Skip to content

Added reduce_all layoutinfer & a reduce_all case #543

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
sunshinemyson merged 1 commit into from
Jan 9, 2023
Merged

Added reduce_all layoutinfer & a reduce_all case #543

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/tim/transform/layout_inference.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -181,6 +181,7 @@ void LayoutInferContext::UpdateGraphOutputMap(const std::shared_ptr<vx::Tensor>&
REGIST_LAYOUT_INFERENCE(VSI_NN_REDUCE_PROD, ReduceProd); \
REGIST_LAYOUT_INFERENCE(VSI_NN_REDUCE_ANY, ReduceAny); \
REGIST_LAYOUT_INFERENCE(VSI_NN_REDUCE_SUM, ReduceSum); \
REGIST_LAYOUT_INFERENCE(VSI_NN_REDUCE_ALL, ReduceAll); \
default: \
VSILOGW("Op %d: Default layout inference pass for reduce.", reduce_type);\
assert(false); \
Expand Down
1 change: 1 addition & 0 deletions src/tim/transform/ops/reduce_layout_inference.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,7 @@ using ReduceAnyLayoutInfer = ReduceLayoutInfer<tim::vx::ops::ReduceAny>;
using ReduceProdLayoutInfer = ReduceLayoutInfer<tim::vx::ops::ReduceProd>;
using ReduceMeanLayoutInfer = ReduceLayoutInfer<tim::vx::ops::ReduceMean>;
using ReduceSumLayoutInfer = ReduceLayoutInfer<tim::vx::ops::ReduceSum>;
using ReduceAllLayoutInfer = ReduceLayoutInfer<tim::vx::ops::ReduceAll>;
} // namespace transform
} // namespace tim

Expand Down
101 changes: 101 additions & 0 deletions src/tim/vx/ops/reduce_sum_test.cc → src/tim/vx/ops/reduce_test.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,4 +128,105 @@ TEST(Reduce_sum, KeepDims) {
std::vector<float> output(golden.size());
EXPECT_TRUE(output_tensor->CopyDataFromTensor(output.data()));
EXPECT_TRUE(ArraysMatch(golden, output, 1e-5f));
}

TEST(Reduce_all, KeepDims) {
auto ctx = tim::vx::Context::Create();

auto graph = ctx->CreateGraph();
tim::vx::ShapeType input_shape({2, 3, 2});
tim::vx::ShapeType output_shape({1, 3, 1});

tim::vx::TensorSpec input_spec(tim::vx::DataType::BOOL8, input_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::BOOL8, output_shape,
tim::vx::TensorAttribute::OUTPUT);

std::vector<int32_t> axis = {2, 0};
auto input_tensor = graph->CreateTensor(input_spec);
auto output_tensor = graph->CreateTensor(output_spec);
auto reduce_all = graph->CreateOperation<tim::vx::ops::ReduceAll>(axis, true);
(*reduce_all).BindInputs({input_tensor}).BindOutputs({output_tensor});

bool in_data[] = {true, true, true, true, true, true,
true, true, false, true, true, true};
bool golden[] = {true, false, true};
bool* p_in = in_data;
input_tensor->CopyDataToTensor(p_in, 12 * sizeof(bool));

EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());

bool output[3 * sizeof(bool)];
bool* p_output = output;
EXPECT_TRUE(output_tensor->CopyDataFromTensor(p_output));
for (int i = 0; i < 3; ++i) {
EXPECT_TRUE(golden[i] == output[i]);
}
}

TEST(Reduce_all, NotKeepDims) {
auto ctx = tim::vx::Context::Create();

auto graph = ctx->CreateGraph();
tim::vx::ShapeType input_shape({2, 3, 2});
tim::vx::ShapeType output_shape({2});

tim::vx::TensorSpec input_spec(tim::vx::DataType::BOOL8, input_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::BOOL8, output_shape,
tim::vx::TensorAttribute::OUTPUT);

std::vector<int32_t> axis = {1, 2, 2, 2};
auto input_tensor = graph->CreateTensor(input_spec);
auto output_tensor = graph->CreateTensor(output_spec);
auto reduce_all =
graph->CreateOperation<tim::vx::ops::ReduceAll>(axis, false);
(*reduce_all).BindInputs({input_tensor}).BindOutputs({output_tensor});

bool in_data[] = {true, true, true, true, true, false,
true, true, true, true, true, true};
bool golden[] = {true, false};
bool* p_in = in_data;
input_tensor->CopyDataToTensor(p_in, 12 * sizeof(bool));

EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());

bool output[2 * sizeof(bool)];
bool* p_output = output;
EXPECT_TRUE(output_tensor->CopyDataFromTensor(p_output));
for (int i = 0; i < 2; ++i) {
EXPECT_TRUE(golden[i] == output[i]);
}
}

TEST(Reduce_max, NotKeepDims) {
auto ctx = tim::vx::Context::Create();

auto graph = ctx->CreateGraph();
tim::vx::ShapeType input_shape({2, 3});
tim::vx::ShapeType output_shape({3});

tim::vx::TensorSpec input_spec(tim::vx::DataType::FLOAT32, input_shape,
tim::vx::TensorAttribute::INPUT);
tim::vx::TensorSpec output_spec(tim::vx::DataType::FLOAT32, output_shape,
tim::vx::TensorAttribute::OUTPUT);
auto input_tensor = graph->CreateTensor(input_spec);
auto output_tensor = graph->CreateTensor(output_spec);
std::vector<int32_t> axis = {0};
auto reduce_sum = graph->CreateOperation<tim::vx::ops::ReduceMax>(axis, false);
(*reduce_sum).BindInputs({input_tensor}).BindOutputs({output_tensor});

std::vector<float> in_data = {-1.0f, -2.0f, 3.0f, 4.0f, 5.0f, -6.0f};
std::vector<float> golden = {-1.0f, 4.0f, 5.0f};

EXPECT_TRUE(input_tensor->CopyDataToTensor(in_data.data(), in_data.size()));

EXPECT_TRUE(graph->Compile());
EXPECT_TRUE(graph->Run());

std::vector<float> output(golden.size());
EXPECT_TRUE(output_tensor->CopyDataFromTensor(output.data()));
EXPECT_EQ(golden, output);
}