Home
last modified time | relevance | path

Searched refs:output_index (Results 1 – 25 of 106) sorted by relevance

12345

/third_party/mindspore/mindspore/ccsrc/runtime/framework/actor/
Doutput_actor.cc23 TensorPtr CreateOutputTensor(const AnfNodePtr &output_node, size_t output_index, size_t output_posi… in CreateOutputTensor() argument
26 << ", output index: " << output_index << ", output position: " << output_position; in CreateOutputTensor()
30 auto type_id = AnfAlgo::GetOutputInferDataType(output_node, output_index); in CreateOutputTensor()
32 auto shape = AnfAlgo::GetOutputInferShape(output_node, output_index); in CreateOutputTensor()
35 tensor->set_padding_type(AnfAlgo::GetOutputReshapeType(output_node, output_index)); in CreateOutputTensor()
38 const auto &device_tensor = AnfAlgo::GetMutableOutputAddr(output_node, output_index, false); in CreateOutputTensor()
86 auto output_index = output_nodes_[i].second; in UpdateOutputDeviceAddress() local
88 const auto &device_tensor = AnfAlgo::GetMutableOutputAddr(output_node, output_index, false); in UpdateOutputDeviceAddress()
100 AnfAlgo::SetOutputAddr(new_device_tensor, output_index, output_node.get()); in UpdateOutputDeviceAddress()
108 void OutputActor::CollectOutput(const AnfNodePtr &output_node, size_t output_index, size_t output_p… in CollectOutput() argument
[all …]
/third_party/mindspore/mindspore/ccsrc/runtime/device/ascend/executor/
Daicpu_ext_info_handle.cc167 bool AicpuExtInfoHandler::UpdateOutputShapeAndType(uint32_t output_index, const NotNull<AnfNodePtr>… in UpdateOutputShapeAndType() argument
168 if (output_index >= output_num_) { in UpdateOutputShapeAndType()
169 …MS_LOG(ERROR) << "output_index:" << output_index << " >= output_num_:" << output_num_ << ", node: … in UpdateOutputShapeAndType()
173 auto shape = AnfAlgo::GetOutputDeviceShape(anf_node, output_index); in UpdateOutputShapeAndType()
174 auto max_shape = AnfAlgo::GetOutputMaxShape(anf_node, output_index); in UpdateOutputShapeAndType()
190 if (output_index >= output_shape_and_type_.size()) { in UpdateOutputShapeAndType()
191 MS_LOG(ERROR) << "Invalid output_index: " << output_index in UpdateOutputShapeAndType()
198 return UpdateShapeAndType(tmp_shape, NOT_NULL(output_shape_and_type_[output_index])); in UpdateOutputShapeAndType()
201 bool AicpuExtInfoHandler::GetOutputShapeAndType(uint32_t output_index, NotNull<std::vector<int64_t>… in GetOutputShapeAndType() argument
203 MS_LOG(DEBUG) << "Get " << node_name_ << " Output:" << output_index << " Shape And Type"; in GetOutputShapeAndType()
[all …]
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/
Dgpu_kernel_factory.cc100 for (size_t output_index = 0; output_index < kernel_info->GetOutputNum(); output_index++) { in ReducePrecision() local
101 if (kernel_info->GetOutputDeviceType(output_index) == kNumberTypeInt64 && in ReducePrecision()
102 …(iter->second)[attr_index].first.GetOutputAttr(output_index % attr_size).first == kNumberTypeInt32… in ReducePrecision()
103 builder->SetOutputDeviceType(kNumberTypeInt32, output_index); in ReducePrecision()
104 …G(WARNING) << "Kernel [" << kernel_name << "] does not support int64, cast output " << output_index in ReducePrecision()
163 for (size_t output_index = 0; output_index < kernel_info->GetOutputNum(); output_index++) { in GpuKernelAttrCheck() local
164 if (kernel_info->GetOutputDeviceType(output_index) != in GpuKernelAttrCheck()
165 (iter->second)[attr_index].first.GetOutputAttr(output_index % attr_size).first) { in GpuKernelAttrCheck()
/third_party/mindspore/mindspore/ccsrc/backend/optimizer/ascend/format_type/
Ddeal_ref_and_split_unsupported_transdata.cc65 … const CNodePtr &cnode, const size_t output_index, in AddRefNodePairToKernelGraph() argument
70 session::AnfWithOutIndex final_pair = std::make_pair(cnode, output_index); in AddRefNodePairToKernelGraph()
100 … const CNodePtr &cnode, size_t output_index, in AddAdditionalToRefOutput() argument
105 size_t final_index = output_index; in AddAdditionalToRefOutput()
