| /external/tensorflow/tensorflow/stream_executor/ |
| D | stream.h | 61 class DeviceMemory; variable
222 const DeviceMemory<float> &x, const DeviceMemory<float> &scale,
223 const DeviceMemory<float> &offset,
224 const DeviceMemory<float> &estimated_mean,
225 const DeviceMemory<float> &estimated_variance,
228 DeviceMemory<float> *y, DeviceMemory<float> *batch_mean,
229 DeviceMemory<float> *batch_var, DeviceMemory<float> *saved_mean,
230 DeviceMemory<float> *saved_inv_var, bool is_training,
231 std::function<const DeviceMemory<float> &()> var_to_inv_var,
235 const DeviceMemory<float> &y_backprop, const DeviceMemory<float> &x,
[all …]
|
| D | blas.h | 59 class DeviceMemory; variable
179 const DeviceMemory<float> &x, int incx,
180 DeviceMemory<float> *result) = 0;
182 const DeviceMemory<double> &x, int incx,
183 DeviceMemory<double> *result) = 0;
185 const DeviceMemory<std::complex<float>> &x, int incx,
186 DeviceMemory<float> *result) = 0;
188 const DeviceMemory<std::complex<double>> &x, int incx,
189 DeviceMemory<double> *result) = 0;
193 const DeviceMemory<float> &x, int incx,
[all …]
|
| D | dnn.h | 920 Stream* stream, const DeviceMemory<float>& x,
921 const DeviceMemory<float>& scale, const DeviceMemory<float>& offset,
922 const DeviceMemory<float>& estimated_mean,
923 const DeviceMemory<float>& estimated_variance,
926 DeviceMemory<float>* y, DeviceMemory<float>* batch_mean,
927 DeviceMemory<float>* batch_var, DeviceMemory<float>* reserve_space_1,
928 DeviceMemory<float>* reserve_space_2, bool is_training,
929 std::function<const DeviceMemory<float>&()> var_to_inv_var,
937 Stream* stream, const DeviceMemory<Eigen::half>& x,
938 const DeviceMemory<float>& scale, const DeviceMemory<float>& offset,
[all …]
|
| D | stream.cc | 315 const DeviceMemory<float> &x, const DeviceMemory<float> &scale, in ThenBatchNormalizationForward()
316 const DeviceMemory<float> &offset, in ThenBatchNormalizationForward()
317 const DeviceMemory<float> &estimated_mean, in ThenBatchNormalizationForward()
318 const DeviceMemory<float> &estimated_variance, in ThenBatchNormalizationForward()
321 DeviceMemory<float> *y, DeviceMemory<float> *batch_mean, in ThenBatchNormalizationForward()
322 DeviceMemory<float> *batch_var, DeviceMemory<float> *saved_mean, in ThenBatchNormalizationForward()
323 DeviceMemory<float> *saved_inv_var, bool is_training, in ThenBatchNormalizationForward()
324 std::function<const DeviceMemory<float> &()> var_to_inv_var, in ThenBatchNormalizationForward()
343 const DeviceMemory<float> &y_backprop, const DeviceMemory<float> &x, in ThenBatchNormalizationBackward()
344 const DeviceMemory<float> &scale, const DeviceMemory<float> &mean, in ThenBatchNormalizationBackward()
[all …]
|
| D | fft.h | 56 class DeviceMemory; variable
182 const DeviceMemory<std::complex<float>> &input,
183 DeviceMemory<std::complex<float>> *output) = 0;
185 const DeviceMemory<std::complex<double>> &input,
186 DeviceMemory<std::complex<double>> *output) = 0;
190 const DeviceMemory<float> &input,
191 DeviceMemory<std::complex<float>> *output) = 0;
193 const DeviceMemory<double> &input,
194 DeviceMemory<std::complex<double>> *output) = 0;
198 const DeviceMemory<std::complex<float>> &input,
[all …]
|
| D | device_memory.h | 110 class DeviceMemory final : public DeviceMemoryBase {
113 DeviceMemory() : DeviceMemoryBase(nullptr, 0) {} in DeviceMemory() function
