experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

TuningType moved to gpu/common/task.

Browse Source 
PiperOrigin-RevId: 342770347
Change-Id: If0fd6174fb650002e9262fe13cb5a409989a9593
This commit is contained in:
Raman Sarokin 2020-11-16 19:08:01 -08:00 committed by TensorFlower Gardener
parent f0c976a27f
commit faad89dadf
15 changed files with 47 additions and 59 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
tensorflow/lite/delegates/gpu/cl/cl_operation.cc
View File

@ -114,10 +114,11 @@ absl::Status ClOperation::CompileDeserialized(
*creation_context.context, *creation_context.device, &kernel_);
}
absl::Status ClOperation::Tune(const TuningParameters& params) {
absl::Status ClOperation::Tune(TuningType tuning_type, const GpuInfo& gpu_info,
ProfilingCommandQueue* profiling_queue) {
std::vector<int3> possible_work_groups;
operation_->GetPossibleKernelWorkGroups(params.tuning_type, *params.info,
kernel_.info_, &possible_work_groups);
operation_->GetPossibleKernelWorkGroups(tuning_type, gpu_info, kernel_.info_,
&possible_work_groups);
if (possible_work_groups.empty()) {
return absl::NotFoundError(
"Can not found work_group size to launch kernel");
@ -137,8 +138,8 @@ absl::Status ClOperation::Tune(const TuningParameters& params) {
}
RETURN_IF_ERROR(cl_args_.Bind(kernel_.kernel()));
int best_work_group_index;
RETURN_IF_ERROR(params.queue->GetBestWorkGroupIndex(
kernel_, *params.info, work_groups_count, possible_work_groups,
RETURN_IF_ERROR(profiling_queue->GetBestWorkGroupIndex(
kernel_, gpu_info, work_groups_count, possible_work_groups,
&best_work_group_index));
operation_->work_group_size_ = possible_work_groups[best_work_group_index];
operation_->work_groups_count_ = GetWorkGroupsCount(

3
tensorflow/lite/delegates/gpu/cl/cl_operation.h
View File

@ -71,7 +71,8 @@ class ClOperation {
operation_->work_group_size_);
}
absl::Status Tune(const TuningParameters& params);
absl::Status Tune(TuningType tuning_type, const GpuInfo& gpu_info,
ProfilingCommandQueue* profiling_queue);
absl::Status Compile(const CreationContext& creation_context);

22
tensorflow/lite/delegates/gpu/cl/inference_context.cc
View File

@ -183,21 +183,20 @@ absl::Status InferenceContext::InitFromGraph(
RETURN_IF_ERROR(Compile(creation_context));
RETURN_IF_ERROR(UpdateParams());
TuningParameters tuning_parameters;
tuning_parameters.queue = env->profiling_queue();
tuning_parameters.info = &env->device().info_;
TuningType tuning_type = TuningType::kExhaustive;
if (create_info.hints.Check(ModelHints::kFastTuning)) {
tuning_parameters.tuning_type = TuningType::FAST;
tuning_type = TuningType::kFast;
}
if (tuning_parameters.info->IsMali()) {
const MaliInfo& info = tuning_parameters.info->mali_info;
if (env->device().GetInfo().IsMali()) {
const MaliInfo& info = env->device().GetInfo().mali_info;
if (info.IsMaliT6xx()) {
// Mali T628 hangs forever in clFinish when used profiling queue
// TuningType::FAST does not use profiling queue.
tuning_parameters.tuning_type = TuningType::FAST;
tuning_type = TuningType::kFast;
}
}
RETURN_IF_ERROR(Tune(tuning_parameters));
RETURN_IF_ERROR(
Tune(tuning_type, env->device().GetInfo(), env->profiling_queue()));
if (serialized_model) {
for (auto& node : nodes_) {
@ -631,9 +630,12 @@ absl::Status InferenceContext::Compile(
return absl::OkStatus();
}
absl::Status InferenceContext::Tune(const TuningParameters& tuning_parameters) {
absl::Status InferenceContext::Tune(TuningType tuning_type,
const GpuInfo& gpu_info,
ProfilingCommandQueue* profiling_queue) {
for (auto& node : nodes_) {
RETURN_IF_ERROR(node.cl_operation.Tune(tuning_parameters));
RETURN_IF_ERROR(
node.cl_operation.Tune(tuning_type, gpu_info, profiling_queue));
}
return absl::OkStatus();
}