116 auto cur_format = AnfAlgo::GetOutputFormat(cnode, output_index); in AddAdditionalToRefOutput()
117 auto cur_type = AnfAlgo::GetOutputDeviceDataType(cnode, output_index); in AddAdditionalToRefOutput()
118 auto cur_shape = AnfAlgo::GetOutputInferShape(cnode, output_index); in AddAdditionalToRefOutput()
119 auto detail_shape = AnfAlgo::GetOutputDetailShape(cnode, output_index); in AddAdditionalToRefOutput()
144 AddRefNodePairToKernelGraph(func_graph, cnode, output_index, input_index); in AddAdditionalToRefOutput()
188 for (size_t output_index = 0; output_index < output_num; ++output_index) { in DealRefForMultipleOutput() local
[all …]
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/
Dkernel_build_info.cc30 std::string KernelBuildInfo::GetOutputFormat(size_t output_index) const { in GetOutputFormat()
31 if (output_index >= outputs_format_.size()) { in GetOutputFormat()
32 MS_LOG(ERROR) << "The index [" << output_index << "] is exceed the number of output"; in GetOutputFormat()
35 return outputs_format_[output_index]; in GetOutputFormat()
46 TypeId KernelBuildInfo::GetOutputDeviceType(size_t output_index) const { in GetOutputDeviceType()
47 if (output_index >= outputs_device_type_.size()) { in GetOutputDeviceType()
48 MS_LOG(ERROR) << "The index [" << output_index << "] is exceed the number of output"; in GetOutputDeviceType()
51 return outputs_device_type_[output_index]; in GetOutputDeviceType()
90 std::string KernelBuildInfo::GetOutputReshapeType(size_t output_index) const { in GetOutputReshapeType()
94 if (output_index >= output_reshape_type_.size()) { in GetOutputReshapeType()
[all …]
/third_party/mindspore/mindspore/ccsrc/backend/optimizer/pass/
Dreplace_node_by_proxy.cc39 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GenerateKernelBuildInfo() local
40 outputs_device_format.push_back(AnfAlgo::GetOutputFormat(cnode, output_index)); in GenerateKernelBuildInfo()
41 outputs_device_type.push_back(AnfAlgo::GetOutputDeviceDataType(cnode, output_index)); in GenerateKernelBuildInfo()
42 outputs_shape.push_back(AnfAlgo::GetOutputInferShape(cnode, output_index)); in GenerateKernelBuildInfo()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/
Drcwm_small_impl.cu24 <<<1, BLOCK, 0, stream>>>(seedc, input_size, input, output_mask, output_index, k); \
43 __global__ void L2Rcwm(int seedc, int input_size, const K *input, K *output_mask, S *output_index, … in L2Rcwm() argument
121 output_index[i] = shared_V[i]; in L2Rcwm()
126 void RCWMScaleK(int seedc, int input_size, K *input, int k, S *output_index, K *output_mask, cudaSt… in RCWMScaleK() argument
146 void CalRandomChoiceWithMaskSmall(int input_size, int seedc, int count, K *input, S *output_index, … in CalRandomChoiceWithMaskSmall() argument
148 RCWMScaleK<T, S, K>(seedc, input_size, input, count, output_index, output_mask, stream); in CalRandomChoiceWithMaskSmall()
152 … int *output_index, bool *output_mask, cudaStream_t stream);
Drandom_choice_with_mask_impl.cu103 … const int d3, const int d4, const int d5, const T *input, S *output_index) { in Reshape2Index() argument
116 output_index[index_pos++] = 0; in Reshape2Index()
120 output_index[index_pos++] = pos_array[i]; in Reshape2Index()
184 …l__ void MoveToOutput(const int input_shape_size, const int count, const T *input, S *output_index, in MoveToOutput() argument
197 output_index[pos] = index_buff[idx]; in MoveToOutput()
211 output_index[pos] = index_buff[idx]; in MoveToOutput()
220 output_index[pos] = static_cast<S>(0); in MoveToOutput()
231 … const T *input, S *output_index, T *output_mask, S *index_buff, S *mask_buff, S *rank_buff, in CalRandomChoiceWithMask() argument
256 …BLOCKS(count), GET_THREADS, 0, stream>>>(input_shape_size, count, input, output_index, output_mask, in CalRandomChoiceWithMask()
262 … const bool *input, int *output_index, bool *output_mask, int *index_buff,
Dtopk_impl.cu27 …<<<block_num_limit, BLOCK, 0, stream>>>(outer_size, inner_size, input, output, output_index, k_cut…
