saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-29 20:14:01 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

NCHWc optimizer fixes for quantized models (#5203)

Browse source 
This updates the NCHWc transformer to not interfere with quantized convolution models, based on observations from internal models. The tensor type for MaxPool must be float. The input to GlobalAveragePool/GlobalMaxPool must be in NCHWc format.
This commit is contained in:
Tracy Sharpe 2020-09-17 09:52:21 -07:00 committed by Tianlei Wu
parent 84ffdbc467
commit f726af34e0
2 changed files with 83 additions and 48 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
onnxruntime/core/optimizer/nchwc_transformer.cc
View file

@ -465,7 +465,11 @@ void NchwcTransformerImpl::TransformPool(Node& node) {
const size_t nchwc_block_size = MlasNchwcGetBlockSize();
auto* input_shape = input_defs[0]->Shape();
const auto* input_type = input_defs[0]->TypeAsProto();
if ((input_type == nullptr) || (input_type->tensor_type().elem_type() != TensorProto_DataType_FLOAT)) {
return;
}
const auto* input_shape = input_defs[0]->Shape();
if ((input_shape == nullptr) || (input_shape->dim_size() != 4)) {
return;
}
@ -927,9 +931,7 @@ void NchwcTransformerImpl::Transform(Node& node) {
graph_utils::IsSupportedOptypeVersionAndDomain(node, "FusedConv", {1}, kMSDomain)) {
TransformConv(node);
} else if (graph_utils::IsSupportedOptypeVersionAndDomain(node, "MaxPool", {1, 8, 10, 11, 12}) ||
graph_utils::IsSupportedOptypeVersionAndDomain(node, "AveragePool", {1, 7, 10, 11}) ||
graph_utils::IsSupportedOptypeVersionAndDomain(node, "GlobalMaxPool", {1}) ||
graph_utils::IsSupportedOptypeVersionAndDomain(node, "GlobalAveragePool", {1})) {
graph_utils::IsSupportedOptypeVersionAndDomain(node, "AveragePool", {1, 7, 10, 11})) {
TransformPool(node);
} else if (node.GetInputEdgesCount() == 0 && node.InputDefs().size() != 0) {
// The following transforms only run when the input edge count has already
@ -955,6 +957,10 @@ void NchwcTransformerImpl::Transform(Node& node) {
} else if (graph_utils::IsSupportedOptypeVersionAndDomain(node, "Upsample", {9, 13}) ||
graph_utils::IsSupportedOptypeVersionAndDomain(node, "Resize", {10, 11, 13})) {
TransformResize(node);
} else if (graph_utils::IsSupportedOptypeVersionAndDomain(node, "GlobalMaxPool", {1}) ||
graph_utils::IsSupportedOptypeVersionAndDomain(node, "GlobalAveragePool", {1})) {
// Convert these pooling types only if the input is already in NCHWc format.
TransformPool(node);
}
}

117
onnxruntime/test/optimizer/nchwc_optimizer_test.cc
View file

@ -44,6 +44,7 @@ struct NchwcTestHelper {
NchwcTestHelper(Graph& graph) : graph_(graph), fill_value_(0), per_sample_tolerance_(0.0) {
}
template<typename T>
NodeArg* MakeInput(const std::vector<int64_t>& shape, const ONNX_NAMESPACE::TypeProto& type_proto) {
int64_t num_elements = 1;
for (auto& dim : shape) {
@ -51,23 +52,24 @@ struct NchwcTestHelper {
}
OrtValue input_value;
CreateMLValue<float>(TestCPUExecutionProvider()->GetAllocator(0, OrtMemTypeDefault), shape,
FillRandomData(static_cast<size_t>(num_elements)), &input_value);
CreateMLValue<T>(TestCPUExecutionProvider()->GetAllocator(0, OrtMemTypeDefault), shape,
FillRandomData<T>(static_cast<size_t>(num_elements)), &input_value);
std::string name = graph_.GenerateNodeArgName("input");
feeds_.insert(std::make_pair(name, input_value));
return &graph_.GetOrCreateNodeArg(name, &type_proto);
}
template<typename T>
NodeArg* MakeInput(const std::vector<int64_t>& shape) {
ONNX_NAMESPACE::TypeProto type_proto;
type_proto.mutable_tensor_type()->set_elem_type(ONNX_NAMESPACE::TensorProto_DataType_FLOAT);
type_proto.mutable_tensor_type()->set_elem_type(utils::ToTensorProtoElementType<T>());
for (auto& dim : shape) {
type_proto.mutable_tensor_type()->mutable_shape()->add_dim()->set_dim_value(dim);
}
return MakeInput(shape, type_proto);
return MakeInput<T>(shape, type_proto);
}
NodeArg* MakeOutput() {
@ -101,7 +103,7 @@ struct NchwcTestHelper {
NodeArg* MakeInitializer(const std::vector<int64_t>& shape) {
int64_t num_elements = std::accumulate(shape.begin(), shape.end(), int64_t(1), std::multiplies<int64_t>{});
return MakeInitializer(shape, FillRandomData(static_cast<size_t>(num_elements)));
return MakeInitializer(shape, FillRandomData<float>(static_cast<size_t>(num_elements)));
}
NodeArg* Make1DInitializer(const std::vector<float>& data) {
