Huawei_Technology
/
mindspore
Not watched
1
0
Fork 0
Code
Releases 13 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
Browse Source

reduce or raise precision restructure

tags/v1.1.0
liubuyu 5 years ago
parent
0490a08e66
commit
020c71e72c
1 changed files with 78 additions and 238 deletions
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +78
    -238
      mindspore/ccsrc/runtime/device/ascend/kernel_select_ascend.cc

+ 78
- 238
mindspore/ccsrc/runtime/device/ascend/kernel_select_ascend.cc View File

@@ -16,23 +16,23 @@


#include "runtime/device/ascend/kernel_select_ascend.h" #include "runtime/device/ascend/kernel_select_ascend.h"


#include <string>
#include <vector>
#include <memory>
#include <utility>
#include <algorithm> #include <algorithm>
#include <map> #include <map>
#include <memory>
#include <string>
#include <unordered_map> #include <unordered_map>
#include <unordered_set> #include <unordered_set>
#include "utils/ms_utils.h"
#include <utility>
#include <vector>
#include "backend/kernel_compiler/kernel_build_info.h"
#include "backend/kernel_compiler/kernel_query.h"
#include "backend/kernel_compiler/oplib/oplib.h"
#include "backend/kernel_compiler/tbe/tbe_dynaminc_shape_util.h" #include "backend/kernel_compiler/tbe/tbe_dynaminc_shape_util.h"
#include "backend/session/anf_runtime_algorithm.h"
#include "debug/anf_ir_dump.h" #include "debug/anf_ir_dump.h"
#include "frontend/operator/ops.h" #include "frontend/operator/ops.h"
#include "utils/ms_context.h" #include "utils/ms_context.h"
#include "backend/session/anf_runtime_algorithm.h"
#include "backend/kernel_compiler/kernel_query.h"
#include "backend/kernel_compiler/oplib/oplib.h"
#include "backend/kernel_compiler/kernel_build_info.h"
#include "utils/ms_utils.h"


namespace mindspore { namespace mindspore {
namespace device { namespace device {
@@ -172,218 +172,6 @@ void UpdateCurMatchCounts(const kernel::KernelBuildInfo &kernel_build_info, cons
} }
} }


void AddSupportMixedPrecisionDataTypeIndex(TypeId data_type, std::vector<int> *support_index) {
MS_EXCEPTION_IF_NULL(support_index);
int index = kUnSupportMixedDataTypeIndex;
switch (data_type) {
case kNumberTypeFloat16:
index = 0;
break;
case kNumberTypeFloat32:
case kNumberTypeFloat:
index = 1;
break;
default:
break;
}
support_index->push_back(index);
}

void AddKernelInputSupportDataType(const kernel::KernelBuildInfo &kernel_build_info, size_t input_index,
std::vector<int> *support_datatype_index, std::vector<TypeId> *support_datatype) {
MS_EXCEPTION_IF_NULL(support_datatype);
auto data_type = kernel_build_info.GetInputDeviceType(input_index);
support_datatype->push_back(data_type);
AddSupportMixedPrecisionDataTypeIndex(data_type, support_datatype_index);
}

void AddKernelOutputSupportDataType(const kernel::KernelBuildInfo &kernel_build_info, size_t output_index,
std::vector<int> *support_datatype_index, std::vector<TypeId> *support_datatype) {
MS_EXCEPTION_IF_NULL(support_datatype);
auto data_type = kernel_build_info.GetOutputDeviceType(output_index);
support_datatype->push_back(data_type);
AddSupportMixedPrecisionDataTypeIndex(data_type, support_datatype_index);
}