101 … S *output_index, S k_cut, const T &init_K, const int &outer_id, T *shared_K, in TopKStep() argument
171 output_index[outer_id * k_cut + (*k_prime) + i] = shared_V[i] % inner_size; in TopKStep()
178 …void TopKBlock(int outer_size, int inner_size, const T *input, T *output, S *output_index, S k_cut, in TopKBlock() argument
200 …outer_size, inner_size, input, output, output_index, k_cut, init_K, outer_id, shared_K, shared_V, … in TopKBlock()
207 …pK(const int outer_size, const int inner_size, const T *input, S k_cut, T *output, S *output_index, in FastTopK() argument
225 int *output_index, const half init_K, cudaStream_t stream);
227 int *output_index, const float init_K, cudaStream_t stream);
Dgeneral_reduction_impl.cu48 T *output, S *output_index, bool fp16_flag, T init_K) { in ThreadReduction() argument
73 output_index[outer_id * inner_size + inner_id] = threadV; in ThreadReduction()
79 S *output_index, bool fp16_flag, T init_K) { in WarpReduction() argument
119 output_index[outer_id * inner_size + inner_id] = threadV; in WarpReduction()
127 T *output, S *output_index, bool fp16_flag, T init_K) { in Warp4Reduction() argument
208 output_index[outer_id * inner_size + inner_id] = shared_V[groupId * kWarpGroup]; in Warp4Reduction()
216 T *output, S *output_index, bool fp16_flag, T init_K) { in BlockReduction() argument
293 output_index[outer_id * inner_size + inner_id] = threadV; in BlockReduction()
300 S *output_index, cudaStream_t stream) { in GeneralReduction() argument
310 small, outer_size, bound, inner_size, input, output, output_index, fp16_flag, init_K); in GeneralReduction()
[all …]
/third_party/mindspore/mindspore/ccsrc/backend/optimizer/ascend/enhancer/
Dinsert_tensor_move_for_getnext.cc40 for (size_t output_index = 0; output_index < output_num; ++output_index) { in InsertTensorMoveForGetNextOutputs() local
41 auto tuple_get_item = CreatTupleGetItemNode(func_graph, node, output_index); in InsertTensorMoveForGetNextOutputs()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/host/
Dhost_kernel_metadata.cc51 for (size_t output_index = 0; output_index < output_num; ++output_index) { in HostMetadataInfo() local
53 outputs_type.push_back(AnfAlgo::GetOutputInferDataType(kernel_node, output_index)); in HostMetadataInfo()
/third_party/mindspore/mindspore/ccsrc/runtime/device/cpu/
Dkernel_select_cpu.cc65 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GetOutputDtypes() local
67 dtype = AnfAlgo::GetOutputInferDataType(kernel_node, output_index); in GetOutputDtypes()
74 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GetOutputFormat() local
104 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GetOutputFormatsAndDtypes() local
105 output_formats->emplace_back(kernel_attr.GetOutputAttr(output_index).second); in GetOutputFormatsAndDtypes()
106 auto dtype = kernel_attr.GetOutputAttr(output_index).first; in GetOutputFormatsAndDtypes()
/third_party/mindspore/tests/ut/python/parallel/
Dtest_semi_auto_two_subgraphs.py57 def __init__(self, network, output_index): argument
60 self.output_index = output_index
63 predict = self.network(x1)[self.output_index]
91 self.loss_net_w = IthOutputCell(network, output_index=0)
92 self.loss_net_d = IthOutputCell(network, output_index=1)
Dtest_auto_parallel_for_loop_multi_subgraph.py78 def __init__(self, network, output_index): argument
81 self.output_index = output_index
84 predict = self.network(x)[self.output_index]
109 self.loss_net_w = IthOutputCell(network, output_index=0)
110 self.loss_net_d = IthOutputCell(network, output_index=1)
Dtest_auto_parallel_double_subgraphs.py61 def __init__(self, network, output_index): argument
64 self.output_index = output_index
67 predict = self.network(x)[self.output_index]
92 self.loss_net_w = IthOutputCell(network, output_index=0)
93 self.loss_net_d = IthOutputCell(network, output_index=1)
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/aicpu/