114 DeviceMemory(std::nullptr_t) : DeviceMemory() {} in DeviceMemory() function
118 explicit DeviceMemory(const DeviceMemoryBase &other) in DeviceMemory() function
132 static DeviceMemory<ElemT> MakeFromByteSize(void *opaque, uint64 bytes) { in MakeFromByteSize()
133 return DeviceMemory<ElemT>(opaque, bytes); in MakeFromByteSize()
154 DeviceMemory(void *opaque, uint64 size) : DeviceMemoryBase(opaque, size) {} in DeviceMemory() function
223 const DeviceMemory<ElemT> &cref() const { return wrapped_; } in cref()
228 DeviceMemory<ElemT> *ptr() { return &wrapped_; } in ptr()
229 const DeviceMemory<ElemT> *ptr() const { return &wrapped_; } in ptr()
[all …]
|
| D | stream_executor_pimpl.h | 113 DeviceMemory<T> AllocateArray(uint64 element_count);
124 DeviceMemory<T> AllocateScalar() { in AllocateScalar()
137 DeviceMemory<T> AllocateZeroed();
154 DeviceMemory<T> AllocateSubBuffer(DeviceMemory<T> *parent,
160 ScopedDeviceMemory<T> AllocateOwnedSubBuffer(DeviceMemory<T> *parent, in AllocateOwnedSubBuffer()
174 port::StatusOr<DeviceMemory<T>> GetSymbol(const string &symbol_name);
263 port::Status SynchronousMemcpyD2H(const DeviceMemory<T> &device_src, in SynchronousMemcpyD2H()
668 inline DeviceMemory<T> StreamExecutor::AllocateArray(uint64 element_count) { in AllocateArray()
671 return DeviceMemory<T>::MakeFromByteSize(opaque, bytes); in AllocateArray()
675 inline port::StatusOr<DeviceMemory<T>> StreamExecutor::GetSymbol( in GetSymbol()
[all …]
|
| D | rng.h | 30 class DeviceMemory; variable
55 DeviceMemory<float> *v) = 0;
57 DeviceMemory<double> *v) = 0;
59 DeviceMemory<std::complex<float>> *v) = 0;
61 DeviceMemory<std::complex<double>> *v) = 0;
66 DeviceMemory<float> *v) { in DoPopulateRandGaussian()
72 double stddev, DeviceMemory<double> *v) { in DoPopulateRandGaussian()
|
| D | temporary_device_memory.h | 113 DeviceMemory<T>* mutable_device_memory() { in mutable_device_memory()
115 return reinterpret_cast<DeviceMemory<T>*>( in mutable_device_memory()
120 const DeviceMemory<T>& device_memory() const { in device_memory()
122 return reinterpret_cast<const DeviceMemory<T>&>( in device_memory()
|
| /external/tensorflow/tensorflow/stream_executor/cuda/ |
| D | cuda_dnn.h | 67 const DeviceMemory<Eigen::half>& input_data,
69 const DeviceMemory<Eigen::half>& input_h_data,
71 const DeviceMemory<Eigen::half>& input_c_data,
72 const DeviceMemory<Eigen::half>& params,
74 DeviceMemory<Eigen::half>* output_data,
76 DeviceMemory<Eigen::half>* output_h_data,
78 DeviceMemory<Eigen::half>* output_c_data, bool is_training,
84 const DeviceMemory<float>& input_data,
86 const DeviceMemory<float>& input_h_data,
88 const DeviceMemory<float>& input_c_data,
[all …]
|
| D | cuda_blas.cc | 642 const DeviceMemory<float> &x, int incx, in DoBlasAsum()
643 DeviceMemory<float> *result) { in DoBlasAsum()
650 const DeviceMemory<double> &x, int incx, in DoBlasAsum()
651 DeviceMemory<double> *result) { in DoBlasAsum()
658 const DeviceMemory<std::complex<float>> &x, int incx, in DoBlasAsum()
659 DeviceMemory<float> *result) { in DoBlasAsum()
666 const DeviceMemory<std::complex<double>> &x, int incx, in DoBlasAsum()
667 DeviceMemory<double> *result) { in DoBlasAsum()
674 const DeviceMemory<float> &x, int incx, in DoBlasAxpy()
675 DeviceMemory<float> *y, int incy) { in DoBlasAxpy()
[all …]
|
| D | cuda_dnn.cc | 990 DeviceMemory<uint8> state_memory; in CudnnDropoutDescriptor()