void AddNodeInputDataType(const CNodePtr &kernel_node, size_t input_index,
std::vector<int> *node_mix_precision_datatype_index,
std::vector<TypeId> *node_mix_precision_datatype) {
AnfNodePtr cur_input = AnfAlgo::GetInputNode(kernel_node, input_index);
MS_EXCEPTION_IF_NULL(cur_input);
MS_EXCEPTION_IF_NULL(node_mix_precision_datatype);
TypeId input_origin_type = AnfAlgo::GetPrevNodeOutputInferDataType(kernel_node, input_index);
AddSupportMixedPrecisionDataTypeIndex(input_origin_type, node_mix_precision_datatype_index);
node_mix_precision_datatype->push_back(input_origin_type);
}

void AddNodeOutputDataType(const CNodePtr &kernel_node, size_t output_index,
std::vector<int> *node_mix_precision_datatype_index,
std::vector<TypeId> *node_mix_precision_datatype) {
MS_EXCEPTION_IF_NULL(node_mix_precision_datatype);
auto output_origin_type = AnfAlgo::GetOutputInferDataType(kernel_node, output_index);
AddSupportMixedPrecisionDataTypeIndex(output_origin_type, node_mix_precision_datatype_index);
node_mix_precision_datatype->push_back(output_origin_type);
}

void CheckDataTypeInputs(const std::vector<int> &node_mix_precision_datatype_index,
const std::vector<TypeId> &node_mix_precision_datatype,
const std::map<size_t, std::vector<TypeId>> &kernel_support_datatypes,
std::map<size_t, std::vector<int>> *kernel_match_datatype_idx) {
if (node_mix_precision_datatype_index.size() != node_mix_precision_datatype.size()) {
MS_LOG(EXCEPTION) << "Node datatype index size " << node_mix_precision_datatype_index.size() << " != datatype size "
<< node_mix_precision_datatype.size();
}
MS_EXCEPTION_IF_NULL(kernel_match_datatype_idx);
if (kernel_support_datatypes.size() != kernel_match_datatype_idx->size()) {
MS_LOG(EXCEPTION) << "Kernel datatype index size " << kernel_match_datatype_idx->size() << " != datatype size "
<< kernel_support_datatypes.size();
}
}

bool RaiseDataTypePrecisionSelect(const std::vector<int> &node_mix_precision_datatype_index,
const std::vector<TypeId> &node_mix_precision_datatype,
const std::map<size_t, std::vector<TypeId>> &kernel_support_datatypes,
std::map<size_t, std::vector<int>> *kernel_match_datatype_idx) {
MS_EXCEPTION_IF_NULL(kernel_match_datatype_idx);
CheckDataTypeInputs(node_mix_precision_datatype_index, node_mix_precision_datatype, kernel_support_datatypes,
kernel_match_datatype_idx);
for (size_t i = 0; i < node_mix_precision_datatype_index.size(); ++i) {
if (node_mix_precision_datatype[i] == kTypeUnknown) {
continue;
}
auto iter = kernel_match_datatype_idx->begin();
while (iter != kernel_match_datatype_idx->end()) {
if (node_mix_precision_datatype_index[i] == kUnSupportMixedDataTypeIndex) {
auto find_iter = kernel_support_datatypes.find(iter->first);
if (find_iter == kernel_support_datatypes.end()) {
MS_LOG(EXCEPTION) << "Kernel datatype index:%lu can not be found " << iter->first;
}
if (i >= find_iter->second.size()) {
MS_LOG(EXCEPTION) << "Node index " << i << "kernel datatype size " << find_iter->second.size();
}
if (node_mix_precision_datatype[i] != find_iter->second[i]) {
iter = kernel_match_datatype_idx->erase(iter);
} else {
++iter;
}
continue;
}
auto datatype_indexes = iter->second;
if (i >= datatype_indexes.size()) {
MS_LOG(EXCEPTION) << "Node datatype index: " << i << " kernel support size " << datatype_indexes.size();
}
if (datatype_indexes[i] < node_mix_precision_datatype_index[i]) {
iter = kernel_match_datatype_idx->erase(iter);
} else {
++iter;
}
}
}
return !kernel_match_datatype_idx->empty();
}

