Merge pull request !7066 from 张学同/to_mergetags/v1.1.0
| @@ -92,7 +92,7 @@ int ActivationFp16CPUKernel::DoActivation(int task_id) { | |||||
| return error_code; | return error_code; | ||||
| } | } | ||||
| int ActivationRun(void *cdata, int task_id) { | |||||
| int ActivationFp16Run(void *cdata, int task_id) { | |||||
| auto activation_kernel = reinterpret_cast<ActivationFp16CPUKernel *>(cdata); | auto activation_kernel = reinterpret_cast<ActivationFp16CPUKernel *>(cdata); | ||||
| auto error_code = activation_kernel->DoActivation(task_id); | auto error_code = activation_kernel->DoActivation(task_id); | ||||
| if (error_code != RET_OK) { | if (error_code != RET_OK) { | ||||
| @@ -115,7 +115,7 @@ int ActivationFp16CPUKernel::Run() { | |||||
| return ret; | return ret; | ||||
| } | } | ||||
| int error_code = ParallelLaunch(this->context_->thread_pool_, ActivationRun, this, thread_count_); | |||||
| int error_code = ParallelLaunch(this->context_->thread_pool_, ActivationFp16Run, this, thread_count_); | |||||
| if (error_code != RET_OK) { | if (error_code != RET_OK) { | ||||
| MS_LOG(ERROR) << "Activation function error error_code[" << error_code << "]"; | MS_LOG(ERROR) << "Activation function error error_code[" << error_code << "]"; | ||||
| FreeTmpBuffer(); | FreeTmpBuffer(); | ||||
| @@ -30,7 +30,7 @@ using mindspore::schema::PrimitiveType_Cast; | |||||
| namespace mindspore::kernel { | namespace mindspore::kernel { | ||||
| namespace { | namespace { | ||||
| int CastRun(void *cdata, int task_id) { | |||||
| int CastFp16Run(void *cdata, int task_id) { | |||||
| if (cdata == nullptr) { | if (cdata == nullptr) { | ||||
| MS_LOG(ERROR) << "input cdata is nullptr!"; | MS_LOG(ERROR) << "input cdata is nullptr!"; | ||||
| return RET_ERROR; | return RET_ERROR; | ||||
| @@ -91,7 +91,7 @@ int CastFp16CPUKernel::Run() { | |||||
| if (data_num_ == 0) { | if (data_num_ == 0) { | ||||
| return RET_OK; | return RET_OK; | ||||
| } | } | ||||
| return ParallelLaunch(this->context_->thread_pool_, CastRun, this, op_parameter_->thread_num_); | |||||
| return ParallelLaunch(this->context_->thread_pool_, CastFp16Run, this, op_parameter_->thread_num_); | |||||
| } | } | ||||
| kernel::LiteKernel *CpuCastFp16KernelCreator(const std::vector<lite::Tensor *> &inputs, | kernel::LiteKernel *CpuCastFp16KernelCreator(const std::vector<lite::Tensor *> &inputs, | ||||
| @@ -50,7 +50,7 @@ int CropFp16CPUKernel::DoExecute(int task_id) { | |||||
| return RET_OK; | return RET_OK; | ||||
| } | } | ||||
| static int CropRun(void *cdata, int task_id) { | |||||
| static int CropFp16Run(void *cdata, int task_id) { | |||||
| auto g_kernel = reinterpret_cast<CropFp16CPUKernel *>(cdata); | auto g_kernel = reinterpret_cast<CropFp16CPUKernel *>(cdata); | ||||
| auto ret = g_kernel->DoExecute(task_id); | auto ret = g_kernel->DoExecute(task_id); | ||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| @@ -79,7 +79,7 @@ int CropFp16CPUKernel::Run() { | |||||
| return RET_ERROR; | return RET_ERROR; | ||||
| } | } | ||||
| ret = ParallelLaunch(this->context_->thread_pool_, CropRun, this, thread_count_); | |||||
| ret = ParallelLaunch(this->context_->thread_pool_, CropFp16Run, this, thread_count_); | |||||
| if (out_tensors_.at(kOutputIndex)->data_type() == kNumberTypeFloat32) { | if (out_tensors_.at(kOutputIndex)->data_type() == kNumberTypeFloat32) { | ||||
| Float16ToFloat32(output_ptr_, reinterpret_cast<float *>(out_tensors_.at(kOutputIndex)->MutableData()), | Float16ToFloat32(output_ptr_, reinterpret_cast<float *>(out_tensors_.at(kOutputIndex)->MutableData()), | ||||
| @@ -65,7 +65,7 @@ int ReduceFp16CPUKernel::CallReduceUnit(int task_id) { | |||||
| return ret; | return ret; | ||||
| } | } | ||||
| static int ReduceImpl(void *cdata, int task_id) { | |||||
| static int ReduceFp16Impl(void *cdata, int task_id) { | |||||
| auto reduce = reinterpret_cast<ReduceFp16CPUKernel *>(cdata); | auto reduce = reinterpret_cast<ReduceFp16CPUKernel *>(cdata); | ||||
| auto error_code = reduce->CallReduceUnit(task_id); | auto error_code = reduce->CallReduceUnit(task_id); | ||||
| if (error_code != RET_OK) { | if (error_code != RET_OK) { | ||||
| @@ -102,7 +102,7 @@ int ReduceFp16CPUKernel::Run() { | |||||
| outer_size_ = outer_sizes_[i]; | outer_size_ = outer_sizes_[i]; | ||||
