| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/nn/ |
| D | flatten_grad_gpu_kernel.h | 46 …cudaMemcpyAsync(output, input, input_size_, cudaMemcpyDeviceToDevice, reinterpret_cast<cudaStream_… in Launch()
68 if (input_size_ == 0) { in Init()
69 input_size_ = 1; in Init()
71 input_size_ *= shape[i]; in Init()
73 input_size_ = input_size_ * sizeof(T); in Init()
80 input_size_ = 0; in ResetResource()
89 input_size_list_.push_back(input_size_); in InitSizeLists()
90 output_size_list_.push_back(input_size_); in InitSizeLists()
98 size_t input_size_; variable
|
| D | binary_cross_entropy_grad_kernel.h | 31 …BinaryCrossEntropyGradGpuKernel() : input_size_(1), reduction_(1), weight_defined_(false), is_null… in BinaryCrossEntropyGradGpuKernel()
51 if (input_size_ > 0) { in Launch()
52 BinaryCrossEntropyLossGrad(input_size_, reduction_, input_x, input_y, weight, dloss, dx, in Launch()
69 input_size_ *= input_shape[i]; in Init()
83 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
84 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
86 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
91 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
93 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
97 size_t input_size_;
|
| D | gelu_grad_kernel.h | 31 GeLUGpuGradKernel() : input_size_(0), is_null_input_(false) {} in GeLUGpuGradKernel()
46 …GeluGradKernel(input_size_ / sizeof(T), dy_addr, x_addr, dx_addr, reinterpret_cast<cudaStream_t>(s… in Launch()
52 input_size_ = sizeof(T); in Init()
61 input_size_ *= dim; in Init()
69 input_size_list_.push_back(input_size_); in InitSizeLists()
70 input_size_list_.push_back(input_size_); in InitSizeLists()
71 input_size_list_.push_back(input_size_); in InitSizeLists()
72 output_size_list_.push_back(input_size_); in InitSizeLists()
79 size_t input_size_; variable
|
| D | binary_cross_entropy_gpu_kernel.h | 31 …BinaryCrossEntropyGpuKernel() : weight_defined_(false), input_size_(1), reduction_(1), workspace_s… in BinaryCrossEntropyGpuKernel()
49 if (input_size_ > 0) { in Launch()
50 BinaryCrossEntropyLoss(input_size_, reduction_, input_x, input_y, weight, loss, tmp_loss, in Launch()
67 input_size_ *= input_shape[i]; in Init()
77 workspace_size_ *= input_size_; in Init()
85 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
86 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
88 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
91 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
101 size_t input_size_; variable
|
| D | layer_norm_grad_grad_gpu_kernel.h | 31 …: input_row_(1), input_col_(1), param_dim_(1), input_size_(1), is_null_input_(false), epsilon_() {} in LayerNormGradGradGpuKernel()
59 … cudaMemsetAsync(global_sum1, 0, input_size_, reinterpret_cast<cudaStream_t>(stream_ptr)), in Launch()
62 … cudaMemsetAsync(global_sum2, 0, input_size_, reinterpret_cast<cudaStream_t>(stream_ptr)), in Launch()
118 input_size_ = input_row_ * input_col_ * sizeof(T); in InitSizeLists()
119 input_size_list_.push_back(input_size_); in InitSizeLists()
120 input_size_list_.push_back(input_size_); in InitSizeLists()
124 input_size_list_.push_back(input_size_); in InitSizeLists()
128 output_size_list_.push_back(input_size_); in InitSizeLists()
129 output_size_list_.push_back(input_size_); in InitSizeLists()
132 workspace_size_list_.push_back(input_size_); in InitSizeLists()
[all …]
|
| D | kl_div_loss_grad_kernel.h | 31 KLDivLossGradGpuKernel() : input_size_(1), reduction_(1), is_null_input_(false) {} in KLDivLossGradGpuKernel()
48 …KLDivLossGrad(input_size_, reduction_, input_x, input_y, dloss, dx, dy, reinterpret_cast<cudaStrea… in Launch()
61 input_size_ *= input_shape[i]; in Init()
