/external/tensorflow/tensorflow/stream_executor/cuda/ |
D | cuda_fft.cc | 80 fft::Type type, int batch_count, ScratchAllocator *scratch_allocator) { in Initialize() argument 100 if (scratch_allocator == nullptr) { in Initialize() 188 return UpdateScratchAllocator(stream, scratch_allocator); in Initialize() 192 if (scratch_allocator == nullptr) { in Initialize() 227 return UpdateScratchAllocator(stream, scratch_allocator); in Initialize() 236 ScratchAllocator *scratch_allocator) { in Initialize() argument 241 /*output_distance=*/0, type, 1, scratch_allocator); in Initialize() 245 Stream *stream, ScratchAllocator *scratch_allocator) { in UpdateScratchAllocator() argument 248 scratch_allocator->AllocateBytes(stream, scratch_size_bytes_); in UpdateScratchAllocator() 310 ScratchAllocator *scratch_allocator) { in Create1dPlanWithScratchAllocator() argument [all …]
|
D | cuda_fft.h | 72 int batch_count, ScratchAllocator* scratch_allocator); 77 ScratchAllocator* scratch_allocator); 80 ScratchAllocator *scratch_allocator);
|
D | cuda_dnn.cc | 2215 ScratchAllocator* scratch_allocator) { in AllocateCudnnConvolutionForwardWorkspace() argument 2243 if (TF_PREDICT_FALSE(!scratch_allocator)) { in AllocateCudnnConvolutionForwardWorkspace() 2248 return scratch_allocator->AllocateBytes(stream, size_in_bytes); in AllocateCudnnConvolutionForwardWorkspace() 2258 ScratchAllocator* scratch_allocator) { in AllocateCudnnConvolutionBackwardDataWorkspace() argument 2288 if (TF_PREDICT_FALSE(!scratch_allocator)) { in AllocateCudnnConvolutionBackwardDataWorkspace() 2293 return scratch_allocator->AllocateBytes(stream, size_in_bytes); in AllocateCudnnConvolutionBackwardDataWorkspace() 2303 ScratchAllocator* scratch_allocator) { in AllocateCudnnConvolutionBackwardFilterWorkspace() argument 2333 if (TF_PREDICT_FALSE(!scratch_allocator)) { in AllocateCudnnConvolutionBackwardFilterWorkspace() 2338 return scratch_allocator->AllocateBytes(stream, size_in_bytes); in AllocateCudnnConvolutionBackwardFilterWorkspace() 2346 const CudnnTensorDescriptor& output_nd, ScratchAllocator* scratch_allocator, in GetCudnnConvolutionForwardAlgorithm() argument [all …]
|
D | cuda_dnn.h | 283 DeviceMemory<double>* output_data, ScratchAllocator* scratch_allocator, 297 DeviceMemory<float>* output_data, ScratchAllocator* scratch_allocator, 315 ScratchAllocator* scratch_allocator, 329 DeviceMemory<int8>* output_data, ScratchAllocator* scratch_allocator, 594 ScratchAllocator* scratch_allocator, 660 ScratchAllocator* scratch_allocator, dnn::AlgorithmDesc* algorithm_desc,
|
D | BUILD | 208 "//tensorflow/stream_executor:scratch_allocator", 246 "//tensorflow/stream_executor:scratch_allocator", 291 "//tensorflow/stream_executor:scratch_allocator",
|
/external/tensorflow/tensorflow/stream_executor/ |
D | fft.h | 110 ScratchAllocator *scratch_allocator) = 0; 115 ScratchAllocator *scratch_allocator) = 0; 120 bool in_place_fft, ScratchAllocator *scratch_allocator) = 0; 167 ScratchAllocator *scratch_allocator) = 0; 176 Stream *stream, Plan *plan, ScratchAllocator *scratch_allocator) = 0; 225 ScratchAllocator *scratch_allocator) override; \ 228 bool in_place_fft, ScratchAllocator *scratch_allocator) override; \ 231 fft::Type type, bool in_place_fft, ScratchAllocator *scratch_allocator) \ 242 bool in_place_fft, int batch_count, ScratchAllocator *scratch_allocator) \ 245 ScratchAllocator *scratch_allocator) \
|
D | stream.h | 311 DeviceMemory<double> *output, ScratchAllocator *scratch_allocator, 322 DeviceMemory<float> *output, ScratchAllocator *scratch_allocator, 333 DeviceMemory<Eigen::half> *output, ScratchAllocator *scratch_allocator, 347 DeviceMemory<double> *output, ScratchAllocator *scratch_allocator, 361 DeviceMemory<float> *output, ScratchAllocator *scratch_allocator, 376 DeviceMemory<Eigen::half> *output, ScratchAllocator *scratch_allocator, 390 ScratchAllocator *scratch_allocator, 412 ScratchAllocator *scratch_allocator, 424 ScratchAllocator *scratch_allocator, 436 ScratchAllocator *scratch_allocator, [all …]
