DEV/flutter_flutter
1
0
Fork 0
mirror of https://github.com/flutter/flutter.git synced 2026-02-20 02:29:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

Avoid truncation to zero when resizing threadgroups (flutter/engine#40502)

Browse Source 
[Impeller] Avoid truncation to zero when resizing threadgroups
This commit is contained in:
Dan Field 2023-03-21 15:11:04 -07:00 committed by GitHub
parent c3f475f8c5
commit 885013ccdd
2 changed files with 83 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
engine/src/flutter/impeller/renderer/backend/metal/compute_pass_mtl.mm
View File

@ -252,8 +252,8 @@ bool ComputePassMTL::EncodeCommands(const std::shared_ptr<Allocator>& allocator,
while (width * height >
static_cast<int64_t>(
pass_bindings.GetPipeline().maxTotalThreadsPerThreadgroup)) {
width /= 2;
height /= 2;
width = std::max(1LL, width / 2);
height = std::max(1LL, height / 2);
}
auto size = MTLSizeMake(width, height, 1);
[encoder dispatchThreadgroups:size threadsPerThreadgroup:size];

81
engine/src/flutter/impeller/renderer/compute_unittests.cc
View File

@ -204,5 +204,86 @@ TEST_P(ComputeTest, MultiStageInputAndOutput) {
latch.Wait();
}
TEST_P(ComputeTest, CanCorrectlyDownScaleLargeGridSize) {
using CS = SampleComputeShader;
auto context = GetContext();
ASSERT_TRUE(context);
ASSERT_TRUE(context->GetDeviceCapabilities().SupportsCompute());
using SamplePipelineBuilder = ComputePipelineBuilder<CS>;
auto pipeline_desc =
SamplePipelineBuilder::MakeDefaultPipelineDescriptor(*context);
ASSERT_TRUE(pipeline_desc.has_value());
auto compute_pipeline =
context->GetPipelineLibrary()->GetPipeline(pipeline_desc).Get();
ASSERT_TRUE(compute_pipeline);
auto cmd_buffer = context->CreateCommandBuffer();
auto pass = cmd_buffer->CreateComputePass();
ASSERT_TRUE(pass && pass->IsValid());
static constexpr size_t kCount = 5;
// Intentionally making the grid size obscenely large. No GPU will tolerate
// this.
pass->SetGridSize(ISize(std::numeric_limits<int64_t>::max(), 1));
pass->SetThreadGroupSize(ISize(std::numeric_limits<int64_t>::max(), 1));
ComputeCommand cmd;
cmd.label = "Compute";
cmd.pipeline = compute_pipeline;
CS::Info info{.count = kCount};
CS::Input0<kCount> input_0;
CS::Input1<kCount> input_1;
for (size_t i = 0; i < kCount; i++) {
input_0.elements[i] = Vector4(2.0 + i, 3.0 + i, 4.0 + i, 5.0 * i);
input_1.elements[i] = Vector4(6.0, 7.0, 8.0, 9.0);
}
input_0.fixed_array[1] = IPoint32(2, 2);
input_1.fixed_array[0] = UintPoint32(3, 3);
input_0.some_int = 5;
input_1.some_struct = CS::SomeStruct{.vf = Point(3, 4), .i = 42};
auto output_buffer = CreateHostVisibleDeviceBuffer<CS::Output<kCount>>(
context, "Output Buffer");
CS::BindInfo(cmd, pass->GetTransientsBuffer().EmplaceUniform(info));
CS::BindInput0(cmd,
pass->GetTransientsBuffer().EmplaceStorageBuffer(input_0));
CS::BindInput1(cmd,
pass->GetTransientsBuffer().EmplaceStorageBuffer(input_1));
CS::BindOutput(cmd, output_buffer->AsBufferView());
ASSERT_TRUE(pass->AddCommand(std::move(cmd)));
ASSERT_TRUE(pass->EncodeCommands());
fml::AutoResetWaitableEvent latch;
ASSERT_TRUE(
cmd_buffer->SubmitCommands([&latch, output_buffer, &input_0,
&input_1](CommandBuffer::Status status) {
EXPECT_EQ(status, CommandBuffer::Status::kCompleted);
auto view = output_buffer->AsBufferView();
EXPECT_EQ(view.range.length, sizeof(CS::Output<kCount>));
CS::Output<kCount>* output =
reinterpret_cast<CS::Output<kCount>*>(view.contents);
EXPECT_TRUE(output);
for (size_t i = 0; i < kCount; i++) {
Vector4 vector = output->elements[i];
Vector4 computed = input_0.elements[i] * input_1.elements[i];
EXPECT_EQ(vector, Vector4(computed.x + 2 + input_1.some_struct.i,
computed.y + 3 + input_1.some_struct.vf.x,
computed.z + 5 + input_1.some_struct.vf.y,
computed.w));
}
latch.Signal();
}));
latch.Wait();
}
} // namespace testing
} // namespace impeller