bool CanDataTypeReduce(const std::vector<int> &datatype_indexes, int check_index,
const std::vector<int> &node_mix_precision_datatype_index) {
auto check_index_tmp = IntToSize(check_index);
if (check_index_tmp < datatype_indexes.size() && check_index_tmp < node_mix_precision_datatype_index.size()) {
return datatype_indexes[check_index] != kUnSupportMixedDataTypeIndex &&
datatype_indexes[check_index] <= node_mix_precision_datatype_index[check_index];
}
MS_LOG(EXCEPTION) << "Check index " << check_index << "is outof range";
}

bool RaiseOrReduceDataTypePrecisionSelect(const std::vector<int> &node_mix_precision_datatype_index,
const std::vector<TypeId> &node_mix_precision_datatype,
const std::map<size_t, std::vector<TypeId>> &kernel_support_datatypes,
std::map<size_t, std::vector<int>> *kernel_match_datatype_idx) {
MS_EXCEPTION_IF_NULL(kernel_match_datatype_idx);
CheckDataTypeInputs(node_mix_precision_datatype_index, node_mix_precision_datatype, kernel_support_datatypes,
kernel_match_datatype_idx);
for (size_t i = 0; i < node_mix_precision_datatype_index.size(); ++i) {
if (node_mix_precision_datatype[i] == kTypeUnknown) {
continue;
}
auto iter = kernel_match_datatype_idx->begin();
while (iter != kernel_match_datatype_idx->end()) {
if (node_mix_precision_datatype_index[i] == kUnSupportMixedDataTypeIndex) {
auto find_iter = kernel_support_datatypes.find(iter->first);
if (find_iter == kernel_support_datatypes.end()) {
MS_LOG(EXCEPTION) << "Kernel datatype index:%lu can not be found " << iter->first;
}
if (i >= find_iter->second.size()) {
MS_LOG(EXCEPTION) << "Node index " << i << " >= kernel datatype size " << find_iter->second.size();
}
if (node_mix_precision_datatype[i] != find_iter->second[i]) {
iter = kernel_match_datatype_idx->erase(iter);
} else {
++iter;
}
continue;
}
auto datatype_indexes = iter->second;
if (i >= datatype_indexes.size()) {
MS_LOG(EXCEPTION) << "Index " << i << "> kernel datatype indexes size " << datatype_indexes.size();
}
if (!CanDataTypeReduce(datatype_indexes, i, node_mix_precision_datatype_index)) {
iter = kernel_match_datatype_idx->erase(iter);
} else {
++iter;
}
}
}
return !kernel_match_datatype_idx->empty();
}

void AddNodeAndKernelDataType(const CNodePtr &kernel_node, const kernel::KernelBuildInfo &kernel_build_info,
std::vector<int> *support_indexes, std::vector<TypeId> *node_mix_precision_datatype,
std::vector<TypeId> *support_datatypes,
std::vector<int> *node_mix_precision_datatype_index) {
MS_EXCEPTION_IF_NULL(node_mix_precision_datatype);
bool add_node_datatype_flag = false;
if (node_mix_precision_datatype->empty()) {
add_node_datatype_flag = true;
}
for (size_t input_index = 0; input_index < kernel_build_info.GetInputNum(); ++input_index) {
AddKernelInputSupportDataType(kernel_build_info, input_index, support_indexes, support_datatypes);
if (add_node_datatype_flag) {
AddNodeInputDataType(kernel_node, input_index, node_mix_precision_datatype_index, node_mix_precision_datatype);
}
}
// Check output data type
for (size_t output_index = 0; output_index < kernel_build_info.GetOutputNum(); ++output_index) {
AddKernelOutputSupportDataType(kernel_build_info, output_index, support_indexes, support_datatypes);
if (add_node_datatype_flag) {
AddNodeOutputDataType(kernel_node, output_index, node_mix_precision_datatype_index, node_mix_precision_datatype);
}
}
}