75 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
76 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
77 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
78 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
80 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
87 size_t input_size_;
|
| D | softplus_grad_gpu_kernel.h | 31 SoftplusGpuGradKernel() : input_size_(0) {} in SoftplusGpuGradKernel()
46 …SoftplusGrad(input_size_ / sizeof(T), dy_addr, x_addr, dx_addr, reinterpret_cast<cudaStream_t>(str… in Launch()
52 input_size_ = sizeof(T); in Init()
61 input_size_ *= dim; in Init()
69 input_size_list_.push_back(input_size_); in InitSizeLists()
70 input_size_list_.push_back(input_size_); in InitSizeLists()
71 output_size_list_.push_back(input_size_); in InitSizeLists()
79 size_t input_size_; variable
|
| D | flatten_gpu_kernel.h | 30 FlattenGpuFwdKernel() : input_size_(0), is_null_input_(false) {} in FlattenGpuFwdKernel()
45 …cudaMemcpyAsync(output, input, input_size_, cudaMemcpyDeviceToDevice, reinterpret_cast<cudaStream_… in Launch()
60 input_size_ = sizeof(T); in Init()
62 input_size_ *= shape[i]; in Init()
68 input_size_ = 0; in ResetResource()
77 input_size_list_.push_back(input_size_); in InitSizeLists()
78 output_size_list_.push_back(input_size_); in InitSizeLists()
86 size_t input_size_; variable
|
| D | kl_div_loss_gpu_kernel.h | 31 KLDivLossGpuKernel() : input_size_(1), reduction_(1), is_null_input_(false), workspace_size_(0) {} in KLDivLossGpuKernel()
47 …KLDivLoss(input_size_, reduction_, input_x, input_y, loss, tmp_loss, reinterpret_cast<cudaStream_t… in Launch()
60 input_size_ *= input_shape[i]; in Init()
70 workspace_size_ *= input_size_; in Init()
78 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
79 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
81 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
89 size_t input_size_;
|
| D | hsigmoid_grad_gpu_kernel.h | 46 CalHSigmoidGrad(input_size_, input, x, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
67 input_size_ = 1; in Init()
69 input_size_ *= input_shape[i]; in Init()
76 input_size_ = 1; in ResetResource()
85 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
86 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
87 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
91 size_t input_size_;
|
| D | hswish_grad_gpu_kernel.h | 46 CalHSwishGrad(input_size_, input, x, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
67 input_size_ = 1; in Init()
69 input_size_ *= input_shape[i]; in Init()
76 input_size_ = 1; in ResetResource()
85 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
86 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
87 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
91 size_t input_size_;
|
| D | gelu_kernel.h | 31 GeluGpuKernel() : input_size_(0), is_null_input_(false) {} in GeluGpuKernel()
45 …Gelu(input_size_ / sizeof(T), input_addr, output_addr, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
51 input_size_ = sizeof(T); in Init()
60 input_size_ *= dim; in Init()
68 input_size_list_.push_back(input_size_); in InitSizeLists()
69 output_size_list_.push_back(input_size_); in InitSizeLists()
76 size_t input_size_; variable
|
| D | softplus_gpu_kernel.h | 31 SoftplusGpuKernel() : is_null_input_(false), input_size_(0) {} in SoftplusGpuKernel()
45 …Softplus(input_size_ / sizeof(T), input_addr, output_addr, reinterpret_cast<cudaStream_t>(stream_p… in Launch()
51 input_size_ = sizeof(T); in Init()
60 input_size_ *= dim; in Init()
68 input_size_list_.push_back(input_size_); in InitSizeLists()
69 output_size_list_.push_back(input_size_); in InitSizeLists()
77 size_t input_size_; variable
|