3
tensorflow/lite/delegates/gpu/cl/inference_context.h
View File

@ -133,7 +133,8 @@ class InferenceContext {
void BindMemoryToOperations();
absl::Status Compile(const CreationContext& creation_context);
absl::Status Tune(const TuningParameters& tuning_parameters);
absl::Status Tune(TuningType tuning_type, const GpuInfo& gpu_info,
ProfilingCommandQueue* profiling_queue);
absl::Status UpdateParams();
// performance hacks

15
tensorflow/lite/delegates/gpu/cl/kernels/BUILD
View File

@ -580,7 +580,6 @@ cc_library(
hdrs = ["fully_connected.h"],
deps = [
":gpu_operation",
":tuning_parameters",
":util",
"//tensorflow/lite/delegates/gpu/cl:buffer",
"//tensorflow/lite/delegates/gpu/cl:cl_kernel",
@ -624,9 +623,9 @@ cc_library(
srcs = ["gpu_operation.cc"],
hdrs = ["gpu_operation.h"],
deps = [
":tuning_parameters",
":util",
":work_group_picking",
"//tensorflow/lite/delegates/gpu/cl:cl_kernel",
"//tensorflow/lite/delegates/gpu/cl:device_info",
"//tensorflow/lite/delegates/gpu/cl:serialization_cc_fbs",
"//tensorflow/lite/delegates/gpu/common:access_type",
@ -638,6 +637,7 @@ cc_library(
"//tensorflow/lite/delegates/gpu/common/task:buffer_desc",
"//tensorflow/lite/delegates/gpu/common/task:gpu_tensor",
"//tensorflow/lite/delegates/gpu/common/task:tensor_desc",
"//tensorflow/lite/delegates/gpu/common/task:tuning_type",
"@com_google_absl//absl/strings",
],
)
@ -1204,15 +1204,6 @@ cc_test(
],
)
cc_library(
name = "tuning_parameters",
hdrs = ["tuning_parameters.h"],
deps = [
"//tensorflow/lite/delegates/gpu/cl:cl_command_queue",
"//tensorflow/lite/delegates/gpu/cl:device_info",
],
)
cc_library(
name = "resize",
srcs = ["resize.cc"],
@ -1306,12 +1297,12 @@ cc_library(
srcs = ["work_group_picking.cc"],
hdrs = ["work_group_picking.h"],
deps = [
":tuning_parameters",
"//tensorflow/lite/delegates/gpu/cl:cl_kernel",
"//tensorflow/lite/delegates/gpu/cl:device_info",
"//tensorflow/lite/delegates/gpu/common:types",
"//tensorflow/lite/delegates/gpu/common:util",
"//tensorflow/lite/delegates/gpu/common:workgroup_selection",
"//tensorflow/lite/delegates/gpu/common/task:tuning_type",
],
)

2
tensorflow/lite/delegates/gpu/cl/kernels/converter.cc
View File

@ -54,7 +54,7 @@ class OpenClConverterImpl : public TensorObjectConverter {
const int3 grid = int3(tensor->Width() * tensor->Batch(), tensor->Height(),
tensor->Slices());
std::vector<int3> work_groups;
GetPossibleWorkGroupsConv(TuningType::FAST, gpu_info_, kernel_.info_, grid,
GetPossibleWorkGroupsConv(TuningType::kFast, gpu_info_, kernel_.info_, grid,
&work_groups);
const int3 work_group_size = work_groups[0];
const int3 work_groups_count = GetWorkGroupsCount(grid, work_group_size);

1
tensorflow/lite/delegates/gpu/cl/kernels/fully_connected.h
View File

@ -27,7 +27,6 @@ limitations under the License.
#include "tensorflow/lite/delegates/gpu/cl/cl_kernel.h"
#include "tensorflow/lite/delegates/gpu/cl/device_info.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/gpu_operation.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/tuning_parameters.h"
#include "tensorflow/lite/delegates/gpu/cl/texture2d.h"
#include "tensorflow/lite/delegates/gpu/common/data_type.h"
#include "tensorflow/lite/delegates/gpu/common/operations.h"

