Home
last modified time | relevance | path

Searched refs:out_shape_ (Results 1 – 25 of 73) sorted by relevance

123

/third_party/mindspore/mindspore/lite/src/runtime/kernel/opencl/kernel/
Dsoftmax.cc67 out_shape_ = GpuTensorInfo(out_tensors_[0]); in Prepare()
69 if (out_shape_.H == 1 && out_shape_.W == 1 && axis_ == 3) { in Prepare()
108 global_size_ = {32, out_shape_.N}; in SetGlobalLocal()
112 global_x = out_shape_.Slice; in SetGlobalLocal()
113 global_y = out_shape_.W; in SetGlobalLocal()
115 global_x = out_shape_.Slice; in SetGlobalLocal()
116 global_y = out_shape_.H; in SetGlobalLocal()
118 global_x = out_shape_.W; in SetGlobalLocal()
119 global_y = out_shape_.H; in SetGlobalLocal()
124 global_size_ = {global_x, global_y, out_shape_.N}; in SetGlobalLocal()
[all …]
Dspace_to_depth.cc44 out_shape_ = GpuTensorInfo(out_tensors_[0]); in Prepare()
50 if (in_shape_.C % C4NUM == 0 && out_shape_.C % C4NUM == 0) { in Prepare()
77 cl_int4 cl_out_shape = {static_cast<cl_int>(out_shape_.N), static_cast<cl_int>(out_shape_.H), in SetConstArgs()
78 static_cast<cl_int>(out_shape_.W), static_cast<cl_int>(out_shape_.Slice)}; in SetConstArgs()
94 int co_size = out_shape_.C; in SetConstArgs()
110 global_size_ = {out_shape_.Slice, out_shape_.W, out_shape_.H * out_shape_.N}; in SetGlobalLocal()
Done_hot.cc43 out_shape_ = GpuTensorInfo(out_tensors_[0]); in Prepare()
44 axis_ = out_shape_.AlignAxis(param_->axis_); in Prepare()
98 cl_int4 cl_out_shape = {static_cast<cl_int>(out_shape_.N), static_cast<cl_int>(out_shape_.H), in SetConstArgs()
99 static_cast<cl_int>(out_shape_.W), static_cast<cl_int>(out_shape_.Slice)}; in SetConstArgs()
121 …if (ocl_runtime_->SetKernelArg(kernel_, arg_idx++, static_cast<int>(out_shape_.C)) != CL_SUCCESS) { in SetConstArgs()
133 global_size_ = {out_shape_.Slice, out_shape_.W, out_shape_.H * out_shape_.N}; in SetGlobalLocal()
Dpower.cc67 float unalign_w = static_cast<float>(out_shape_.s[3]); in SetConstArgs()
68 out_shape_.s[3] = UP_DIV(out_shape_.s[3], C4NUM); in SetConstArgs()
72 if (ocl_runtime_->SetKernelArg(kernel_, arg_cn++, out_shape_) != CL_SUCCESS) { in SetConstArgs()
77 if (ocl_runtime_->SetKernelArg(kernel_, arg_cn++, out_shape_) != CL_SUCCESS) { in SetConstArgs()
108 Broadcast2GpuShape(out_shape_.s, output_shape.s, out_tensors_.at(0)->shape().size(), 1); in SetGlobalLocal()
111 uint32_t OH = out_shape_.s[0] * out_shape_.s[1]; in SetGlobalLocal()
112 uint32_t OW = out_shape_.s[2]; in SetGlobalLocal()
113 uint32_t OC = UP_DIV(out_shape_.s[3], C4NUM); in SetGlobalLocal()
Dstack.cc116 Broadcast2GpuShape(out_shape_.s, outshape_tmp.s, out_tensors_[0]->shape().size(), 1); in SetConstArgs()
118 out_shape_.s[3] = UP_DIV(out_shape_.s[3], C4NUM); in SetConstArgs()
123 if (ocl_runtime_->SetKernelArg(kernel_, arg_cn++, out_shape_) != CL_SUCCESS) { in SetConstArgs()
146 OH_ = out_shape_.s[0] * out_shape_.s[1]; in SetGlobalLocal()
147 OW_ = out_shape_.s[2]; in SetGlobalLocal()
148 OC_ = out_shape_.s[3]; in SetGlobalLocal()
150 OH_ = UP_DIV(out_shape_.s[0], C4NUM); in SetGlobalLocal()
151 OW_ = out_shape_.s[3]; in SetGlobalLocal()
153 OH_ = out_shape_.s[0]; in SetGlobalLocal()
