Skip to content

P
Paddle-Lite

PaddlePaddle / Paddle-Lite

通知 331
Star 4
Fork 1
P
Paddle-Lite
提交 59b80b23 编写于 作者: N nhzlx
浏览文件
操作

Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle-Lite into develop

上级 20ba19fe fd9ceec8
显示空白变更内容
内联 并排
Showing with 139 addition and 186 deletion
lite/api/CMakeLists.txt
浏览文件 @ 59b80b23
......@@ -119,27 +119,27 @@ if(WITH_TESTING)
add_dependencies(test_cxx_api extern_lite_download_lite_naive_model_tar_gz)
if(NOT LITE_WITH_LIGHT_WEIGHT_FRAMEWORK)
lite_cc_test(test_googlenet SRCS test_googlenet_lite.cc
DEPS cxx_api mir_passes lite_api_test_helper
DEPS mir_passes lite_api_test_helper paddle_api_full paddle_api_light gflags utils
${ops} ${host_kernels} ${x86_kernels}
ARGS --model_dir=${LITE_MODEL_DIR}/googlenet)
add_dependencies(test_googlenet extern_lite_download_GoogleNet_inference_tar_gz)
lite_cc_test(test_mobilenetv1_lite_x86 SRCS test_mobilenetv1_lite_x86.cc
DEPS cxx_api mir_passes lite_api_test_helper
DEPS mir_passes lite_api_test_helper paddle_api_full paddle_api_light gflags utils
${ops} ${host_kernels} ${x86_kernels}
ARGS --model_dir=${LITE_MODEL_DIR}/mobilenet_v1)
add_dependencies(test_mobilenetv1_lite_x86 extern_lite_download_mobilenet_v1_tar_gz)
lite_cc_test(test_mobilenetv2_lite_x86 SRCS test_mobilenetv2_lite_x86.cc
DEPS cxx_api mir_passes lite_api_test_helper
DEPS mir_passes lite_api_test_helper paddle_api_full paddle_api_light gflags utils
${ops} ${host_kernels} ${x86_kernels}
ARGS --model_dir=${LITE_MODEL_DIR}/mobilenet_v2_relu)
add_dependencies(test_mobilenetv2_lite_x86 extern_lite_download_mobilenet_v2_relu_tar_gz)
lite_cc_test(test_inceptionv4_lite_x86 SRCS test_inceptionv4_lite_x86.cc
DEPS cxx_api mir_passes lite_api_test_helper
DEPS mir_passes lite_api_test_helper paddle_api_full paddle_api_light gflags utils
${ops} ${host_kernels} ${x86_kernels}
ARGS --model_dir=${LITE_MODEL_DIR}/inception_v4_simple)
add_dependencies(test_inceptionv4_lite_x86 extern_lite_download_inception_v4_simple_tar_gz)
lite_cc_test(test_resnet50_lite_x86 SRCS test_resnet50_lite_x86.cc
DEPS cxx_api mir_passes lite_api_test_helper
DEPS mir_passes lite_api_test_helper paddle_api_full paddle_api_light gflags utils
${ops} ${host_kernels} ${x86_kernels}
ARGS --model_dir=${LITE_MODEL_DIR}/resnet50)
add_dependencies(test_resnet50_lite_x86 extern_lite_download_resnet50_tar_gz)
......