| D | apply_gradient_descent_gpu_kernel.h | 47 …CalApplyGradientDescent(input_size_, var, alpha, delta, output, reinterpret_cast<cudaStream_t>(str… in Launch()
68 input_size_ = 1; in Init()
70 input_size_ *= input_shape[i]; in Init()
77 input_size_ = 1; in ResetResource()
86 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
88 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
89 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
93 size_t input_size_;
|
| D | smooth_l1_loss_grad_gpu_kernel.h | 29 SmoothL1LossGradGpuKernel() : input_size_(1), beta_(1.0) {} in SmoothL1LossGradGpuKernel()
46 …SmoothL1LossGrad(input_size_, beta_, prediction, target, dloss, dx, reinterpret_cast<cudaStream_t>… in Launch()
59 input_size_ *= input_shape[i]; in Init()
69 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
70 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
71 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
75 size_t input_size_;
|
| D | smooth_l1_loss_gpu_kernel.h | 29 SmoothL1LossGpuKernel() : input_size_(1), beta_(1.0), is_null_input_(false) {} in SmoothL1LossGpuKernel()
45 …SmoothL1Loss(input_size_, beta_, prediction, target, loss, reinterpret_cast<cudaStream_t>(stream_p… in Launch()
58 input_size_ *= input_shape[i]; in Init()
68 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
69 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
70 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
74 size_t input_size_;
|
| D | hsigmoid_gpu_kernel.h | 45 CalHSigmoid(input_size_, input, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
66 input_size_ = 1; in Init()
68 input_size_ *= input_shape[i]; in Init()
75 input_size_ = 1; in ResetResource()
84 input_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
85 output_size_list_.push_back(input_size_ * sizeof(T)); in InitSizeLists()
89 size_t input_size_;
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/other/ |
| D | assign_gpu_kernel.h | 29 AssignGpuKernel() : input_size_(0), is_null_input_(false) {} in AssignGpuKernel()
45 …cudaMemcpyAsync(var, value, input_size_, cudaMemcpyDeviceToDevice, reinterpret_cast<cudaStream_t>(… in Launch()
49 …cudaMemcpyAsync(output, value, input_size_, cudaMemcpyDeviceToDevice, reinterpret_cast<cudaStream_… in Launch()
67 input_size_ = sizeof(T); in Init()
69 input_size_ = input_size_ * x; in Init()
77 input_size_list_.push_back(input_size_); in InitSizeLists()
78 input_size_list_.push_back(input_size_); in InitSizeLists()
79 output_size_list_.push_back(input_size_); in InitSizeLists()
102 size_t input_size_; variable
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/math/ |
| D | assign_add_gpu_kernel.h | 30 AssignAddGpuFwdKernel() : is_null_input_(false), input_size_(0) {} in AssignAddGpuFwdKernel()
46 CalAssignAdd(input_size_ / sizeof(T), input_addr, input_addr2, output_addr, in Launch()
69 input_size_ = sizeof(T); in Init()
71 input_size_ = i * input_size_; in Init()
79 input_size_list_.push_back(input_size_); in InitSizeLists()
80 input_size_list_.push_back(input_size_); in InitSizeLists()
81 output_size_list_.push_back(input_size_); in InitSizeLists()
90 size_t input_size_; variable
|
| D | float_status_gpu_kernel.h | 37 …FloatStatusGpuKernel() : kernel_name_(OP_INVALID), input_size_(0), output_size_(0), is_null_input_… in FloatStatusGpuKernel()
54 …CalFloatStatus(input_size_ / sizeof(T), input, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
59 … CalIsInf(input_size_ / sizeof(T), input, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
64 … CalIsNan(input_size_ / sizeof(T), input, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
69 … CalIsFinite(input_size_ / sizeof(T), input, output, reinterpret_cast<cudaStream_t>(stream_ptr)); in Launch()