|
D | BUILD | 447 name = "scratch_allocator", 448 srcs = ["scratch_allocator.cc"], 449 hdrs = ["scratch_allocator.h"],
|
D | stream.cc | 450 ScratchAllocator *scratch_allocator, in ThenFusedConvolveWithAlgorithm() argument 466 activation_mode, output_descriptor, output, scratch_allocator, in ThenFusedConvolveWithAlgorithm() 488 ScratchAllocator *scratch_allocator, in ThenFusedConvolveWithAlgorithm() argument 504 activation_mode, output_descriptor, output, scratch_allocator, in ThenFusedConvolveWithAlgorithm() 527 DeviceMemory<Eigen::half> *output, ScratchAllocator *scratch_allocator, in ThenFusedConvolveWithAlgorithm() argument 543 activation_mode, output_descriptor, output, scratch_allocator, in ThenFusedConvolveWithAlgorithm() 565 ScratchAllocator *scratch_allocator, in ThenFusedConvolveWithAlgorithm() argument 581 activation_mode, output_descriptor, output, scratch_allocator, in ThenFusedConvolveWithAlgorithm() 600 ScratchAllocator *scratch_allocator, in ThenConvolveWithAlgorithm() argument 617 scratch_allocator, &algorithm_desc, &scratch_memory) in ThenConvolveWithAlgorithm() [all …]
|
/external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
D | cudnn_conv_runner.cc | 114 se::ScratchAllocator* scratch_allocator, in RunCudnnConvImpl() argument 248 scratch_allocator, algorithm, profile_result); in RunCudnnConvImpl() 259 scratch_allocator, algorithm, profile_result); in RunCudnnConvImpl() 270 scratch_allocator, algorithm, profile_result); in RunCudnnConvImpl() 303 output_descriptor, &output_buf, scratch_allocator, algorithm, in RunCudnnConvImpl() 399 ScratchBufAllocator scratch_allocator(scratch_buf); in RunCudnnConv() local 400 return RunCudnnConv(conv, operand_buffers, result_buffer, &scratch_allocator, in RunCudnnConv() 407 se::ScratchAllocator* scratch_allocator, se::Stream* stream, in RunCudnnConv() argument 420 return RunCudnnConvImpl<Eigen::half>(params, scratch_allocator, stream, in RunCudnnConv() 423 return RunCudnnConvImpl<float>(params, scratch_allocator, stream, in RunCudnnConv() [all …]
|
D | fft_thunk.cc | 117 FftScratchAllocator scratch_allocator(buffer_allocations.device_ordinal(), in ExecuteOnStream() local 150 fft_type_, kInPlaceFft, batch_size, &scratch_allocator); in ExecuteOnStream() 154 stream, fft_plan_.get(), &scratch_allocator); in ExecuteOnStream()
|
D | cusolver_rewriter.cc | 187 ScratchAllocator scratch_allocator(device_ordinal, allocator); in RunOnComputation() local 196 RunOnInstruction(&context, &scratch_allocator, instruction)); in RunOnComputation()
|
D | cudnn_conv_algorithm_picker.cc | 236 ScratchAllocator scratch_allocator(device_ordinal, allocator); in PickBestAlgorithm() local 247 &scratch_allocator, &stream, options); in PickBestAlgorithm() 262 int64 scratch_bytes_used = scratch_allocator.TotalAllocatedBytes(); in PickBestAlgorithm()
|
D | cudnn_conv_runner.h | 62 se::ScratchAllocator* scratch_allocator, se::Stream* stream,
|
/external/tensorflow/tensorflow/stream_executor/rocm/ |
D | rocm_fft.cc | 149 fft::Type type, int batch_count, ScratchAllocator *scratch_allocator) { in Initialize() argument 168 if (scratch_allocator == nullptr) { in Initialize() 264 scratch_allocator->AllocateBytes(stream, size_in_bytes); in Initialize() 281 if (scratch_allocator == nullptr) { in Initialize() 320 scratch_allocator->AllocateBytes(stream, size_in_bytes); in Initialize() 341 ScratchAllocator *scratch_allocator) { in Initialize() argument 346 /*output_distance=*/0, type, 1, scratch_allocator); in Initialize() 390 ScratchAllocator *scratch_allocator) { in Create1dPlanWithScratchAllocator() argument 394 type, scratch_allocator); in Create1dPlanWithScratchAllocator() 419 bool in_place_fft, ScratchAllocator *scratch_allocator) { in Create2dPlanWithScratchAllocator() argument [all …]