@ -161,14 +163,15 @@ struct NchwcTestHelper {
return AddTransposeNode(input_arg, output_arg, {1, 0, 2, 3});
}
std::vector<float> FillRandomData(size_t count) {
template<typename T>
std::vector<T> FillRandomData(size_t count) {
constexpr int min_fill_value = -23;
constexpr int max_fill_value = 23;
std::vector<float> random_data;
std::vector<T> random_data;
random_data.resize(count);
for (size_t n = 0; n < count; n++) {
random_data[n] = static_cast<float>(fill_value_);
random_data[n] = static_cast<T>(fill_value_);
fill_value_++;
if (fill_value_ == max_fill_value) {
fill_value_ = min_fill_value;
@ -186,7 +189,7 @@ struct NchwcTestHelper {
void NchwcOptimizerTester(const std::function<void(NchwcTestHelper& helper)>& build_test_case,
const std::function<void(NchwcInferenceSession& session)>& check_nchwc_graph,
int opset_version = 11) {
int opset_version = 12) {
// Ignore the test if NCHWc is not supported by the platform.
if (MlasNchwcGetBlockSize() <= 1) {
return;
@ -251,7 +254,7 @@ void NchwcOptimizerTester(const std::function<void(NchwcTestHelper& helper)>& bu
TEST(NchwcOptimizerTests, ConvNchw) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({16, 3, 112, 112});
auto* input_arg = helper.MakeInput<float>({16, 3, 112, 112});
auto* output_arg = helper.MakeOutput();
auto* conv_output_arg = output_arg;
@ -291,7 +294,7 @@ TEST(NchwcOptimizerTests, ConvNchw) {
TEST(NchwcOptimizerTests, ConvNchwc) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({16, 64, 28, 28});
auto* input_arg = helper.MakeInput<float>({16, 64, 28, 28});
auto* output_arg = helper.MakeOutput();
auto* conv_output_arg = output_arg;
@ -329,7 +332,7 @@ TEST(NchwcOptimizerTests, ConvNchwc) {
TEST(NchwcOptimizerTests, ConvNchwcGrouped) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({16, 48, 28, 28});
auto* input_arg = helper.MakeInput<float>({16, 48, 28, 28});
auto* output_arg = helper.MakeOutput();
auto* conv_output_arg = output_arg;
@ -364,7 +367,7 @@ TEST(NchwcOptimizerTests, ConvNchwcGrouped) {
TEST(NchwcOptimizerTests, ConvDepthwise) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({16, 96, 28, 28});
auto* input_arg = helper.MakeInput<float>({16, 96, 28, 28});
auto* output_arg = helper.MakeOutput();
auto* conv_output_arg = output_arg;
@ -399,7 +402,7 @@ TEST(NchwcOptimizerTests, ConvDepthwise) {
TEST(NchwcOptimizerTests, ConvPointwise) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({16, 64, 28, 42});
auto* input_arg = helper.MakeInput<float>({16, 64, 28, 42});
auto* output_arg = helper.MakeOutput();
auto* conv_output_arg = output_arg;
@ -432,7 +435,7 @@ TEST(NchwcOptimizerTests, ConvPointwise) {
TEST(NchwcOptimizerTests, ConvMaxPool) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 34, 34});
auto* input_arg = helper.MakeInput<float>({1, 48, 34, 34});
auto* conv_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -456,7 +459,7 @@ TEST(NchwcOptimizerTests, ConvMaxPool) {
TEST(NchwcOptimizerTests, ConvMaxPoolDilations) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 66, 77});
auto* input_arg = helper.MakeInput<float>({1, 48, 66, 77});
auto* conv_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -481,7 +484,7 @@ TEST(NchwcOptimizerTests, ConvMaxPoolDilations) {
TEST(NchwcOptimizerTests, ConvAveragePool) {
auto test_case = [&](bool count_include_pad) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 34, 34});
auto* input_arg = helper.MakeInput<float>({1, 48, 34, 34});
auto* conv_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -513,7 +516,7 @@ TEST(NchwcOptimizerTests, ConvAveragePool) {
TEST(NchwcOptimizerTests, ConvGlobalPool) {
auto test_case = [&](const std::string& op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 96, 54, 54});
auto* input_arg = helper.MakeInput<float>({1, 96, 54, 54});
auto* conv_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -543,7 +546,7 @@ TEST(NchwcOptimizerTests, ConvGlobalPool) {
TEST(NchwcOptimizerTests, ConvAddFusion) {
auto test_case = [&](const std::string& op_type, int opset_version, bool do_relu) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 32, 28, 28});
auto* input_arg = helper.MakeInput<float>({1, 32, 28, 28});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* conv2_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -575,7 +578,7 @@ TEST(NchwcOptimizerTests, ConvAddFusion) {
// Verify that Add or Sum can be fused into a preceding NCHWc Conv node,
// with an optional Relu node following.