void PrecisionReduce(const std::vector<int> &node_mix_precision_datatype_index,
const std::vector<TypeId> &node_mix_precision_datatype,
const std::map<size_t, std::vector<TypeId>> &kernel_support_datatype,
std::map<size_t, std::vector<int>> *kernel_match_datatype_idx, bool *precision_reduce) {
MS_EXCEPTION_IF_NULL(kernel_match_datatype_idx);
auto context_ptr = MsContext::GetInstance();
MS_EXCEPTION_IF_NULL(context_ptr);
MS_EXCEPTION_IF_NULL(precision_reduce);
std::map<size_t, std::vector<int>> kernel_match_datatype_idx_copy = *kernel_match_datatype_idx;
// raise precision
bool selected_ret = RaiseDataTypePrecisionSelect(node_mix_precision_datatype_index, node_mix_precision_datatype,
kernel_support_datatype, kernel_match_datatype_idx);
if (selected_ret) {
*precision_reduce = false;
return;
}
if (context_ptr->get_param<bool>(MS_CTX_ENABLE_REDUCE_PRECISION)) {
selected_ret = RaiseOrReduceDataTypePrecisionSelect(node_mix_precision_datatype_index, node_mix_precision_datatype,
kernel_support_datatype, &kernel_match_datatype_idx_copy);
}
if (selected_ret) {
*precision_reduce = true;
*kernel_match_datatype_idx = kernel_match_datatype_idx_copy;
}
}

void PrintRaiseOrReducePrecisionSelectedInfo(const CNodePtr &cnode, void PrintRaiseOrReducePrecisionSelectedInfo(const CNodePtr &cnode,
const std::shared_ptr<kernel::KernelBuildInfo> &selected_kernel_build_info, const std::shared_ptr<kernel::KernelBuildInfo> &selected_kernel_build_info,
bool precision_reduce) { bool precision_reduce) {
@@ -434,30 +222,82 @@ std::vector<std::shared_ptr<kernel::KernelBuildInfo>> FilteredKernelInfoByDtype(
return result; return result;
} }


bool TagRaiseReduce(const std::shared_ptr<kernel::KernelBuildInfo> &kernel_build_info, const CNodePtr &cnode,
const std::map<TypeId, TypeId> &type_map) {
MS_EXCEPTION_IF_NULL(cnode);
MS_EXCEPTION_IF_NULL(kernel_build_info);
size_t flag_in = 0;
size_t flag_out = 0;
for (size_t input_index = 0; input_index < kernel_build_info->GetInputNum(); ++input_index) {
auto in_dtype = AnfAlgo::GetPrevNodeOutputInferDataType(cnode, input_index);
auto device_dtype = kernel_build_info->GetInputDeviceType(input_index);
if (device_dtype == kNumberTypeFloat || device_dtype == kNumberTypeFloat32) {
device_dtype = kNumberTypeFloat32;
}
auto iter = type_map.find(in_dtype);
if (iter != type_map.end() && iter->second != device_dtype && in_dtype != device_dtype) {
return false;
}
if (iter == type_map.end() && in_dtype != device_dtype) {
flag_in += 1;
}
}

for (size_t output_index = 0; output_index < kernel_build_info->GetOutputNum(); ++output_index) {
auto in_dtype = AnfAlgo::GetOutputInferDataType(cnode, output_index);
auto device_dtype = kernel_build_info->GetOutputDeviceType(output_index);
if (device_dtype == kNumberTypeFloat || device_dtype == kNumberTypeFloat32) {
device_dtype = kNumberTypeFloat32;
}
auto iter = type_map.find(in_dtype);
if (iter != type_map.end() && iter->second != device_dtype && in_dtype != device_dtype) {
return false;
}
if (iter == type_map.end() && in_dtype != device_dtype) {
flag_out += 1;
}
}
if (flag_in == kernel_build_info->GetInputNum() || flag_out == kernel_build_info->GetOutputNum()) {
return false;
}
return true;
}