3
tensorflow/lite/delegates/gpu/cl/kernels/gpu_operation.h
View File

@ -19,8 +19,8 @@ limitations under the License.
#include <string>
#include <vector>
#include "tensorflow/lite/delegates/gpu/cl/cl_kernel.h"
#include "tensorflow/lite/delegates/gpu/cl/device_info.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/tuning_parameters.h"
#include "tensorflow/lite/delegates/gpu/cl/serialization_generated.h"
#include "tensorflow/lite/delegates/gpu/common/data_type.h"
#include "tensorflow/lite/delegates/gpu/common/precision.h"
@ -29,6 +29,7 @@ limitations under the License.
#include "tensorflow/lite/delegates/gpu/common/task/buffer_desc.h"
#include "tensorflow/lite/delegates/gpu/common/task/gpu_tensor.h"
#include "tensorflow/lite/delegates/gpu/common/task/tensor_desc.h"
#include "tensorflow/lite/delegates/gpu/common/task/tuning_type.h"
#include "tensorflow/lite/delegates/gpu/common/types.h"
namespace tflite {

1
tensorflow/lite/delegates/gpu/cl/kernels/special/BUILD
View File

@ -36,7 +36,6 @@ cc_library(
"//tensorflow/lite/delegates/gpu/cl:tensor",
"//tensorflow/lite/delegates/gpu/cl:texture2d",
"//tensorflow/lite/delegates/gpu/cl/kernels:gpu_operation",
"//tensorflow/lite/delegates/gpu/cl/kernels:tuning_parameters",
"//tensorflow/lite/delegates/gpu/cl/kernels:util",
"//tensorflow/lite/delegates/gpu/common:data_type",
"//tensorflow/lite/delegates/gpu/common:operations",

1
tensorflow/lite/delegates/gpu/cl/kernels/special/fc_fc_add.h
View File

@ -27,7 +27,6 @@ limitations under the License.
#include "tensorflow/lite/delegates/gpu/cl/cl_kernel.h"
#include "tensorflow/lite/delegates/gpu/cl/device_info.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/gpu_operation.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/tuning_parameters.h"
#include "tensorflow/lite/delegates/gpu/cl/texture2d.h"
#include "tensorflow/lite/delegates/gpu/common/data_type.h"
#include "tensorflow/lite/delegates/gpu/common/operations.h"

8
tensorflow/lite/delegates/gpu/cl/kernels/winograd.cc
View File

@ -285,11 +285,11 @@ void Winograd4x4To36::GetPossibleKernelWorkGroups(
TuningType tuning_type, const GpuInfo& gpu_info,
const KernelInfo& kernel_info, std::vector<int3>* work_groups) const {
switch (tuning_type) {
case TuningType::EXHAUSTIVE:
case TuningType::kExhaustive:
GetPossibleWorkGroups(tuning_type, gpu_info, kernel_info, grid_size_,
work_groups);
return;
case TuningType::FAST:
case TuningType::kFast:
default:
work_groups->push_back(SelectBestWorkGroup(kernel_info));
return;
@ -481,11 +481,11 @@ void Winograd36To4x4::GetPossibleKernelWorkGroups(
TuningType tuning_type, const GpuInfo& gpu_info,
const KernelInfo& kernel_info, std::vector<int3>* work_groups) const {
switch (tuning_type) {
case TuningType::EXHAUSTIVE:
case TuningType::kExhaustive:
GetPossibleWorkGroups(tuning_type, gpu_info, kernel_info, grid_size_,
work_groups);
return;
case TuningType::FAST:
case TuningType::kFast:
default:
work_groups->push_back(SelectBestWorkGroup(kernel_info));
return;