std::vector<std::string> op_types{"Add", "Sum"};
static const int opset_versions[] = {7, 10, 11};
static const int opset_versions[] = {7, 10, 11, 12};
for (auto& op_type : op_types) {
for (auto opset_version : opset_versions) {
test_case(op_type, opset_version, false);
@ -586,7 +589,7 @@ TEST(NchwcOptimizerTests, ConvAddFusion) {
TEST(NchwcOptimizerTests, ConvNoBiasAddFusion) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 32, 28, 28});
auto* input_arg = helper.MakeInput<float>({1, 32, 28, 28});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* conv2_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -612,7 +615,7 @@ TEST(NchwcOptimizerTests, ConvNoBiasAddFusion) {
TEST(NchwcOptimizerTests, FusedConvAddFusion) {
auto test_case = [&](bool do_relu1, bool do_relu2, int add_count) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 32, 28, 28});
auto* input_arg = helper.MakeInput<float>({1, 32, 28, 28});
auto* add1_input_arg = helper.MakeIntermediate();
auto* add2_input_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -659,7 +662,7 @@ TEST(NchwcOptimizerTests, FusedConvAddFusion) {
TEST(NchwcOptimizerTests, ConvBinary) {
auto test_case = [&](const std::string& op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 32, 23, 23});
auto* input_arg = helper.MakeInput<float>({1, 32, 23, 23});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* conv2_output_arg = helper.MakeIntermediate();
auto* relu1_output_arg = helper.MakeIntermediate();
@ -697,7 +700,7 @@ TEST(NchwcOptimizerTests, ConvBinary) {
TEST(NchwcOptimizerTests, ConvConcat) {
auto test_case = [&](int axis, int channel_count, int reorder_output_count) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 17, 34});
auto* input_arg = helper.MakeInput<float>({1, 48, 17, 34});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* conv2_output_arg = helper.MakeIntermediate();
auto* conv3_output_arg = helper.MakeIntermediate();
@ -733,7 +736,7 @@ TEST(NchwcOptimizerTests, ConvConcat) {
TEST(NchwcOptimizerTests, ConvReuseWeightsOIHWBiBo) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 64, 7, 7});
auto* input_arg = helper.MakeInput<float>({1, 64, 7, 7});
auto* output1_arg = helper.MakeOutput();
auto* output2_arg = helper.MakeOutput();
auto* output3_arg = helper.MakeOutput();
@ -774,10 +777,10 @@ TEST(NchwcOptimizerTests, ConvReuseWeightsOIHWBiBo) {
TEST(NchwcOptimizerTests, ConvReuseWeightsOIHWBo) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input1_arg = helper.MakeInput({1, 64, 7, 7});
auto* input2_arg = helper.MakeInput({1, 64, 7, 7});
auto* input3_arg = helper.MakeInput({1, 1, 7, 7});
auto* input4_arg = helper.MakeInput({1, 1, 7, 7});
auto* input1_arg = helper.MakeInput<float>({1, 64, 7, 7});
auto* input2_arg = helper.MakeInput<float>({1, 64, 7, 7});