Daicpu_kernel_build.cc231 for (size_t output_index = 0; output_index < output_num; output_index++) { in SetNodeOutputs() local
234 std::vector<size_t> output_shape = AnfAlgo::GetOutputDeviceShape(anf_node, output_index); in SetNodeOutputs()
242 TypeId output_type = AnfAlgo::GetOutputDeviceDataType(anf_node, output_index); in SetNodeOutputs()
343 for (size_t output_index = 0; output_index < output_num; output_index++) { in SetExtInfoOutputShapeType() local
344 std::vector<size_t> output_shape = AnfAlgo::GetOutputDeviceShape(anf_node, output_index); in SetExtInfoOutputShapeType()
345 TypeId output_type = AnfAlgo::GetOutputDeviceDataType(anf_node, output_index); in SetExtInfoOutputShapeType()
347 outputs[output_index].type = output_data_type; in SetExtInfoOutputShapeType()
351 outputs[output_index].dims[output_shape_index] = SizeToLong(output_shape[output_shape_index]); in SetExtInfoOutputShapeType()
354 outputs[output_index].dims[output_shape_index] = LLONG_MIN; in SetExtInfoOutputShapeType()
Daicpu_kernel_metadata.cc67 for (size_t output_index = 0; output_index < output_num; ++output_index) { in AicpuMetadataInfoForSpecialNodes() local
69 (void)outputs_type.emplace_back(AnfAlgo::GetOutputInferDataType(kernel_node, output_index)); in AicpuMetadataInfoForSpecialNodes()
/third_party/mindspore/mindspore/ccsrc/runtime/device/ascend/
Dkernel_select_ascend.cc68 for (size_t output_index = 0; output_index < kernel_build_info.GetOutputNum(); ++output_index) { in MatchInferOutputDataType() local
69 …if (kernel_build_info.GetOutputDeviceType(output_index) != AnfAlgo::GetOutputInferDataType(cnode, … in MatchInferOutputDataType()
174 for (size_t output_index = 0; output_index < output_num; ++output_index) { in UpdateCurMatchCounts() local
176 if (kernel_build_info.GetOutputDeviceType(output_index) == in UpdateCurMatchCounts()
177 AnfAlgo::GetOutputInferDataType(kernel_node, output_index)) { in UpdateCurMatchCounts()
180 if (kernel_build_info.GetOutputFormat(output_index) == pri_match_format) { in UpdateCurMatchCounts()
271 for (size_t output_index = 0; output_index < kernel_build_info->GetOutputNum(); ++output_index) { in TagRaiseReduce() local
272 auto in_dtype = AnfAlgo::GetOutputInferDataType(cnode, output_index); in TagRaiseReduce()
273 auto device_dtype = kernel_build_info->GetOutputDeviceType(output_index); in TagRaiseReduce()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/
Done_hot_cpu_kernel.cc64 size_t output_index = stride_num * depth_ * stride_ + i % stride_; in Launch() local
68 output[output_index] = on_value; in Launch()
70 output[output_index] = off_value; in Launch()
72 output_index += stride_; in Launch()
/third_party/mindspore/mindspore/ccsrc/fl/server/kernel/
Doptimizer_kernel.h78 for (size_t output_index = 0; output_index < output_num; ++output_index) { in InitServerKernelInputOutputSize() local
79 std::vector<size_t> shape = AnfAlgo::GetOutputInferShape(kernel_node, output_index); in InitServerKernelInputOutputSize()
/third_party/mindspore/mindspore/ccsrc/backend/optimizer/gpu/
Dadd_relu_v2_fusion.cc43 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GenerateKernelBuildInfo() local
44 outputs_type.push_back(AnfAlgo::GetOutputInferDataType(node, output_index)); in GenerateKernelBuildInfo()
Dmatmul_biasadd_fusion.cc43 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GenerateKernelBuildInfo() local
44 outputs_type.push_back(AnfAlgo::GetOutputInferDataType(node, output_index)); in GenerateKernelBuildInfo()
Dadd_relu_grad_v2_fusion.cc43 for (size_t output_index = 0; output_index < output_num; ++output_index) { in GenerateKernelBuildInfo() local
44 outputs_type.push_back(AnfAlgo::GetOutputInferDataType(node, output_index)); in GenerateKernelBuildInfo()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/hccl/
Dhccl_kernel_metadata.cc97 for (size_t output_index = 0; output_index < output_num; ++output_index) { in HcclMetadataInfo() local
98 outputs_format.emplace_back(GetKernelFormat(kernel_node, output_index)); in HcclMetadataInfo()

12345