lite/api/test_googlenet_lite.cc
浏览文件 @ 59b80b23
......@@ -15,40 +15,43 @@
#include <gflags/gflags.h>
#include <gtest/gtest.h>
#include <vector>
#include "lite/api/cxx_api.h"
#include "lite/api/lite_api_test_helper.h"
#include "lite/api/paddle_api.h"
#include "lite/api/paddle_use_kernels.h"
#include "lite/api/paddle_use_ops.h"
#include "lite/api/paddle_use_passes.h"
#include "lite/api/test_helper.h"
#include "lite/core/op_registry.h"
#include "lite/core/tensor.h"
#include "lite/utils/cp_logging.h"
namespace paddle {
namespace lite {
#ifdef LITE_WITH_X86
TEST(CXXApi, test_lite_googlenet) {
lite::Predictor predictor;
std::vector<Place> valid_places({Place{TARGET(kX86), PRECISION(kFloat)}});
lite_api::CxxConfig config;
config.set_model_dir(FLAGS_model_dir);
config.set_valid_places({lite_api::Place{TARGET(kX86), PRECISION(kFloat)},
lite_api::Place{TARGET(kHost), PRECISION(kFloat)}});
auto predictor = lite_api::CreatePaddlePredictor(config);
// LOG(INFO)<<"FLAGS_eval_googlenet_dir:"<<FLAGS_test_lite_googlenet_dir;
std::string model_dir = FLAGS_model_dir;
predictor.Build(model_dir, "", "", valid_places);
auto* input_tensor = predictor.GetInput(0);
input_tensor->Resize(DDim(std::vector<DDim::value_type>({1, 3, 224, 224})));
auto input_tensor = predictor->GetInput(0);
std::vector<int64_t> input_shape{1, 3, 224, 224};
input_tensor->Resize(input_shape);
auto* data = input_tensor->mutable_data<float>();
for (int i = 0; i < input_tensor->dims().production(); i++) {
int input_num = 1;
for (int i = 0; i < input_shape.size(); ++i) {
input_num *= input_shape[i];
}
for (int i = 0; i < input_num; i++) {
data[i] = 1;
}
for (int i = 0; i < FLAGS_warmup; ++i) {
predictor.Run();
predictor->Run();
}
auto start = GetCurrentUS();
for (int i = 0; i < FLAGS_repeats; ++i) {
predictor.Run();
predictor->Run();
}
LOG(INFO) << "================== Speed Report ===================";
......@@ -56,8 +59,8 @@ TEST(CXXApi, test_lite_googlenet) {
<< ", warmup: " << FLAGS_warmup << ", repeats: " << FLAGS_repeats
<< ", spend " << (GetCurrentUS() - start) / FLAGS_repeats / 1000.0
<< " ms in average.";
auto* out = predictor.GetOutput(0);
// TODO(FrostML): open this check.
/*auto out = predictor->GetOutput(0);
std::vector<float> results(
{0.00034298553, 0.0008200012, 0.0005046297, 0.000839279,
0.00052616704, 0.0003447803, 0.0010877076, 0.00081762316,
......@@ -67,9 +70,9 @@ TEST(CXXApi, test_lite_googlenet) {
for (size_t i = 0; i < results.size(); ++i) {
EXPECT_NEAR(out->data<float>()[i * 51], results[i], 1e-5);
}
ASSERT_EQ(out->dims().size(), 2);
ASSERT_EQ(out->dims()[0], 1);
ASSERT_EQ(out->dims()[1], 1000);
ASSERT_EQ(out->shape().size(), 2);
ASSERT_EQ(out->shape()[0], 1);
ASSERT_EQ(out->shape()[1], 1000);*/
}
#endif
} // namespace lite
......
lite/api/test_inceptionv4_lite_x86.cc
浏览文件 @ 59b80b23
......@@ -12,64 +12,46 @@
// See the License for the specific language governing permissions and
// limitations under the License.