auto* input3_arg = helper.MakeInput<float>({1, 1, 7, 7});
auto* input4_arg = helper.MakeInput<float>({1, 1, 7, 7});
auto* output1_arg = helper.MakeOutput();
auto* output2_arg = helper.MakeOutput();
auto* output3_arg = helper.MakeOutput();
@ -831,7 +834,7 @@ TEST(NchwcOptimizerTests, ShapeInferencing) {
type_proto.mutable_tensor_type()->mutable_shape()->add_dim()->set_dim_param("input_height");
type_proto.mutable_tensor_type()->mutable_shape()->add_dim()->set_dim_param("input_width");
auto* input_arg = helper.MakeInput({1, 3, 50, 100}, type_proto);
auto* input_arg = helper.MakeInput<float>({1, 3, 50, 100}, type_proto);
auto* output_arg = helper.MakeOutput();
// With these padding and kernel arguments, the shape along each spatial
@ -887,7 +890,7 @@ TEST(NchwcOptimizerTests, ShapeInferencing2) {
type_proto.mutable_tensor_type()->mutable_shape()->add_dim()->set_dim_param("input_height");
type_proto.mutable_tensor_type()->mutable_shape()->add_dim()->set_dim_param("input_width");
auto* input_arg = helper.MakeInput({1, 1, 49, 98}, type_proto);
auto* input_arg = helper.MakeInput<float>({1, 1, 49, 98}, type_proto);
auto* output_arg = helper.MakeOutput();
auto* conv1_output_arg = helper.MakeIntermediate();
@ -925,7 +928,7 @@ TEST(NchwcOptimizerTests, ShapeInferencing2) {
TEST(NchwcOptimizerTests, MixedOutputUsage) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({6, 5, 11, 11});
auto* input_arg = helper.MakeInput<float>({6, 5, 11, 11});
auto* output_arg = helper.MakeOutput();
auto* conv1_output_arg = helper.MakeIntermediate();
@ -957,24 +960,24 @@ TEST(NchwcOptimizerTests, MixedOutputUsage) {
TEST(NchwcOptimizerTests, TensorAlignment) {
auto build_test_case = [&](NchwcTestHelper& helper) {
// Input channel count must currently be a multiple of the NCHWc block size.
auto* input1_arg = helper.MakeInput({1, 60, 28, 42});
auto* input1_arg = helper.MakeInput<float>({1, 60, 28, 42});
auto* output1_arg = helper.MakeOutput();
helper.AddConvNode(input1_arg, output1_arg, {128, 60, 1, 1});
// Grouped input channel count must be a multiple of the NCHWc block size.
auto* input2_arg = helper.MakeInput({1, 48, 28, 42});
auto* input2_arg = helper.MakeInput<float>({1, 48, 28, 42});
auto* output2_arg = helper.MakeOutput();
auto& conv2_node = helper.AddConvNode(input2_arg, output2_arg, {128, 12, 3, 3});
conv2_node.AddAttribute("group", static_cast<int64_t>(4));
// Grouped output channel count must be a multiple of the NCHWc block size.
auto* input3_arg = helper.MakeInput({1, 64, 28, 42});
auto* input3_arg = helper.MakeInput<float>({1, 64, 28, 42});
auto* output3_arg = helper.MakeOutput();
auto& conv3_node = helper.AddConvNode(input3_arg, output3_arg, {48, 16, 3, 3});
conv3_node.AddAttribute("group", static_cast<int64_t>(4));
// Channel count must currently be a multiple of the NCHWc block size.