| inner_size_ = inner_sizes_[i]; | inner_size_ = inner_sizes_[i]; | ||||
| axis_size_ = axis_sizes_[i]; | axis_size_ = axis_sizes_[i]; | ||||
| auto error_code = ParallelLaunch(this->context_->thread_pool_, ReduceImpl, this, context_->thread_num_); | |||||
| auto error_code = ParallelLaunch(this->context_->thread_pool_, ReduceFp16Impl, this, context_->thread_num_); | |||||
| if (error_code != RET_OK) { | if (error_code != RET_OK) { | ||||
| FreeTmpBuffer(); | FreeTmpBuffer(); | ||||
| MS_LOG(ERROR) << "Reduce run error, error_code[" << error_code << "]"; | MS_LOG(ERROR) << "Reduce run error, error_code[" << error_code << "]"; | ||||
| @@ -92,7 +92,7 @@ int ScaleFp16CPUKernel::Scale(int task_id) { | |||||
| return RET_OK; | return RET_OK; | ||||
| } | } | ||||
| int ScaleRun(void *cdata, int task_id) { | |||||
| int ScaleFp16Run(void *cdata, int task_id) { | |||||
| auto scale = reinterpret_cast<ScaleFp16CPUKernel *>(cdata); | auto scale = reinterpret_cast<ScaleFp16CPUKernel *>(cdata); | ||||
| auto ret = scale->Scale(task_id); | auto ret = scale->Scale(task_id); | ||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| @@ -121,7 +121,7 @@ int ScaleFp16CPUKernel::Run() { | |||||
| return ret; | return ret; | ||||
| } | } | ||||
| ret = ParallelLaunch(this->context_->thread_pool_, ScaleRun, this, op_parameter_->thread_num_); | |||||
| ret = ParallelLaunch(this->context_->thread_pool_, ScaleFp16Run, this, op_parameter_->thread_num_); | |||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| MS_LOG(ERROR) << "Scale error error_code[" << ret << "]"; | MS_LOG(ERROR) << "Scale error error_code[" << ret << "]"; | ||||
| return RET_ERROR; | return RET_ERROR; | ||||
| @@ -65,7 +65,7 @@ int SplitFp16CPUKernel::Split(int task_id) { | |||||
| return RET_OK; | return RET_OK; | ||||
| } | } | ||||
| static int SplitRun(void *cdata, int task_id) { | |||||
| static int SplitFp16Run(void *cdata, int task_id) { | |||||
| auto g_kernel = reinterpret_cast<SplitFp16CPUKernel *>(cdata); | auto g_kernel = reinterpret_cast<SplitFp16CPUKernel *>(cdata); | ||||
| auto ret = g_kernel->Split(task_id); | auto ret = g_kernel->Split(task_id); | ||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| @@ -94,7 +94,7 @@ int SplitFp16CPUKernel::Run() { | |||||
| return RET_ERROR; | return RET_ERROR; | ||||
| } | } | ||||
| } | } | ||||
| ret = ParallelLaunch(this->context_->thread_pool_, SplitRun, this, thread_n_num_); | |||||
| ret = ParallelLaunch(this->context_->thread_pool_, SplitFp16Run, this, thread_n_num_); | |||||
| for (int i = 0; i < param->num_split_; i++) { | for (int i = 0; i < param->num_split_; i++) { | ||||
| if (out_tensors_.at(i)->data_type() == kNumberTypeFloat32) { | if (out_tensors_.at(i)->data_type() == kNumberTypeFloat32) { | ||||
| Float16ToFloat32(output_ptr_[i], reinterpret_cast<float *>(out_tensors_.at(i)->MutableData()), | Float16ToFloat32(output_ptr_[i], reinterpret_cast<float *>(out_tensors_.at(i)->MutableData()), | ||||
| @@ -117,7 +117,7 @@ int TransposeFp16CPUKernel::TransposeParallel(int task_id) { | |||||
| return RET_OK; | return RET_OK; | ||||
| } | } | ||||
| static int TransposeRun(void *cdata, int task_id) { | |||||
| static int TransposeFp16Run(void *cdata, int task_id) { | |||||
| auto g_kernel = reinterpret_cast<TransposeFp16CPUKernel *>(cdata); | auto g_kernel = reinterpret_cast<TransposeFp16CPUKernel *>(cdata); | ||||
| auto ret = g_kernel->TransposeParallel(task_id); | auto ret = g_kernel->TransposeParallel(task_id); | ||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| @@ -162,7 +162,7 @@ int TransposeFp16CPUKernel::Run() { | |||||
| in_shape_ = const_cast<int *>(in_tensor->shape().data()); | in_shape_ = const_cast<int *>(in_tensor->shape().data()); | ||||
| out_shape_ = const_cast<int *>(out_tensor->shape().data()); | out_shape_ = const_cast<int *>(out_tensor->shape().data()); | ||||
| ret = ParallelLaunch(this->context_->thread_pool_, TransposeRun, this, thread_h_num_); | |||||
| ret = ParallelLaunch(this->context_->thread_pool_, TransposeFp16Run, this, thread_h_num_); | |||||
| if (ret != RET_OK) { | if (ret != RET_OK) { | ||||
| MS_LOG(ERROR) << "Tranpose error error_code[" << ret << "]"; | MS_LOG(ERROR) << "Tranpose error error_code[" << ret << "]"; | ||||
| FreeFp16Buffer(); | FreeFp16Buffer(); | ||||