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include <gflags/gflags.h>
#include <gtest/gtest.h>
#include <vector>
#include "lite/api/cxx_api.h"
#include "lite/api/lite_api_test_helper.h"
#include "lite/api/paddle_api.h"
#include "lite/api/paddle_use_kernels.h"
#include "lite/api/paddle_use_ops.h"
#include "lite/api/paddle_use_passes.h"
#include "lite/api/test_helper.h"
#include "lite/core/op_registry.h"
#include "lite/core/tensor.h"
#include "lite/utils/cp_logging.h"
namespace paddle {
namespace lite {
TEST(InceptionV4, test_inceptionv4_lite_x86) {
lite::Predictor predictor;
std::vector<Place> valid_places({Place{TARGET(kX86), PRECISION(kFloat)}});
lite_api::CxxConfig config;
config.set_model_dir(FLAGS_model_dir);
config.set_valid_places({lite_api::Place{TARGET(kX86), PRECISION(kFloat)},
lite_api::Place{TARGET(kHost), PRECISION(kFloat)}});
auto predictor = lite_api::CreatePaddlePredictor(config);
// LOG(INFO)<<"FLAGS_eval_googlenet_dir:"<<FLAGS_test_lite_googlenet_dir;
std::string model_dir = FLAGS_model_dir;
std::vector<std::string> passes({"static_kernel_pick_pass",
"variable_place_inference_pass",
"type_target_cast_pass",
"variable_place_inference_pass",
"io_copy_kernel_pick_pass",
"variable_place_inference_pass",
"runtime_context_assign_pass"});
predictor.Build(model_dir, "", "", valid_places, passes);
auto* input_tensor = predictor.GetInput(0);
input_tensor->Resize(DDim(std::vector<DDim::value_type>({1, 3, 224, 224})));
auto input_tensor = predictor->GetInput(0);
std::vector<int64_t> input_shape{1, 3, 224, 224};
input_tensor->Resize(input_shape);
auto* data = input_tensor->mutable_data<float>();
for (int i = 0; i < input_tensor->dims().production(); i++) {
int input_num = 1;
for (int i = 0; i < input_shape.size(); ++i) {
input_num *= input_shape[i];
}
for (int i = 0; i < input_num; i++) {
data[i] = 1;
}
for (int i = 0; i < FLAGS_warmup; ++i) {
predictor.Run();
predictor->Run();
}
auto start = GetCurrentUS();
for (int i = 0; i < FLAGS_repeats; ++i) {
predictor.Run();
predictor->Run();
}
LOG(INFO) << "================== Speed Report ===================";
......@@ -77,8 +59,8 @@ TEST(InceptionV4, test_inceptionv4_lite_x86) {
<< ", repeats: " << FLAGS_repeats << ", spend "
<< (GetCurrentUS() - start) / FLAGS_repeats / 1000.0
<< " ms in average.";
std::vector<std::vector<float>> results;
// TODO(FrostML): open this check.
/*std::vector<std::vector<float>> results;
// i = 1
results.emplace_back(std::vector<float>(
{0.0011684548, 0.0010390386, 0.0011301535, 0.0010133048,
......@@ -87,19 +69,19 @@ TEST(InceptionV4, test_inceptionv4_lite_x86) {
0.0009782845, 0.0009230255, 0.0010548076, 0.0010974824,
0.0010612885, 0.00089107914, 0.0010112736, 0.00097655767}));
auto* out = predictor.GetOutput(0);
ASSERT_EQ(out->dims().size(), 2);
ASSERT_EQ(out->dims()[0], 1);
ASSERT_EQ(out->dims()[1], 1000);
auto out = predictor->GetOutput(0);
ASSERT_EQ(out->shape().size(), 2);
ASSERT_EQ(out->shape()[0], 1);
ASSERT_EQ(out->shape()[1], 1000);
int step = 50;
for (int i = 0; i < results.size(); ++i) {
for (int j = 0; j < results[i].size(); ++j) {
EXPECT_NEAR(out->data<float>()[j * step + (out->dims()[1] * i)],
EXPECT_NEAR(out->data<float>()[j * step + (out->shape()[1] * i)],
results[i][j],
1e-6);
}
}
}*/
}
} // namespace lite
......
lite/api/test_mobilenetv1_lite_x86.cc
浏览文件 @ 59b80b23
......@@ -12,62 +12,46 @@
// See the License for the specific language governing permissions and
// limitations under the License.
