/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/ |
D | scatter_arithmetic_cpu_kernel.cc | 54 inner_size_ = 1; in InitKernel() 60 inner_size_ *= input_shape[i]; in InitKernel() 62 input_size_ = input_shape[0] * inner_size_; in InitKernel() 93 auto base_index_updates = i * inner_size_; in ScatterAdd() 94 auto base_index_input = indices[i] * inner_size_; in ScatterAdd() 95 for (size_t j = 0; j < inner_size_; j++) { in ScatterAdd() 104 auto base_index_updates = i * inner_size_; in ScatterSub() 105 auto base_index_input = indices[i] * inner_size_; in ScatterSub() 106 for (size_t j = 0; j < inner_size_; j++) { in ScatterSub() 115 auto base_index_updates = i * inner_size_; in ScatterMul() [all …]
|
D | topk_cpu_kernel.cc | 36 if (inputs[0]->size != outer_size_ * inner_size_ * sizeof(T)) { in LaunchKernel() 50 size_t k_num = IntToSize(std::min<int>(inner_size_, k)); in LaunchKernel() 63 size_t *idx = workspace + i * inner_size_; in LaunchKernel() 64 auto base_input = i * inner_size_; in LaunchKernel() 65 std::iota(idx, idx + inner_size_, base_input); in LaunchKernel() 69 const size_t threshold = inner_size_ * fraction; in LaunchKernel() 72 std::stable_sort(idx, idx + inner_size_, comparator); in LaunchKernel() 74 std::nth_element(idx, idx + SizeToLong(k_num), idx + inner_size_, comparator); in LaunchKernel() 78 std::nth_element(idx, idx + SizeToLong(k_num), idx + inner_size_, comparator); in LaunchKernel() 101 inner_size_ = x_shape_[x_shape_.size() - 1]; in InitKernel() [all …]
|
D | cpu_kernel.h | 211 axis_offset_ = outer_index * axis_size_ * inner_size_ + inner_index; in SetOffset() 213 inline size_t GetPos(size_t i) const { return axis_offset_ + i * inner_size_; } in GetPos() 214 inline size_t RevertPos(size_t i) const { return (i - axis_offset_) / inner_size_; } in RevertPos() 218 inline size_t InnerSize() const { return inner_size_; } in InnerSize() 223 size_t inner_size_{0};
|
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/base/ |
D | strided_slice.cc | 119 …uint8_t *cur_in_ptr = input_ptr_ + (caled_num * in_shape[split_axis_] + begin_index) * inner_size_; in FastRunImpl() 120 uint8_t *cur_out_ptr = output_ptr_ + caled_num * out_shape[split_axis_] * inner_size_; in FastRunImpl() 128 …in_ptr, cur_out_ptr, out_shape[split_axis_], param_->strides_[split_axis_], cur_outer, inner_size_, in FastRunImpl() 129 in_shape[split_axis_] * inner_size_); in FastRunImpl() 132 …*cur_in_ptr = input_ptr_ + (caled_num * param_->strides_[split_axis_] + begin_index) * inner_size_; in FastRunImpl() 133 uint8_t *cur_out_ptr = output_ptr_ + caled_num * inner_size_; in FastRunImpl() 141 …FastStride(cur_in_ptr, cur_out_ptr, cal_axis_num, param_->strides_[split_axis_], 1, inner_size_, 0… in FastRunImpl() 163 inner_size_ = inner_ * sizeof(int8_t); in FastRun() 166 inner_size_ = inner_ * sizeof(float); in FastRun() 169 inner_size_ = inner_ * sizeof(int16_t); in FastRun() [all …]
|
D | one_hot_base.cc | 85 inner_size_ = indices->ElementsNum() / outer_size_; in ReSize() 173 one_hot_param->inner_size_ = inner_size_; in InitParamsAndOnOffValue()
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/arrays/ |
D | topk_gpu_kernel.h | 33 …: sorted_(false), is_null_input_(false), outer_size_(1), inner_size_(1), k_(1), input_shape_size_(… in TopKGpuKernel() 64 …Cast(outer_size_ * inner_size_, input_addr, casted_float32_input, reinterpret_cast<cudaStream_t>(s… in Launch() 67 …FastTopK(outer_size_, inner_size_, casted_float32_input, k_cut, casted_float32_top_k_output, indic… in Launch() 79 FastTopK(outer_size_, inner_size_, input_addr, k_cut, output_addr, indices, init_k, in Launch() 99 inner_size_ = input_shapes[input_shapes.size() - 1]; in Init() 110 input_size_list_.push_back(outer_size_ * inner_size_ * sizeof(T)); in InitSizeLists() 117 workspace_size_list_.push_back(outer_size_ * inner_size_ * sizeof(float)); in InitSizeLists() 126 size_t inner_size_; variable