154 OW_ = out_shape_.s[1]; in SetGlobalLocal()
Dprelu.cc111 out_shape_.s[3] = UP_DIV(out_shape_.s[3], C4NUM); in SetConstArgs()
112 if (ocl_runtime_->SetKernelArg(kernel_, arg_idx++, out_shape_) != CL_SUCCESS) { in SetConstArgs()
125 OH = out_shape_.s[0] * out_shape_.s[1]; in SetGlobalLocal()
126 OW = out_shape_.s[2]; in SetGlobalLocal()
127 OC = out_shape_.s[3]; in SetGlobalLocal()
141 Broadcast2GpuShape(out_shape_.s, output_shape.s, out_tensors_.at(0)->shape().size(), 1); in Prepare()
Darithmetic.cc81 global_size_ = {out_shape_.width, out_shape_.height}; in SetGlobalLocal()
83 global_size_ = {out_shape_.Slice, out_shape_.W, out_shape_.H * out_shape_.N}; in SetGlobalLocal()
120 …cl_int4 out_shape = {static_cast<int>(out_shape_.N), static_cast<int>(out_shape_.H), static_cast<i… in SetConstArgs()
121 static_cast<int>(out_shape_.Slice)}; in SetConstArgs()
165 out_shape_ = GpuTensorInfo(out_tensors_[0]); in Prepare()
Dconcat.cc131 Broadcast2GpuShape(out_shape_.s, output_shape_.s, out_tensors_[0]->shape().size(), 1); in SetConstArgs()
163 out_shape_.s[3] = UP_DIV(out_shape_.s[3], C4NUM); in SetConstArgs()
164 if (ocl_runtime_->SetKernelArg(kernel_, arg_cn++, out_shape_) != CL_SUCCESS) { in SetConstArgs()
174 OH = out_shape_.s[0] * out_shape_.s[1]; in SetGlobalLocal()
175 OW = out_shape_.s[2]; in SetGlobalLocal()
179 OH = out_shape_.s[0] * out_shape_.s[1]; in SetGlobalLocal()
180 OW = out_shape_.s[2]; in SetGlobalLocal()
181 OC = out_shape_.s[3]; in SetGlobalLocal()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/fp16/
Dcrop_fp16.c44 const int out_batch = para->out_shape_[0]; in Fp16Crop1D()
62 const int out_batch = para->out_shape_[0]; in Fp16Crop2D()
63 const int out_height = para->out_shape_[1]; in Fp16Crop2D()
86 const int out_batch = para->out_shape_[0]; in Fp16Crop3D()
87 const int out_height = para->out_shape_[1]; in Fp16Crop3D()
88 const int out_width = para->out_shape_[2]; in Fp16Crop3D()
121 const int out_batch = para->out_shape_[0]; in Fp16Crop4D()
122 const int out_height = para->out_shape_[1]; in Fp16Crop4D()
123 const int out_width = para->out_shape_[2]; in Fp16Crop4D()
124 const int out_channel = para->out_shape_[3]; in Fp16Crop4D()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/int8/
Dleaky_relu_int8.cc72 if (quant_prelu_parm_.out_shape_ != nullptr) { in ~LeakyReluInt8CPUKernel()
73 free(quant_prelu_parm_.out_shape_); in ~LeakyReluInt8CPUKernel()
74 quant_prelu_parm_.out_shape_ = nullptr; in ~LeakyReluInt8CPUKernel()
99 if (quant_prelu_parm_.out_shape_ != nullptr) { in ReSize()
100 free(quant_prelu_parm_.out_shape_); in ReSize()
101 quant_prelu_parm_.out_shape_ = nullptr; in ReSize()
103 quant_prelu_parm_.out_shape_ = reinterpret_cast<int *>(malloc(output_dim * sizeof(int))); in ReSize()
104 if (quant_prelu_parm_.out_shape_ == nullptr) { in ReSize()
108 …memcpy(reinterpret_cast<void *>(const_cast<int *>(quant_prelu_parm_.out_shape_)), output_shape.dat… in ReSize()
Dtranspose_int8.cc78 CHECK_NULL_RETURN(out_shape_); in DoTranspose()
80 …TransposeDimsInt8(in_ptr_, out_ptr_, out_shape_, transpose_param_, task_id, op_parameter_->thread_… in DoTranspose()
128 memcpy(out_shape_, out_dims.data(), out_dims.size() * sizeof(int)); in Run()
