamplers) {
- std::vector<wgpu::BindGroupLayoutEntry> entries(numTexturesAndSamplers);
- for (int i = 0; i < numTexturesAndSamplers;) {
- entries[i].binding = static_cast<uint32_t>(i);
- entries[i].visibility = wgpu::ShaderStage::Fragment;
- entries[i].sampler.type = wgpu::SamplerBindingType::Filtering;
- ++i;
- entries[i].binding = i;
- entries[i].visibility = wgpu::ShaderStage::Fragment;
- entries[i].texture.sampleType = wgpu::TextureSampleType::Float;
- entries[i].texture.viewDimension = wgpu::TextureViewDimension::e2D;
- entries[i].texture.multisampled = false;
- ++i;
- }
+ if (numTexturesAndSamplers == 2) {
+ // Common case: single texture + sampler.
+ groupLayouts[1] =
+ resourceProvider->getOrCreateSingleTextureSamplerBindGroupLayout();
+ } else {
+ std::vector<wgpu::BindGroupLayoutEntry> entries(numTexturesAndSamplers);
+ for (int i = 0; i < numTexturesAndSamplers;) {
+ entries[i].binding = static_cast<uint32_t>(i);
+ entries[i].visibility = wgpu::ShaderStage::Fragment;
+ entries[i].sampler.type = wgpu::SamplerBindingType::Filtering;
+ ++i;
+ entries[i].binding = i;
+ entries[i].visibility = wgpu::ShaderStage::Fragment;
+ entries[i].texture.sampleType = wgpu::TextureSampleType::Float;
+ entries[i].texture.viewDimension = wgpu::TextureViewDimension::e2D;
+ entries[i].texture.multisampled = false;
+ ++i;
+ }
- wgpu::BindGroupLayoutDescriptor groupLayoutDesc;
+ wgpu::BindGroupLayoutDescriptor groupLayoutDesc;
#if defined(SK_DEBUG)
- groupLayoutDesc.label = step->name();
+ groupLayoutDesc.label = step->name();
#endif
- groupLayoutDesc.entryCount = entries.size();
- groupLayoutDesc.entries = entries.data();
- groupLayouts[1] = device.CreateBindGroupLayout(&groupLayoutDesc);
+ groupLayoutDesc.entryCount = entries.size();
+ groupLayoutDesc.entries = entries.data();
+ groupLayouts[1] = device.CreateBindGroupLayout(&groupLayoutDesc);
+ }
if (!groupLayouts[1]) {
return {};
}
@@ -554,7 +560,8 @@
step->primitiveType(),
depthStencilSettings.fStencilReferenceValue,
/*hasStepUniforms=*/!step->uniforms().empty(),
- /*hasPaintUniforms=*/fsSkSLInfo.fNumPaintUniforms > 0));
+ /*hasPaintUniforms=*/fsSkSLInfo.fNumPaintUniforms > 0,
+ numTexturesAndSamplers));
}
DawnGraphicsPipeline::DawnGraphicsPipeline(const skgpu::graphite::SharedContext* sharedContext,
@@ -564,14 +571,16 @@
PrimitiveType primitiveType,
uint32_t refValue,
bool hasStepUniforms,
- bool hasPaintUniforms)
+ bool hasPaintUniforms,
+ int numFragmentTexturesAndSamplers)
: GraphicsPipeline(sharedContext, pipelineInfo)
, fAsyncPipelineCreation(std::move(asyncCreationInfo))
, fGroupLayouts(std::move(groupLayouts))
, fPrimitiveType(primitiveType)
, fStencilReferenceValue(refValue)
, fHasStepUniforms(hasStepUniforms)
- , fHasPaintUniforms(hasPaintUniforms) {}
+ , fHasPaintUniforms(hasPaintUniforms)
+ , fNumFragmentTexturesAndSamplers(numFragmentTexturesAndSamplers) {}
void DawnGraphicsPipeline::freeGpuData() {
fAsyncPipelineCreation = nullptr;
diff --git a/src/gpu/graphite/dawn/DawnGraphicsPipeline.h b/src/gpu/graphite/dawn/DawnGraphicsPipeline.h
index b7fb93c..1bdc1d4 100644
--- a/src/gpu/graphite/dawn/DawnGraphicsPipeline.h
+++ b/src/gpu/graphite/dawn/DawnGraphicsPipeline.h
@@ -65,6 +65,7 @@
PrimitiveType primitiveType() const { return fPrimitiveType; }
bool hasStepUniforms() const { return fHasStepUniforms; }
bool hasPaintUniforms() const { return fHasPaintUniforms; }
+ int numTexturesAndSamplers() const { return fNumFragmentTexturesAndSamplers; }
const wgpu::RenderPipeline& dawnRenderPipeline() const;
using BindGroupLayouts = std::array<wgpu::BindGroupLayout, kBindGroupCount>;
@@ -80,7 +81,8 @@
PrimitiveType primitiveType,
uint32_t refValue,