|
D | scatter_functor_gpu_kernel.h | 53 ScatterFunc(scatter_functor_type_, inner_size_, indices_size_, indices, updates, input, in Launch() 83 inner_size_ = 1; in Init() 85 inner_size_ *= input_shape[i]; in Init() 87 input_size_ = input_shape[0] * inner_size_; in Init() 93 updates_size_ = indices_size_ * inner_size_; in Init() 100 inner_size_ = 0; in ResetResource() 119 size_t inner_size_; variable
|
D | argmax_gpu_kernel.h | 35 inner_size_(0), in ArgmaxGpuKernel() 52 …CalArgmax(input, bound_, outer_size_, inner_size_, output, reinterpret_cast<cudaStream_t>(stream_p… in Launch() 90 inner_size_ = 1; in Init() 92 inner_size_ *= shape[i]; in Init() 113 size_t inner_size_; variable
|
D | unsorted_segment_min_gpu_kernel.h | 46 …CalUnsortedSegmentMin(input_addr, indices_addr, num_segments_, outer_size_, inner_size_, output_ad… in Launch() 91 inner_size_ = 1; in Init() 93 inner_size_ *= input_shapes[i]; in Init() 102 inner_size_ = 1; in ResetResource() 122 size_t inner_size_; variable
|
D | in_top_k_gpu_kernel.h | 59 if (k_ >= static_cast<int64_t>(inner_size_)) { in Launch() 80 …FastTopK(outer_size_, inner_size_, casted_float32_input, static_cast<int32_t>(k_), top_k_output_de… in Launch() 87 …FastTopK(outer_size_, inner_size_, predictions_device, static_cast<int32_t>(k_), top_k_output_devi… in Launch() 130 inner_size_ = input_shape_[1]; in Init() 151 inner_size_ = 0; in ResetResource() 187 size_t inner_size_; variable
|
D | unsorted_segment_max_gpu_kernel.h | 51 …CalUnsortedSegmentMax(input_addr, indices_addr, num_segments_, outer_size_, inner_size_, output_ad… in Launch() 97 inner_size_ = 1; in Init() 99 inner_size_ *= input_shapes[i]; in Init() 108 inner_size_ = 1; in ResetResource() 128 size_t inner_size_; variable
|
D | sort_gpu_kernel.h | 92 …FastTopK(outer_size_, inner_size_, intermediate_input_device, static_cast<int32_t>(input_shape_[ax… in Launch() 171 inner_size_ = input_shape_[axis_]; in Init() 172 outer_size_ = input_size_ / inner_size_; in Init() 196 inner_size_ = 0; in ResetResource() 235 size_t inner_size_; variable
|
/third_party/mindspore/mindspore/lite/micro/coder/opcoders/base/ |
D | strided_slice_base_coder.cc | 117 << " + " << (caled_num * in_shape[split_axis_] + begin_index) * inner_size_ << ";\n"; in DoFastCode() 120 << " + " << caled_num * out_shape[split_axis_] * inner_size_ << ";\n"; in DoFastCode() 122 strided_slice_parameter_->strides_[split_axis_], cur_outer, inner_size_, in DoFastCode() 123 in_shape.at(split_axis_) * inner_size_); in DoFastCode() 134 …(caled_num * strided_slice_parameter_->strides_[split_axis_] + begin_index) * inner_size_ << ";\n"; in DoFastCode() 137 << " + " << caled_num * inner_size_ << ";\n"; in DoFastCode() 139 strided_slice_parameter_->strides_[split_axis_], 1, inner_size_, 0); in DoFastCode() 155 inner_size_ = GetInnerSize(input_tensor_->data_type(), inner_); in DoCode()
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/math/ |
D | index_add_gpu_kernel.h | 37 inner_size_(0), in IndexAddGpuKernel() 55 … CalIndexAdd(dst, index, src, outer_size_, src_axis_size_, dst_axis_size_, inner_size_, use_lock_, in Launch() 89 inner_size_ = 1; in Init() 91 inner_size_ *= src_shape[i]; in Init() 133 size_t inner_size_; variable
|
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp32_grad/ |
D | softmax_grad.cc | 50 inner_size_ = 1; in Init() 52 inner_size_ *= in_shape.at(i); in Init() 54 set_workspace_size(inner_size_ * (1 + in_shape.at(axis)) * sizeof(float)); in Init() 65 float *sum_mul_ = sum_data_ + inner_size_; in Execute()
|