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include <gflags/gflags.h>
#include <gtest/gtest.h>
#include <vector>
#include "lite/api/cxx_api.h"
#include "lite/api/lite_api_test_helper.h"
#include "lite/api/paddle_api.h"
#include "lite/api/paddle_use_kernels.h"
#include "lite/api/paddle_use_ops.h"
#include "lite/api/paddle_use_passes.h"
#include "lite/api/test_helper.h"
#include "lite/core/op_registry.h"
#include "lite/core/tensor.h"
#include "lite/utils/cp_logging.h"
namespace paddle {
namespace lite {
TEST(Mobilenet_v1, test_mobilenetv1_lite_x86) {
lite::Predictor predictor;
std::vector<Place> valid_places({Place{TARGET(kX86), PRECISION(kFloat)}});
lite_api::CxxConfig config;
config.set_model_dir(FLAGS_model_dir);
config.set_valid_places({lite_api::Place{TARGET(kX86), PRECISION(kFloat)},
lite_api::Place{TARGET(kHost), PRECISION(kFloat)}});
auto predictor = lite_api::CreatePaddlePredictor(config);
std::string model_dir = FLAGS_model_dir;
std::vector<std::string> passes({"static_kernel_pick_pass",
"variable_place_inference_pass",
"type_target_cast_pass",
"variable_place_inference_pass",
"io_copy_kernel_pick_pass",
"variable_place_inference_pass",
"runtime_context_assign_pass"});
predictor.Build(model_dir, "", "", valid_places, passes);
auto* input_tensor = predictor.GetInput(0);
input_tensor->Resize(DDim(std::vector<DDim::value_type>({1, 3, 224, 224})));
auto input_tensor = predictor->GetInput(0);
std::vector<int64_t> input_shape{1, 3, 224, 224};
input_tensor->Resize(input_shape);
auto* data = input_tensor->mutable_data<float>();
for (int i = 0; i < input_tensor->dims().production(); i++) {
int input_num = 1;
for (int i = 0; i < input_shape.size(); ++i) {
input_num *= input_shape[i];
}
for (int i = 0; i < input_num; i++) {
data[i] = 1;
}
for (int i = 0; i < FLAGS_warmup; ++i) {
predictor.Run();
predictor->Run();
}
auto start = GetCurrentUS();
for (int i = 0; i < FLAGS_repeats; ++i) {
predictor.Run();
predictor->Run();
}
LOG(INFO) << "================== Speed Report ===================";
......@@ -75,8 +59,8 @@ TEST(Mobilenet_v1, test_mobilenetv1_lite_x86) {
<< ", repeats: " << FLAGS_repeats << ", spend "
<< (GetCurrentUS() - start) / FLAGS_repeats / 1000.0
<< " ms in average.";
std::vector<std::vector<float>> results;
// TODO(FrostML): open this check.
/*std::vector<std::vector<float>> results;
// i = 1
results.emplace_back(std::vector<float>(
{0.00019130898, 9.467885e-05, 0.00015971427, 0.0003650665,
......@@ -84,19 +68,19 @@ TEST(Mobilenet_v1, test_mobilenetv1_lite_x86) {
0.0010323516, 0.00010079765, 0.00011006987, 0.0017364529,
0.0048292773, 0.0013995157, 0.0018453331, 0.0002428986,
0.00020211363, 0.00013668182, 0.0005855956, 0.00025901722}));
auto* out = predictor.GetOutput(0);
ASSERT_EQ(out->dims().size(), 2);
ASSERT_EQ(out->dims()[0], 1);
ASSERT_EQ(out->dims()[1], 1000);
auto out = predictor->GetOutput(0);
ASSERT_EQ(out->shape().size(), 2);
ASSERT_EQ(out->shape()[0], 1);
ASSERT_EQ(out->shape()[1], 1000);
int step = 50;
for (int i = 0; i < results.size(); ++i) {
for (int j = 0; j < results[i].size(); ++j) {
EXPECT_NEAR(out->data<float>()[j * step + (out->dims()[1] * i)],
EXPECT_NEAR(out->data<float>()[j * step + (out->shape()[1] * i)],
results[i][j],
1e-6);
}
}
}*/
}
} // namespace lite
......
lite/api/test_mobilenetv2_lite_x86.cc
浏览文件 @ 59b80b23
......@@ -12,66 +12,47 @@
// See the License for the specific language governing permissions and
// limitations under the License.
// Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include <gflags/gflags.h>
#include <gtest/gtest.h>
#include <vector>
#include "lite/api/cxx_api.h"
#include "lite/api/lite_api_test_helper.h"
#include "lite/api/paddle_api.h"
#include "lite/api/paddle_use_kernels.h"
#include "lite/api/paddle_use_ops.h"
#include "lite/api/paddle_use_passes.h"
#include "lite/api/test_helper.h"
#include "lite/core/op_registry.h"
#include "lite/core/tensor.h"
#include "lite/utils/cp_logging.h"
// for googlenet
namespace paddle {
namespace lite {
TEST(Mobilenet_v2, test_mobilenetv2_lite_x86) {
lite::Predictor predictor;
std::vector<Place> valid_places({Place{TARGET(kX86), PRECISION(kFloat)},
Place{TARGET(kHost), PRECISION(kFloat)}});
lite_api::CxxConfig config;
config.set_model_dir(FLAGS_model_dir);
config.set_valid_places({lite_api::Place{TARGET(kX86), PRECISION(kFloat)},
lite_api::Place{TARGET(kHost), PRECISION(kFloat)}});
auto predictor = lite_api::CreatePaddlePredictor(config);
// LOG(INFO)<<"FLAGS_eval_googlenet_dir:"<<FLAGS_test_lite_googlenet_dir;
std::string model_dir = FLAGS_model_dir;
std::vector<std::string> passes({"static_kernel_pick_pass",
"variable_place_inference_pass",
"type_target_cast_pass",
"variable_place_inference_pass",
"io_copy_kernel_pick_pass",
"variable_place_inference_pass",
"runtime_context_assign_pass"});
predictor.Build(model_dir, "", "", valid_places, passes);
auto* input_tensor = predictor.GetInput(0);
input_tensor->Resize(DDim(std::vector<DDim::value_type>({1, 3, 224, 224})));
auto input_tensor = predictor->GetInput(0);
std::vector<int64_t> input_shape{1, 3, 224, 224};
input_tensor->Resize(input_shape);
auto* data = input_tensor->mutable_data<float>();
for (int i = 0; i < input_tensor->dims().production(); i++) {
int input_num = 1;
for (int i = 0; i < input_shape.size(); ++i) {
input_num *= input_shape[i];
}
for (int i = 0; i < input_num; i++) {
data[i] = 1;
}
for (int i = 0; i < FLAGS_warmup; ++i) {
predictor.Run();
predictor->Run();
}
auto start = GetCurrentUS();
for (int i = 0; i < FLAGS_repeats; ++i) {
predictor.Run();
predictor->Run();
}
LOG(INFO) << "================== Speed Report ===================";
......@@ -79,8 +60,8 @@ TEST(Mobilenet_v2, test_mobilenetv2_lite_x86) {
<< ", repeats: " << FLAGS_repeats << ", spend "
<< (GetCurrentUS() - start) / FLAGS_repeats / 1000.0
<< " ms in average.";
std::vector<std::vector<float>> results;
// TODO(FrostML): open this check.
/*std::vector<std::vector<float>> results;
// i = 1
results.emplace_back(std::vector<float>(
{0.00017082224, 5.699624e-05, 0.000260885, 0.00016412718,
......@@ -88,19 +69,19 @@ TEST(Mobilenet_v2, test_mobilenetv2_lite_x86) {
0.0009059976, 9.5378724e-05, 5.386537e-05, 0.0006427285,
0.0070957416, 0.0016094646, 0.0018807327, 0.00010506048,
6.823785e-05, 0.00012269315, 0.0007806194, 0.00022354358}));
auto* out = predictor.GetOutput(0);
ASSERT_EQ(out->dims().size(), 2);
ASSERT_EQ(out->dims()[0], 1);
ASSERT_EQ(out->dims()[1], 1000);
auto out = predictor->GetOutput(0);
ASSERT_EQ(out->shape().size(), 2);
ASSERT_EQ(out->shape()[0], 1);
ASSERT_EQ(out->shape()[1], 1000);
int step = 50;
for (int i = 0; i < results.size(); ++i) {
for (int j = 0; j < results[i].size(); ++j) {
EXPECT_NEAR(out->data<float>()[j * step + (out->dims()[1] * i)],
EXPECT_NEAR(out->data<float>()[j * step + (out->shape()[1] * i)],
results[i][j],
1e-6);
}
}
}*/
}
} // namespace lite
......