90 input_size_ = sizeof(T); in Init()
92 input_size_ = input_size_ * x; in Init()
104 output_size_ = input_size_ / sizeof(T) * sizeof(bool); in Init()
112 input_size_list_.push_back(input_size_); in InitSizeLists()
135 size_t input_size_; variable
|
| D | equalcount_gpu_kernel.h | 30 …EqualCountGpuKernel() : input_size_(0), output_size_(0), workspace_size_(0), is_null_input_(false)… in EqualCountGpuKernel()
46 int size = SizeToInt(input_size_ / sizeof(T)); in Launch()
64 input_size_ = sizeof(T); in Init()
73 input_size_ *= input_shape[i]; in Init()
81 input_size_list_.push_back(input_size_); in InitSizeLists()
82 input_size_list_.push_back(input_size_); in InitSizeLists()
88 size_t input_size_;
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/quant/ |
| D | fake_learned_scale_quant_perlayer_grad_gpu_kernel.cc | 22 …: input_size_(0), workspace_size_(0), quant_num_(1), quant_delay_(0), global_step_(0), neg_trunc_(… in FakeLearnedScaleQuantPerLayerGradGpuKernel()
62 input_size_ = sizeof(float); in Init()
64 input_size_ *= input_shape[i]; in Init()
71 input_size_list_.push_back(input_size_); // gradient in InitSizeLists()
72 input_size_list_.push_back(input_size_); // input in InitSizeLists()
75 output_size_list_.push_back(input_size_); // grad_input in InitSizeLists()
77 workspace_size_list_.push_back(input_size_); // input_div_alpha in InitSizeLists()
78 workspace_size_list_.push_back(input_size_); // input_quant in InitSizeLists()
119 … cudaMemcpyAsync(grad_input, gradient, input_size_, cudaMemcpyDeviceToDevice, in Launch()
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/arrays/ |
| D | range_gpu_kernel.h | 28 …RangeGPUKernel() : input_size_(0), output_size_(0), start_(0.), limit_(1.), delta_(1.), is_null_in… in RangeGPUKernel()
42 int size = SizeToInt(input_size_ / sizeof(T)); in Launch()
66 input_size_ = 1; in Init()
68 input_size_ *= input_shape[i]; in Init()
70 input_size_ *= sizeof(T); in Init()
71 output_size_ = input_size_; in Init()
81 input_size_list_.push_back(input_size_); in InitSizeLists()
90 size_t input_size_; variable
|
| D | sort_gpu_kernel.h | 79 Negative(intermediate_input_device, intermediate_output_device, input_size_, in Launch()
86 … CalTranspose(input_size_, intermediate_input_device, input_shape_device, perm_device, input_rank_, in Launch()
102 …CalTranspose(input_size_, intermediate_input_device, transposed_shape_device, perm_device, input_r… in Launch()
106 …CalTranspose(input_size_, temp_indices_device, transposed_shape_device, perm_device, input_rank_, … in Launch()
112 Negative(intermediate_input_device, intermediate_output_device, input_size_, in Launch()
147 input_size_ = 1; in Init()
149 input_size_ *= input_shape_[i]; in Init()
172 outer_size_ = input_size_ / inner_size_; in Init()
187 input_size_ = 0; in ResetResource()
205 size_t input_bytes = input_size_ * sizeof(T); in InitSizeLists()
[all …]
|
| /third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/cpu/ |
| D | unique_cpu_kernel.h | 39 IndexType input_size_{0};
60 size_t input_size_{0};
81 if (params->input_size_ < 1) { in CalculateEachBucketSize()
84 for (IndexType i = 0; i < params->input_size_; ++i) { in CalculateEachBucketSize()
102 IndexType input_size = params->input_size_; in SplitAndCalculateBucketSize()
122 segments[i]->input_size_ = data_size; in SplitAndCalculateBucketSize()
143 if (segment->input_size_ < 1) { in SegmentToBuckets()
146 for (IndexType i = 0; i < segment->input_size_; ++i) { in SegmentToBuckets()
156 if (bucket_index >= bucket->input_size_) { in SegmentToBuckets()
202 bucket->input_size_ = bucket_data_size[i]; in GatherSegmentsToBuckets()
[all …]
|