133 return DoTransposeInt8(in_ptr_, out_ptr_, out_shape_, transpose_param_); in Run()
Dadd_int8.cc128 if (arith_para_->out_shape_[i] == -1) { in ReSize()
129 memcpy(arith_para_->out_shape_, output->shape().data(), output->shape().size() * sizeof(int)); in ReSize()
146 in_size_ *= arith_para_->out_shape_[i]; in ReSize()
148 out_size_ *= arith_para_->out_shape_[i]; in ReSize()
154 ComputeStrides(arith_para_->out_shape_, arith_para_->out_strides_, arith_para_->ndim_); in ReSize()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp32/
Dtranspose_fp32.cc79 if (this->out_shape_ != nullptr) { in ReSize()
80 free(this->out_shape_); in ReSize()
81 this->out_shape_ = nullptr; in ReSize()
84 out_shape_ = reinterpret_cast<int *>(malloc(out_shape.size() * sizeof(int))); in ReSize()
85 if (out_shape_ == nullptr) { in ReSize()
89 memcpy(out_shape_, out_shape.data(), in_shape.size() * sizeof(int)); in ReSize()
94 if (this->out_shape_ != nullptr) { in ~TransposeCPUKernel()
95 free(this->out_shape_); in ~TransposeCPUKernel()
112 …oseFp32(static_cast<const float *>(in_data_), static_cast<float *>(out_data_), out_shape_, param_); in TransposeDim2to6()
116 …static_cast<const float *>(in_data_), static_cast<float *>(out_data_), out_shape_, param_, task_id, in TransposeDimGreaterThan6()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/opencl/kernel/int8/
Darithmetic_int8.cc82 global_size_ = {out_shape_.width, out_shape_.height}; in SetGlobalLocal()
84 global_size_ = {out_shape_.Slice, out_shape_.W, out_shape_.H * out_shape_.N}; in SetGlobalLocal()
121 …cl_int4 out_shape = {static_cast<int>(out_shape_.N), static_cast<int>(out_shape_.H), static_cast<i… in SetConstArgs()
122 static_cast<int>(out_shape_.Slice)}; in SetConstArgs()
184 out_shape_ = GpuTensorInfo(out_tensors_[0]); in Prepare()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/int8/
Dcrop_int8.c40 const int out_batch = para->out_shape_[0]; in Int8Crop1D()
80 const int out_batch = para->out_shape_[0]; in Int8Crop2D()
81 const int out_height = para->out_shape_[1]; in Int8Crop2D()
125 const int out_batch = para->out_shape_[0]; in Int8Crop3D()
126 const int out_height = para->out_shape_[1]; in Int8Crop3D()
127 const int out_width = para->out_shape_[2]; in Int8Crop3D()
181 const int out_batch = para->out_shape_[0]; in Int8Crop4D()
182 const int out_height = para->out_shape_[1]; in Int8Crop4D()
183 const int out_width = para->out_shape_[2]; in Int8Crop4D()
184 const int out_channel = para->out_shape_[3]; in Int8Crop4D()
/third_party/mindspore/mindspore/lite/micro/coder/opcoders/nnacl/fp32/
Dtranspose_fp32_coder.cc50 out_shape_ = in Resize()
52 MS_CHECK_PTR(out_shape_); in Resize()
53 memcpy(out_shape_, out_shape.data(), in_shape.size() * sizeof(int)); in Resize()
136 *(dim_size + i - 1) = *(dim_size + i) * out_shape_[i]; in DoCode()
145 …code.CodeFunction("TransposeDimsFp32", input_tensor_, output_tensor_, out_shape_, "dim_size", "pos… in DoCode()
150 code.CodeFunction("DoTransposeFp32", input_tensor_, output_tensor_, out_shape_, "&trans_param"); in DoCode()
Dtile_fp32_coder.cc43 tile_param_->out_shape_[i] = output_tensor_->shape().at(i); in Resize()