D | sparse_softmax_cross_entropy_with_logits.cc | 100 SoftMaxP1(ins, losses, sum_data, task_id * stride, count, length, inner_size_); in Execute() 103 SoftMaxP2(ins, losses, sum_data, task_id * stride, count, length, inner_size_); in Execute() 147 inner_size_ = inner_size; in Run()
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp32/ |
D | instance_norm_fp32.c | 32 const float *src_b = src_data + b * param->channel_ * param->inner_size_; in InstanceNorm() 33 float *dst_b = dst_data + b * param->channel_ * param->inner_size_; in InstanceNorm() 35 const float *src = src_b + c * param->inner_size_; in InstanceNorm() 36 float *dst = dst_b + c * param->inner_size_; in InstanceNorm() 42 for (; index <= param->inner_size_ - C8NUM; index += C8NUM) { in InstanceNorm() 55 for (; index <= param->inner_size_ - C4NUM; index += C4NUM) { in InstanceNorm() 76 for (; index < param->inner_size_; index++) { in InstanceNorm() 81 mean /= (float)param->inner_size_; in InstanceNorm() 82 squ_m /= (float)param->inner_size_; in InstanceNorm() 89 for (; index <= param->inner_size_ - C8NUM; index += C8NUM) { in InstanceNorm() [all …]
|
D | scale_fp32.c | 97 if (scale_param->inner_size_ == 1) { in DoScale() 101 scale_param->inner_size_); in DoScale() 199 if (scale_param->inner_size_ == 1) { in DoScaleRelu() 203 scale_param->inner_size_); in DoScaleRelu() 305 if (scale_param->inner_size_ == 1) { in DoScaleRelu6() 309 scale_param->inner_size_); in DoScaleRelu6()
|
/third_party/mindspore/mindspore/lite/micro/coder/opcoders/nnacl/fp32/ |
D | reduce_fp32_coder.cc | 104 inner_size_ = inner_sizes_.at(i); in DoCode() 107 …code.CodeFunction(reduce, outer_size_, inner_size_, axis_size_, src_addr, dst_addr, 0, thread_num_… in DoCode() 109 …code.CodeFunction(int_reduce, outer_size_, inner_size_, axis_size_, src_addr, dst_addr, 0, thread_… in DoCode()
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp16/ |
D | scale_fp16.c | 74 if (scale_param->inner_size_ == 1) { in DoScaleFp16() 78 scale_param->inner_size_); in DoScaleFp16() 146 if (scale_param->inner_size_ == 1) { in Fp16DoScaleRelu() 150 scale_param->inner_size_); in Fp16DoScaleRelu() 220 if (scale_param->inner_size_ == 1) { in DoScaleRelu6Fp16() 224 scale_param->inner_size_); in DoScaleRelu6Fp16()
|
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp32/ |
D | reduce_fp32.cc | 66 reducer_(outer_size_, inner_size_, axis_size_, static_cast<const float *>(src_data_), in CallReduceUnit() 73 bool_reducer_(outer_size_, inner_size_, axis_size_, static_cast<const bool *>(src_data_), in CallReduceUnit() 80 int_reducer_(outer_size_, inner_size_, axis_size_, static_cast<const int *>(src_data_), in CallReduceUnit() 123 inner_size_ = inner_sizes_.at(i); in Run()
|
D | instance_norm_fp32.cc | 42 param_->inner_size_ = in_tensor->Height() * in_tensor->Width(); in ReSize() 91 …PackNHWCToNC4HW4Fp32(src_data_, tmp_src_data_, param_->batch_, param_->inner_size_, param_->channe… in Run()
|
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp16/ |
D | reduce_fp16.cc | 68 return reducer_(outer_size_, inner_size_, axis_size_, fp16_src_data_, fp16_dst_data_, task_id, in CallReduceUnit() 95 inner_size_ = inner_sizes_.at(i); in Run() 110 inner_size_ = inner_sizes_.back(); in Run()
|
/third_party/mindspore/mindspore/lite/micro/coder/opcoders/nnacl/int8/ |
D | reduce_int8_coder.cc | 225 inner_size_ = inner_sizes_.at(i); in DoCode() 228 …code.CodeFunction(reducer_, outer_size_, inner_size_, axis_size_, begin_src_data_src, dst_addr, pt… in DoCode() 231 …code.CodeFunction(last_reducer_, outer_size_, inner_size_, axis_size_, begin_src_data_src, dst_add… in DoCode()
|
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/ |
D | instance_norm_parameter.h | 29 int inner_size_; member
|