| /external/tensorflow/tensorflow/compiler/xla/stream_executor/cuda/ |
| D | cuda_dnn.cc | 2891 const dnn::AlgorithmDesc& algorithm_desc, in AllocateCudnnConvolutionForwardWorkspace() argument
2893 if (IsTensorMathOpSet(conv) != algorithm_desc.tensor_ops_enabled()) { in AllocateCudnnConvolutionForwardWorkspace()
2901 if (algorithm_desc.workspace_size()) { in AllocateCudnnConvolutionForwardWorkspace()
2902 size_in_bytes = *algorithm_desc.workspace_size(); in AllocateCudnnConvolutionForwardWorkspace()
2909 /*algo=*/ToConvForwardAlgo(algorithm_desc), in AllocateCudnnConvolutionForwardWorkspace()
2940 const dnn::AlgorithmDesc& algorithm_desc, in AllocateCudnnConvolutionBackwardDataWorkspace() argument
2942 if (IsTensorMathOpSet(conv) != algorithm_desc.tensor_ops_enabled()) { in AllocateCudnnConvolutionBackwardDataWorkspace()
2950 if (algorithm_desc.workspace_size()) { in AllocateCudnnConvolutionBackwardDataWorkspace()
2951 size_in_bytes = *algorithm_desc.workspace_size(); in AllocateCudnnConvolutionBackwardDataWorkspace()
2959 /*algo=*/ToConvBackwardDataAlgo(algorithm_desc), in AllocateCudnnConvolutionBackwardDataWorkspace()
[all …]
|
| D | cuda_dnn.h | 234 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc,
257 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc,
344 dnn::AlgorithmDesc algorithm_desc, DeviceMemory<uint8> scratch_memory,
632 ScratchAllocator* scratch_allocator, dnn::AlgorithmDesc* algorithm_desc,
|
| /external/tensorflow/tensorflow/compiler/xla/stream_executor/ |
| D | dnn.h | 1330 ScratchAllocator* scratch_allocator, AlgorithmDesc* algorithm_desc, in PrepareForConvolution() argument
1336 scratch_allocator, algorithm_desc, scratch_memory); in PrepareForConvolution()
1380 AlgorithmDesc algorithm_desc, DeviceMemory<uint8> scratch_memory,
1403 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc,
1425 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc,
2626 ScratchAllocator* scratch_allocator, AlgorithmDesc* algorithm_desc, in DoPrepareForConvolution() argument
2628 *algorithm_desc = {}; in DoPrepareForConvolution()
|
| D | stream.h | 348 dnn::AlgorithmDesc algorithm_desc; in ConvolveWithAlgorithm() local
353 algorithm_config, scratch_allocator, &algorithm_desc, in ConvolveWithAlgorithm()
359 convolution_descriptor, algorithm_desc, in ConvolveWithAlgorithm()
394 const dnn::AlgorithmDesc &algorithm_desc, dnn::ConvolutionKind kind, in ConvolveRunnerFromDesc() argument
405 this, algorithm_desc, kind, element_type, output_type, input_descriptor, in ConvolveRunnerFromDesc()
411 const dnn::AlgorithmDesc &algorithm_desc, dnn::ConvolutionKind kind, in FusedConvolveRunnerFromDesc() argument
426 this, algorithm_desc, kind, element_type, bias_type, output_type, in FusedConvolveRunnerFromDesc()
|
| D | dnn.cc | 133 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc, in ConvolveRunnerFromDesc() argument
159 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc, in FusedConvolveRunnerFromDesc() argument
|
| /external/tensorflow/tensorflow/stream_executor/rocm/ |
| D | rocm_dnn.h | 254 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc,
348 dnn::AlgorithmDesc algorithm_desc, DeviceMemory<uint8> scratch_memory,
770 ScratchAllocator* scratch_allocator, dnn::AlgorithmDesc* algorithm_desc,
|
| D | rocm_dnn.cc | 2911 ScratchAllocator* scratch_allocator, dnn::AlgorithmDesc* algorithm_desc, in DoPrepareForConvolution() argument
2919 *algorithm_desc = *input_algo_desc; in DoPrepareForConvolution()
3119 dnn::AlgorithmDesc algorithm_desc, DeviceMemory<uint8> scratch_memory, in DoConvolve() argument
3123 ConvolveRunnerFromDesc(stream, algorithm_desc, kind, element_type, in DoConvolve()
3187 Stream* stream, const dnn::AlgorithmDesc& algorithm_desc, in ConvolveRunnerFromDesc() argument
3200 auto workspace_size = algorithm_desc.workspace_size(); in ConvolveRunnerFromDesc()
3207 parent_, miopen_.get(), algorithm_desc.algo_id(), *workspace_size, kind, in ConvolveRunnerFromDesc()
|