lite/api/test_resnet50_lite_x86.cc
浏览文件 @ 59b80b23
......@@ -15,47 +15,43 @@
#include <gflags/gflags.h>
#include <gtest/gtest.h>
#include <vector>
#include "lite/api/cxx_api.h"
#include "lite/api/lite_api_test_helper.h"
#include "lite/api/paddle_api.h"
#include "lite/api/paddle_use_kernels.h"
#include "lite/api/paddle_use_ops.h"
#include "lite/api/paddle_use_passes.h"
#include "lite/api/test_helper.h"
#include "lite/core/op_registry.h"
#include "lite/utils/cp_logging.h"
namespace paddle {
namespace lite {
TEST(Resnet50, test_resnet50_lite_x86) {
// DeviceInfo::Init();
// DeviceInfo::Global().SetRunMode(lite_api::LITE_POWER_HIGH, FLAGS_threads);
lite::Predictor predictor;
std::vector<Place> valid_places({Place{TARGET(kX86), PRECISION(kFloat)},
Place{TARGET(kHost), PRECISION(kFloat)}});
lite_api::CxxConfig config;
config.set_model_dir(FLAGS_model_dir);
config.set_valid_places({lite_api::Place{TARGET(kX86), PRECISION(kFloat)},
lite_api::Place{TARGET(kHost), PRECISION(kFloat)}});
auto predictor = lite_api::CreatePaddlePredictor(config);
std::string model_dir = FLAGS_model_dir;
std::vector<std::string> passes({"static_kernel_pick_pass",
"variable_place_inference_pass",
"type_target_cast_pass",
"variable_place_inference_pass",
"io_copy_kernel_pick_pass",
"variable_place_inference_pass",
"runtime_context_assign_pass"});
predictor.Build(model_dir, "", "", valid_places, passes);
auto* input_tensor = predictor.GetInput(0);
input_tensor->Resize(DDim(std::vector<DDim::value_type>({1, 3, 224, 224})));
auto input_tensor = predictor->GetInput(0);
std::vector<int64_t> input_shape{1, 3, 224, 224};
input_tensor->Resize(input_shape);
auto* data = input_tensor->mutable_data<float>();
auto item_size = input_tensor->dims().production();
for (int i = 0; i < item_size; i++) {
int input_num = 1;
for (int i = 0; i < input_shape.size(); ++i) {
input_num *= input_shape[i];
}
for (int i = 0; i < input_num; i++) {
data[i] = 1;
}
for (int i = 0; i < FLAGS_warmup; ++i) {
predictor.Run();
predictor->Run();
}
auto start = GetCurrentUS();
for (int i = 0; i < FLAGS_repeats; ++i) {
predictor.Run();
predictor->Run();
}
LOG(INFO) << "================== Speed Report ===================";
......@@ -64,7 +60,8 @@ TEST(Resnet50, test_resnet50_lite_x86) {
<< ", spend " << (GetCurrentUS() - start) / FLAGS_repeats / 1000.0
<< " ms in average.";
std::vector<std::vector<float>> results;
// TODO(FrostML): open this check.