std::vector<std::shared_ptr<kernel::KernelBuildInfo>> FilterRaisedOrReducePrecisionMatchedKernelInfo( std::vector<std::shared_ptr<kernel::KernelBuildInfo>> FilterRaisedOrReducePrecisionMatchedKernelInfo(
const CNodePtr &cnode, const std::vector<std::shared_ptr<kernel::KernelBuildInfo>> &kernel_info_list, const CNodePtr &cnode, const std::vector<std::shared_ptr<kernel::KernelBuildInfo>> &kernel_info_list,
bool *precision_reduce) { bool *precision_reduce) {
std::vector<std::shared_ptr<kernel::KernelBuildInfo>> filtered_kernel_info_list; std::vector<std::shared_ptr<kernel::KernelBuildInfo>> filtered_kernel_info_list;
std::map<size_t, std::vector<int>> kernel_match_datatype_idx;
std::map<size_t, std::vector<TypeId>> kernel_support_datatype;
std::vector<int> node_mix_precision_datatype_index;
std::vector<TypeId> node_mix_precision_datatype;
const std::map<TypeId, TypeId> raise_map = {{kNumberTypeFloat16, kNumberTypeFloat32}};
const std::map<TypeId, TypeId> reduce_map = {{kNumberTypeInt64, kNumberTypeInt32},
{kNumberTypeFloat, kNumberTypeFloat16},
{kNumberTypeFloat32, kNumberTypeFloat16}};
// raise precision
for (size_t info_index = 0; info_index < kernel_info_list.size(); ++info_index) { for (size_t info_index = 0; info_index < kernel_info_list.size(); ++info_index) {
std::vector<int> support_indexes;
std::vector<TypeId> support_datatypes;
MS_EXCEPTION_IF_NULL(kernel_info_list[info_index]); MS_EXCEPTION_IF_NULL(kernel_info_list[info_index]);
AddNodeAndKernelDataType(cnode, *kernel_info_list[info_index], &support_indexes, &node_mix_precision_datatype,
&support_datatypes, &node_mix_precision_datatype_index);
kernel_match_datatype_idx[info_index] = support_indexes;
kernel_support_datatype[info_index] = support_datatypes;
if (TagRaiseReduce(kernel_info_list[info_index], cnode, raise_map)) {
filtered_kernel_info_list.push_back(kernel_info_list[info_index]);
}
}

if (!filtered_kernel_info_list.empty()) {
*precision_reduce = false;
return filtered_kernel_info_list;
}

// reduce precision
auto context_ptr = MsContext::GetInstance();
MS_EXCEPTION_IF_NULL(context_ptr);
if (context_ptr->get_param<bool>(MS_CTX_ENABLE_REDUCE_PRECISION)) {
for (size_t info_index = 0; info_index < kernel_info_list.size(); ++info_index) {
MS_EXCEPTION_IF_NULL(kernel_info_list[info_index]);
if (TagRaiseReduce(kernel_info_list[info_index], cnode, reduce_map)) {
filtered_kernel_info_list.push_back(kernel_info_list[info_index]);
}
}
}
if (!filtered_kernel_info_list.empty()) {
*precision_reduce = true;
} }
PrecisionReduce(node_mix_precision_datatype_index, node_mix_precision_datatype, kernel_support_datatype,
&kernel_match_datatype_idx, precision_reduce);
std::transform(
kernel_match_datatype_idx.begin(), kernel_match_datatype_idx.end(), std::back_inserter(filtered_kernel_info_list),
[&](const std::pair<size_t, std::vector<int>> &matched_idx) -> std::shared_ptr<kernel::KernelBuildInfo> {
return kernel_info_list[matched_idx.first];
});
return filtered_kernel_info_list; return filtered_kernel_info_list;
} }
} // namespace } // namespace


Write Preview
Loading…
Cancel
Save