|
D | rocm_fft.h | 71 int batch_count, ScratchAllocator *scratch_allocator); 76 ScratchAllocator *scratch_allocator);
|
D | BUILD | 172 "//tensorflow/stream_executor:scratch_allocator", 199 "//tensorflow/stream_executor:scratch_allocator", 236 "//tensorflow/stream_executor:scratch_allocator",
|
D | rocm_dnn.h | 274 DeviceMemory<double>* output_data, ScratchAllocator* scratch_allocator, 288 DeviceMemory<float>* output_data, ScratchAllocator* scratch_allocator, 306 ScratchAllocator* scratch_allocator, 320 DeviceMemory<int8>* output_data, ScratchAllocator* scratch_allocator, 767 ScratchAllocator* scratch_allocator, dnn::AlgorithmDesc* algorithm_desc,
|
/external/tensorflow/tensorflow/core/kernels/ |
D | conv_ops_3d.cc | 454 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in launch() local 460 output_desc, &output_ptr, &scratch_allocator, in launch() 471 scratch_allocator.TotalByteSize()); in launch() 484 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in launch() local 489 &output_ptr, &scratch_allocator, in launch()
|
D | conv_ops_fused_impl.h | 526 DnnScratchAllocator scratch_allocator(ConvolveScratchSize(), context); 530 launch(se::dnn::AlgorithmConfig(profile_algorithm), &scratch_allocator, 540 scratch_allocator.TotalByteSize()); 777 DnnScratchAllocator* scratch_allocator, 789 scratch_allocator, algorithm_config, profile_result) 806 DnnScratchAllocator scratch_allocator(ConvolveScratchSize(), context); 807 bool cudnn_launch_status = launch(algorithm_config, &scratch_allocator,
|
D | conv_grad_ops_3d.cc | 1370 DnnScratchAllocator scratch_allocator(ConvolveBackwardDataScratchSize, in Compute() local 1377 conv_desc, input_desc, &in_backprop_ptr, &scratch_allocator, in Compute() 1386 if (scratch_allocator.TotalByteSize() == 0 && in Compute() 1407 DnnScratchAllocator scratch_allocator(ConvolveBackwardDataScratchSize, in Compute() local 1413 conv_desc, input_desc, &in_backprop_ptr, &scratch_allocator, in Compute() 1776 DnnScratchAllocator scratch_allocator(ConvolveBackwardFilterScratchSize, in Compute() local 1784 &scratch_allocator, AlgorithmConfig(profile_algorithm), in Compute() 1793 if (scratch_allocator.TotalByteSize() == 0 && in Compute() 1814 DnnScratchAllocator scratch_allocator(ConvolveBackwardFilterScratchSize, in Compute() local 1820 filter_desc, &filter_backprop_ptr, &scratch_allocator, in Compute()
|
D | batch_matmul_op_impl.h | 385 CublasScratchAllocator scratch_allocator(context); 392 batch_size, &scratch_allocator) 477 CublasScratchAllocator scratch_allocator(context); 484 batch_size, &scratch_allocator)
|
D | conv_ops.cc | 864 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in operator ()() local 870 output_desc, &output_ptr, &scratch_allocator, in operator ()() 881 scratch_allocator.TotalByteSize()); in operator ()() 894 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in operator ()() local 899 &output_ptr, &scratch_allocator, in operator ()()
|
/external/tensorflow/tensorflow/contrib/fused_conv/kernels/ |
D | fused_conv2d_bias_activation_op.cc | 718 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in launch() local 726 output_desc, &output_ptr, &scratch_allocator, in launch() 737 scratch_allocator.TotalByteSize()); in launch() 753 DnnScratchAllocator scratch_allocator(ConvolveScratchSize, ctx); in launch() local 760 &output_ptr, &scratch_allocator, algorithm_config, in launch()
|
/external/tensorflow/tensorflow/core/platform/default/build_config/ |
D | BUILD | 46 "//tensorflow/stream_executor:scratch_allocator", 96 "//tensorflow/stream_executor:scratch_allocator",
|