46 ComputeStrides(tile_param_->out_shape_, tile_param_->out_strides_, tile_param_->in_dim_); in Resize()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/base/
Darithmetic_base.c22 param->multiples0_[i] = param->out_shape_[i] / param->in_shape0_[i]; in CalcMultiplesAndStrides()
25 param->multiples1_[i] = param->out_shape_[i] / param->in_shape1_[i]; in CalcMultiplesAndStrides()
31 ComputeStrides(param->out_shape_, param->out_strides_, param->ndim_); in CalcMultiplesAndStrides()
/third_party/mindspore/mindspore/lite/micro/coder/opcoders/nnacl/int8/
Dadd_int8_coder.cc107 if (arith_para_->out_shape_[i] == -1) { in ReSize()
108 …MS_CHECK_RET_CODE(memcpy_s(arith_para_->out_shape_, std::extent<decltype(arith_para_->out_shape_)>… in ReSize()
127 in_size_ *= arith_para_->out_shape_[i]; in ReSize()
129 out_size_ *= arith_para_->out_shape_[i]; in ReSize()
135 ComputeStrides(arith_para_->out_shape_, arith_para_->out_strides_, arith_para_->ndim_); in ReSize()
/third_party/mindspore/mindspore/lite/test/ut/nnacl/infer/
Dmax_min_grad_infer_test.cc64 ASSERT_EQ(parameter->out_shape_[0], 7); in TEST_F()
65 ASSERT_EQ(parameter->out_shape_[1], 8); in TEST_F()
66 ASSERT_EQ(parameter->out_shape_[2], 9); in TEST_F()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp16_grad/
Dbias_fp16_grad.cc39 bias_param->out_shape_[i] = 1; // 1 dimension for N,H,W, in ReSize()
41 bias_param->out_shape_[bias_param->ndim_ - 1] = dims[bias_param->ndim_ - 1]; in ReSize()
44 bias_param->out_shape_[i] = 0; in ReSize()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp32_grad/
Dbias_grad.cc35 bias_param->out_shape_[i] = 1; // 1 dimension for N,H,W, in ReSize()
37 bias_param->out_shape_[bias_param->ndim_ - 1] = dims[bias_param->ndim_ - 1]; in ReSize()
40 bias_param->out_shape_[i] = 0; in ReSize()
Darithmetic_grad.cc84 ReduceSumByAxes(dy, arithmeticParameter_->out_shape_, dx1, arithmeticParameter_->in_shape0_, in ArithmeticGradAdd()
90 ReduceSumByAxes(dy, arithmeticParameter_->out_shape_, dx2, arithmeticParameter_->in_shape1_, in ArithmeticGradAdd()
101 ReduceSumByAxes(dy, arithmeticParameter_->out_shape_, dx1, arithmeticParameter_->in_shape0_, in ArithmeticGradSub()
109 ReduceSumByAxes(dy, arithmeticParameter_->out_shape_, dx2, arithmeticParameter_->in_shape1_, in ArithmeticGradSub()
218 arithmeticParameter_->out_shape_, dx1, dx2, arithmeticParameter_->ndim_); in ArithmeticGradMaximum()
232 arithmeticParameter_->out_shape_, dx1, dx2, arithmeticParameter_->ndim_); in ArithmeticGradMinimum()
/third_party/mindspore/mindspore/lite/src/runtime/kernel/arm/fp16/
Dtranspose_fp16.cc37 …poseFp16(static_cast<const float16_t *>(in_data_), static_cast<float16_t *>(out_data_), out_shape_, in TransposeDim2to6()
42 …(static_cast<const float16_t *>(in_data_), static_cast<float16_t *>(out_data_), out_shape_, param_, in TransposeDimGreaterThan6()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/nnacl/infer/
Darithmetic_grad_infer.c70 param->out_shape_[i] = out_shape[i]; in ArithmeticGradInferShape()
87 param->out_shape_[i] = out_shape[i]; in ArithmeticGradInferShape()
94 param->out_shape_[i] = out_shape[i]; in ArithmeticGradInferShape()

123