auto* input4_arg = helper.MakeInput({1, 60, 12, 12});
auto* input4_arg = helper.MakeInput<float>({1, 60, 12, 12});
auto* output4_arg = helper.MakeOutput();
auto& pool_node = helper.AddNode("MaxPool", {input4_arg}, {output4_arg});
pool_node.AddAttribute("kernel_shape", std::vector<int64_t>{2, 2});
@ -996,7 +999,7 @@ TEST(NchwcOptimizerTests, TensorAlignment) {
TEST(NchwcOptimizerTests, IntermediatesAsGraphOutputs) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 34, 34});
auto* input_arg = helper.MakeInput<float>({1, 48, 34, 34});
auto* conv_output_arg = helper.MakeOutput();
auto* output_arg = helper.MakeOutput();
@ -1028,7 +1031,7 @@ TEST(NchwcOptimizerTests, IntermediatesAsGraphOutputs) {
TEST(NchwcOptimizerTests, BatchNormalization) {
auto test_case = [&](bool training_outputs) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 1, 23, 21});
auto* input_arg = helper.MakeInput<float>({1, 1, 23, 21});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* conv2_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -1104,7 +1107,7 @@ TEST(NchwcOptimizerTests, BatchNormalization) {
TEST(NchwcOptimizerTests, ConvReorderOutputNhwc) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 64, 28, 32});
auto* input_arg = helper.MakeInput<float>({1, 64, 28, 32});
auto* conv_output_arg = helper.MakeIntermediate();
auto* nhwc_output_arg = helper.MakeOutput();
@ -1126,7 +1129,7 @@ TEST(NchwcOptimizerTests, ConvReorderOutputNhwc) {
TEST(NchwcOptimizerTests, ConvReorderOutputBoth) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({5, 64, 33, 37});
auto* input_arg = helper.MakeInput<float>({5, 64, 33, 37});
auto* conv_output_arg = helper.MakeIntermediate();
auto* nchw_output_arg = helper.MakeOutput();
auto* nhwc_output_arg = helper.MakeOutput();
@ -1151,7 +1154,7 @@ TEST(NchwcOptimizerTests, ConvReorderOutputBoth) {
TEST(NchwcOptimizerTests, ConvReorderOutputCnhw) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 64, 28, 32});
auto* input_arg = helper.MakeInput<float>({1, 64, 28, 32});
auto* conv_output_arg = helper.MakeIntermediate();
auto* nhwc_output_arg = helper.MakeOutput();
@ -1174,7 +1177,7 @@ TEST(NchwcOptimizerTests, ConvReorderOutputCnhw) {
TEST(NchwcOptimizerTests, Upsample) {
auto test_case = [&](int opset_version, float scale_h, float scale_w) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({3, 16, 27, 15});
auto* input_arg = helper.MakeInput<float>({3, 16, 27, 15});
auto* conv_output_arg = helper.MakeIntermediate();
auto* output_arg = helper.MakeOutput();
@ -1211,7 +1214,7 @@ TEST(NchwcOptimizerTests, Upsample) {
// Verify that upsample nodes can be converted to the NCHWc format for
// various versions of the operator.
static const int opset_versions[] = {9, 10, 11};
static const int opset_versions[] = {9, 10, 11, 12};
for (auto opset_version : opset_versions) {
test_case(opset_version, 1.f, 1.f);
test_case(opset_version, 2.f, 2.f);
@ -1222,7 +1225,7 @@ TEST(NchwcOptimizerTests, Upsample) {
TEST(NchwcOptimizerTests, Activation) {
auto test_case = [&](const std::string& activation_op_type) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto* input_arg = helper.MakeInput({1, 48, 11, 15});
auto* input_arg = helper.MakeInput<float>({1, 48, 11, 15});
auto* conv1_output_arg = helper.MakeIntermediate();
auto* activation_output_arg = helper.MakeIntermediate();
auto* mul_output_arg = helper.MakeIntermediate();
@ -1254,6 +1257,32 @@ TEST(NchwcOptimizerTests, Activation) {
}
}
TEST(NchwcOptimizerTests, MaxPoolTypeCheck) {
auto build_test_case = [&](NchwcTestHelper& helper) {
auto add_pool_node = [&](NchwcTestHelper& helper, NodeArg* input_arg) {
auto* output_arg = helper.MakeOutput();
auto& pool_node = helper.AddNode("MaxPool", {input_arg}, {output_arg});
pool_node.AddAttribute("pads", std::vector<int64_t>{0, 0, 0, 0});
pool_node.AddAttribute("kernel_shape", std::vector<int64_t>{3, 3});
};
const std::vector<int64_t> input_shape{1, 32, 13, 13};
add_pool_node(helper, helper.MakeInput<float>(input_shape));
add_pool_node(helper, helper.MakeInput<uint8_t>(input_shape));
};
auto check_nchwc_graph = [&](NchwcInferenceSession& session) {
auto op_to_count = session.CountOpsInGraph();
EXPECT_EQ(op_to_count["MaxPool"], 1);
EXPECT_EQ(op_to_count["nchwc.MaxPool"], 1);
EXPECT_EQ(op_to_count["nchwc.ReorderInput"], 1);
EXPECT_EQ(op_to_count["nchwc.ReorderOutput"], 1);
};
// Verify that the optimizer checks the type of the MaxPool node.
NchwcOptimizerTester(build_test_case, check_nchwc_graph, 12);
}
#endif
} // namespace test