8
tensorflow/lite/delegates/gpu/cl/kernels/work_group_picking.cc
View File

@ -252,11 +252,11 @@ void GetPossibleWorkGroups(TuningType tuning_type, const GpuInfo& gpu_info,
const KernelInfo& kernel_info, const int3& grid,
std::vector<int3>* work_groups) {
switch (tuning_type) {
case TuningType::FAST:
case TuningType::kFast:
work_groups->push_back(
GetWorkGroup(grid, kernel_info.max_work_group_size));
return;
case TuningType::EXHAUSTIVE: {
case TuningType::kExhaustive: {
GetWorkGroupsAlignedToGrid(gpu_info, kernel_info, grid, work_groups);
return;
}
@ -270,7 +270,7 @@ void GetPossibleWorkGroupsConv(TuningType tuning_type, const GpuInfo& gpu_info,
const KernelInfo& kernel_info, const int3& grid,
std::vector<int3>* work_groups) {
switch (tuning_type) {
case TuningType::FAST: {
case TuningType::kFast: {
int max_z_size = 16;
if (gpu_info.IsAdreno()) {
max_z_size = gpu_info.adreno_info.IsAdreno3xx() ? 16 : 64;
@ -280,7 +280,7 @@ void GetPossibleWorkGroupsConv(TuningType tuning_type, const GpuInfo& gpu_info,
GetWorkGroupConv(grid, kernel_info.max_work_group_size, max_z_size));
return;
}
case TuningType::EXHAUSTIVE: {
case TuningType::kExhaustive: {
GetWorkGroupsAlignedToGrid(gpu_info, kernel_info, grid, work_groups);
return;
}

2
tensorflow/lite/delegates/gpu/cl/kernels/work_group_picking.h
View File

@ -20,7 +20,7 @@ limitations under the License.
#include "tensorflow/lite/delegates/gpu/cl/cl_kernel.h"
#include "tensorflow/lite/delegates/gpu/cl/device_info.h"
#include "tensorflow/lite/delegates/gpu/cl/kernels/tuning_parameters.h"
#include "tensorflow/lite/delegates/gpu/common/task/tuning_type.h"
#include "tensorflow/lite/delegates/gpu/common/types.h"
#include "tensorflow/lite/delegates/gpu/common/workgroup_selection.h"

5
tensorflow/lite/delegates/gpu/common/task/BUILD
View File

@ -100,6 +100,11 @@ cc_library(
],
)
cc_library(
name = "tuning_type",
hdrs = ["tuning_type.h"],
)
cc_library(
name = "util",
srcs = ["util.cc"],

21
tensorflow/lite/delegates/gpu/cl/kernels/tuning_parameters.h → tensorflow/lite/delegates/gpu/common/task/tuning_type.h
View File

@ -1,4 +1,4 @@
/* Copyright 2019 The TensorFlow Authors. All Rights Reserved.
/* Copyright 2020 The TensorFlow 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.
@ -13,26 +13,15 @@ See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/
#ifndef TENSORFLOW_LITE_DELEGATES_GPU_CL_KERNELS_TUNING_PARAMETERS_H_
#define TENSORFLOW_LITE_DELEGATES_GPU_CL_KERNELS_TUNING_PARAMETERS_H_
#include "tensorflow/lite/delegates/gpu/cl/cl_command_queue.h"
#include "tensorflow/lite/delegates/gpu/cl/device_info.h"
#ifndef TENSORFLOW_LITE_DELEGATES_GPU_COMMON_TASK_TUNING_TYPE_H_
#define TENSORFLOW_LITE_DELEGATES_GPU_COMMON_TASK_TUNING_TYPE_H_
namespace tflite {
namespace gpu {
namespace cl {
enum class TuningType { EXHAUSTIVE, FAST };
enum class TuningType { kExhaustive, kFast };
struct TuningParameters {
ProfilingCommandQueue* queue;
const GpuInfo* info;
TuningType tuning_type = TuningType::EXHAUSTIVE;
};
} // namespace cl
} // namespace gpu
} // namespace tflite
#endif // TENSORFLOW_LITE_DELEGATES_GPU_CL_KERNELS_TUNING_PARAMETERS_H_
#endif // TENSORFLOW_LITE_DELEGATES_GPU_COMMON_TASK_TUNING_TYPE_H_