/*std::vector<std::vector<float>> results;
// i = 1
results.emplace_back(std::vector<float>(
{0.00024139918, 0.00020566184, 0.00022418296, 0.00041731037,
......@@ -72,19 +69,19 @@ TEST(Resnet50, test_resnet50_lite_x86) {
0.00072681636, 8.531815e-05, 0.0002129998, 0.0021168243,
0.006387163, 0.0037145028, 0.0012812682, 0.00045948103,
0.00013535398, 0.0002483765, 0.00076759676, 0.0002773295}));
auto* out = predictor.GetOutput(0);
ASSERT_EQ(out->dims().size(), 2);
ASSERT_EQ(out->dims()[0], 1);
ASSERT_EQ(out->dims()[1], 1000);
auto out = predictor->GetOutput(0);
ASSERT_EQ(out->shape().size(), 2);
ASSERT_EQ(out->shape()[0], 1);
ASSERT_EQ(out->shape()[1], 1000);
int step = 50;
for (int i = 0; i < results.size(); ++i) {
for (int j = 0; j < results[i].size(); ++j) {
EXPECT_NEAR(out->data<float>()[j * step + (out->dims()[1] * i)],
EXPECT_NEAR(out->data<float>()[j * step + (out->shape()[1] * i)],
results[i][j],
1e-6);
}
}
}*/
}
} // namespace lite
......
lite/core/program.cc
浏览文件 @ 59b80b23
......@@ -196,11 +196,14 @@ void Instruction::Run() {
if (op_->run_once() && has_run_) {
return;
}
#ifndef LITE_SHUTDOWN_LOG
VLOG(4) << "kernel launch";
#endif
op_->InferShape();
#ifndef LITE_SHUTDOWN_LOG
VLOG(4) << ">> Running kernel: " << op_->op_info()->Repr() << " on Target "
<< TargetToStr(kernel_->target());
#endif
kernel_->Launch();
has_run_ = true;
}
......
lite/kernels/xpu/bridges/conv_op.cc
浏览文件 @ 59b80b23
......@@ -136,7 +136,7 @@ node_map_type ConvConverter(const std::shared_ptr<lite::OpLite> op,
std::shared_ptr<xtcl::xExpr> add_node = nullptr;
if (is_channel_bias) {
add_node = std::make_shared<xtcl::xExpr>(
graph_ctx->builder->CreateBiasAdd(*conv_node, *bias_node, 1));
graph_ctx->builder->CreateBiasAdd(*conv_node, 1, *bias_node));
} else {
add_node = std::make_shared<xtcl::xExpr>(
graph_ctx->builder->CreateBinaryOp("add", *conv_node, *bias_node));
......
lite/kernels/xpu/bridges/elementwise_ops.cc
浏览文件 @ 59b80b23
......@@ -69,7 +69,7 @@ node_map_type ElementwiseConverter(const std::shared_ptr<lite::OpLite> op,
std::shared_ptr<xtcl::xExpr> elementwise_node = nullptr;
if (y_dims.size() == 1) {
elementwise_node = std::make_shared<xtcl::xExpr>(
graph_ctx->builder->CreateBiasAdd(*x_node, *y_node, axis));
graph_ctx->builder->CreateBiasAdd(*x_node, axis, *y_node));
} else if (x_dims.size() == y_dims.size()) {
elementwise_node = std::make_shared<xtcl::xExpr>(
graph_ctx->builder->CreateBinaryOp("add", *x_node, *y_node));
......
lite/kernels/xpu/bridges/mul_op.cc
浏览文件 @ 59b80b23
......@@ -78,8 +78,11 @@ node_map_type MulConverter(const std::shared_ptr<lite::OpLite> op,
// create mul node and set params from op
std::shared_ptr<xtcl::xExpr> mul_node = nullptr;
mul_node = std::make_shared<xtcl::xExpr>(graph_ctx->builder->CreateDense(
*x_node, *y_const_node, static_cast<int>(y_dims[1])));
mul_node = std::make_shared<xtcl::xExpr>(
graph_ctx->builder->CreateDense(*x_node,
static_cast<int>(y_dims[1]),
::xtcl::NullValue<::xtcl::DataType>(),
*y_const_node));
graph_ctx->builder->SetLayer(unique_op_type);
// output converted nodes
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或