bool hasStepUniforms,
- bool hasPaintUniforms);
+ bool hasPaintUniforms,
+ int numFragmentTexturesAndSamplers);
void freeGpuData() override;
@@ -90,6 +92,7 @@
const uint32_t fStencilReferenceValue;
const bool fHasStepUniforms;
const bool fHasPaintUniforms;
+ const int fNumFragmentTexturesAndSamplers;
};
} // namespace skgpu::graphite
diff --git a/src/gpu/graphite/dawn/DawnResourceProvider.cpp b/src/gpu/graphite/dawn/DawnResourceProvider.cpp
index d410654..1cc1d75 100644
--- a/src/gpu/graphite/dawn/DawnResourceProvider.cpp
+++ b/src/gpu/graphite/dawn/DawnResourceProvider.cpp
@@ -24,7 +24,8 @@
namespace {
constexpr int kBufferBindingSizeAlignment = 16;
-constexpr int kMaxNumberOfCachedBindGroups = 32;
+constexpr int kMaxNumberOfCachedBufferBindGroups = 32;
+constexpr int kMaxNumberOfCachedTextureBindGroups = 256;
wgpu::ShaderModule create_shader_module(const wgpu::Device& device, const char* source) {
wgpu::ShaderModuleWGSLDescriptor wgslDesc;
@@ -124,6 +125,25 @@
return uniqueKey;
}
+
+UniqueKey make_texture_bind_group_key(const DawnSampler* sampler, const DawnTexture* texture) {
+ static const UniqueKey::Domain kTextureBindGroupDomain = UniqueKey::GenerateDomain();
+
+ UniqueKey uniqueKey;
+ {
+ UniqueKey::Builder builder(&uniqueKey,
+ kTextureBindGroupDomain,
+ 2,
+ "GraphicsPipelineSingleTextureSamplerBindGroup");
+
+ builder[0] = sampler->uniqueID().asUInt();
+ builder[1] = texture->uniqueID().asUInt();
+
+ builder.finish();
+ }
+
+ return uniqueKey;
+}
} // namespace
DawnResourceProvider::DawnResourceProvider(SharedContext* sharedContext,
@@ -131,7 +151,8 @@
uint32_t recorderID,
size_t resourceBudget)
: ResourceProvider(sharedContext, singleOwner, recorderID, resourceBudget)
- , fUniformBufferBindGroupCache(kMaxNumberOfCachedBindGroups) {}
+ , fUniformBufferBindGroupCache(kMaxNumberOfCachedBufferBindGroups)
+ , fSingleTextureSamplerBindGroups(kMaxNumberOfCachedTextureBindGroups) {}
DawnResourceProvider::~DawnResourceProvider() = default;
@@ -337,6 +358,37 @@
return fUniformBuffersBindGroupLayout;
}
+const wgpu::BindGroupLayout&
+DawnResourceProvider::getOrCreateSingleTextureSamplerBindGroupLayout() {
+ if (fSingleTextureSamplerBindGroupLayout) {
+ return fSingleTextureSamplerBindGroupLayout;
+ }
+
+ std::array<wgpu::BindGroupLayoutEntry, 2> entries;
+
+ entries[0].binding = 0;
+ entries[0].visibility = wgpu::ShaderStage::Fragment;
+ entries[0].sampler.type = wgpu::SamplerBindingType::Filtering;
+
+ entries[1].binding = 1;
+ entries[1].visibility = wgpu::ShaderStage::Fragment;
+ entries[1].texture.sampleType = wgpu::TextureSampleType::Float;
+ entries[1].texture.viewDimension = wgpu::TextureViewDimension::e2D;
+ entries[1].texture.multisampled = false;
+
+ wgpu::BindGroupLayoutDescriptor groupLayoutDesc;
+#if defined(SK_DEBUG)
+ groupLayoutDesc.label = "Single texture + sampler bind group layout";
+#endif
+
+ groupLayoutDesc.entryCount = entries.size();
+ groupLayoutDesc.entries = entries.data();
+ fSingleTextureSamplerBindGroupLayout =
+ this->dawnSharedContext()->device().CreateBindGroupLayout(&groupLayoutDesc);
+
+ return fSingleTextureSamplerBindGroupLayout;
+}
+
const wgpu::Buffer& DawnResourceProvider::getOrCreateNullBuffer() {
if (!fNullBuffer) {
wgpu::BufferDescriptor desc;
@@ -399,4 +451,31 @@
return *fUniformBufferBindGroupCache.insert(key, bindGroup);
}
+const wgpu::BindGroup& DawnResourceProvider::findOrCreateSingleTextureSamplerBindGroup(
+ const DawnSampler* sampler, const DawnTexture* texture) {
+ auto key = make_texture_bind_group_key(sampler, texture);
+ auto* existingBindGroup = fSingleTextureSamplerBindGroups.find(key);
+ if (existingBindGroup) {
+ // cache hit.
+ return *existingBindGroup;
+ }
+
+ std::array<wgpu::BindGroupEntry, 2> entries;
+
+ entries[0].binding = 0;
+ entries[0].sampler = sampler->dawnSampler();
+ entries[1].binding = 1;
+ entries[1].textureView = texture->sampleTextureView();
+
+ wgpu::BindGroupDescriptor desc;
+ desc.layout = getOrCreateSingleTextureSamplerBindGroupLayout();
+ desc.entryCount = entries.size();
+ desc.entries = entries.data();
+
+ const auto& device = this->dawnSharedContext()->device();
+ auto bindGroup = device.CreateBindGroup(&desc);
+
+ return *fSingleTextureSamplerBindGroups.insert(key, bindGroup);
+}
+
} // namespace skgpu::graphite
diff --git a/src/gpu/graphite/dawn/DawnResourceProvider.h b/src/gpu/graphite/dawn/DawnResourceProvider.h
index b2be5ce..3a86a5e 100644
--- a/src/gpu/graphite/dawn/DawnResourceProvider.h
+++ b/src/gpu/graphite/dawn/DawnResourceProvider.h
@@ -15,6 +15,7 @@
namespace skgpu::graphite {
class DawnGraphicsPipeline;
+class DawnSampler;
class DawnSharedContext;
class DawnTexture;
class DawnBuffer;
@@ -37,6 +38,7 @@
sk_sp<DawnBuffer> findOrCreateDawnBuffer(size_t size, BufferType type, AccessPattern);
const wgpu::BindGroupLayout& getOrCreateUniformBuffersBindGroupLayout();
+ const wgpu::BindGroupLayout& getOrCreateSingleTextureSamplerBindGroupLayout();
// Find the cached bind group or create a new one based on the bound buffers and their
// binding sizes (boundBuffersAndSizes) for these uniforms (in order):
@@ -46,6 +48,10 @@
const wgpu::BindGroup& findOrCreateUniformBuffersBindGroup(
const std::array<std::pair<const DawnBuffer*, uint32_t>, 3>& boundBuffersAndSizes);
+ // Find or create a bind group containing the given sampler & texture.
+ const wgpu::BindGroup& findOrCreateSingleTextureSamplerBindGroup(const DawnSampler* sampler,
+ const DawnTexture* texture);
+
private:
sk_sp<GraphicsPipeline> createGraphicsPipeline(const RuntimeEffectDictionary*,
const GraphicsPipelineDesc&,
@@ -69,6 +75,7 @@
skia_private::THashMap<uint64_t, wgpu::RenderPipeline> fBlitWithDrawPipelines;
wgpu::BindGroupLayout fUniformBuffersBindGroupLayout;
+ wgpu::BindGroupLayout fSingleTextureSamplerBindGroupLayout;
wgpu::Buffer fNullBuffer;
@@ -76,8 +83,10 @@
uint32_t operator()(const skgpu::UniqueKey& key) const { return key.hash(); }
};
- using BufferBindGroupCache = SkLRUCache<UniqueKey, wgpu::BindGroup, UniqueKeyHash>;
- BufferBindGroupCache fUniformBufferBindGroupCache;
+ using BindGroupCache = SkLRUCache<UniqueKey, wgpu::BindGroup, UniqueKeyHash>;
+
+ BindGroupCache fUniformBufferBindGroupCache;
+ BindGroupCache fSingleTextureSamplerBindGroups;
};
} // namespace skgpu::graphite
diff --git a/src/gpu/graphite/dawn/DawnTexture.cpp b/src/gpu/graphite/dawn/DawnTexture.cpp
index ed87408..12f0fc3 100644
--- a/src/gpu/graphite/dawn/DawnTexture.cpp
+++ b/src/gpu/graphite/dawn/DawnTexture.cpp
@@ -184,6 +184,12 @@
}
void DawnTexture::freeGpuData() {
+ if (this->ownership() != Ownership::kWrapped && fTexture) {
+ // Destroy the texture even if it is still referenced by other BindGroup or views.
+ // Graphite should already guarantee that all command buffers using this texture (indirectly
+ // via BindGroup or views) are already completed.
+ fTexture.Destroy();
+ }
fTexture = nullptr;
fSampleTextureView = nullptr;
fRenderTextureView = nullptr;