1426 const DeviceMemory<T>& input_data, in ExtractAndCheckRnnForward()
1428 const DeviceMemory<T>& input_h_data, in ExtractAndCheckRnnForward()
1430 const DeviceMemory<T>& input_c_data, const DeviceMemory<T>& params, in ExtractAndCheckRnnForward()
1432 const DeviceMemory<T>& output_data, in ExtractAndCheckRnnForward()
1434 const DeviceMemory<T>& output_h_data, in ExtractAndCheckRnnForward()
1436 const DeviceMemory<T>& output_c_data, RnnModelDims* model_dims) { in ExtractAndCheckRnnForward()
1504 DeviceMemory<uint8>* workspace) { in CreateRnnWorkspace()
1525 *workspace = DeviceMemory<uint8>(); in CreateRnnWorkspace()
1536 const DeviceMemory<T>& input_data, in DoRnnForwardImpl()
[all …]
|
| D | cuda_blas.h | 114 const port::ArraySlice<DeviceMemory<T> *> &a_array, int lda,
115 const port::ArraySlice<DeviceMemory<T> *> &b_array, int ldb, T beta,
116 const port::ArraySlice<DeviceMemory<T> *> &c_array, int ldc,
128 uint64 n, uint64 k, const CompT &alpha, const DeviceMemory<InT> &a,
129 int lda, const DeviceMemory<InT> &b, int ldb, const CompT &beta,
130 DeviceMemory<OutT> *c, int ldc, blas::ComputationType computation_type,
138 uint64 n, uint64 k, const ParamType &alpha, const DeviceMemory<T> &a,
139 int lda, const DeviceMemory<T> &b, int ldb, const ParamType &beta,
140 DeviceMemory<T> *c, int ldc, blas::ProfileResult *output_profile_result);
146 const DeviceMemory<T> &a, int lda,
[all …]
|
| D | cuda_rng.h | 32 class DeviceMemory; variable
58 bool DoPopulateRandUniform(Stream *stream, DeviceMemory<float> *v) override;
59 bool DoPopulateRandUniform(Stream *stream, DeviceMemory<double> *v) override;
61 DeviceMemory<std::complex<float>> *v) override;
63 DeviceMemory<std::complex<double>> *v) override;
65 DeviceMemory<float> *v) override;
67 DeviceMemory<double> *v) override;
75 bool DoPopulateRandUniformInternal(Stream *stream, DeviceMemory<T> *v);
78 DeviceMemory<ElemT> *v, FuncT func);
|
| D | cuda_rng.cc | 153 DeviceMemory<T> *v) { in DoPopulateRandUniformInternal()
190 bool CUDARng::DoPopulateRandUniform(Stream *stream, DeviceMemory<float> *v) { in DoPopulateRandUniform()
194 bool CUDARng::DoPopulateRandUniform(Stream *stream, DeviceMemory<double> *v) { in DoPopulateRandUniform()
199 DeviceMemory<std::complex<float>> *v) { in DoPopulateRandUniform()
204 DeviceMemory<std::complex<double>> *v) { in DoPopulateRandUniform()
211 DeviceMemory<ElemT> *v, in DoPopulateRandGaussianInternal()
233 DeviceMemory<float> *v) { in DoPopulateRandGaussian()
239 DeviceMemory<double> *v) { in DoPopulateRandGaussian()
|
| D | cuda_fft.h | 90 DeviceMemory<uint8> scratch_;
122 const DeviceMemory<InputT> &input,
123 DeviceMemory<OutputT> *output);
129 const DeviceMemory<InputT> &input,
130 DeviceMemory<OutputT> *output);
|
| D | cuda_fft.cc | 465 const DeviceMemory<InputT> &input, in DoFftInternal()
466 DeviceMemory<OutputT> *output) { in DoFftInternal()
492 const DeviceMemory<InputT> &input, in DoFftWithDirectionInternal()
493 DeviceMemory<OutputT> *output) { in DoFftWithDirectionInternal()
520 const DeviceMemory<std::complex<__type>> &input, \
521 DeviceMemory<std::complex<__type>> *output) { \
526 const DeviceMemory<__type> &input, \
527 DeviceMemory<std::complex<__type>> *output) { \
532 const DeviceMemory<std::complex<__type>> &input, \
533 DeviceMemory<__type> *output) { \
|
| D | cuda_helpers.h | 34 class DeviceMemory; variable
42 const T *CUDAMemory(const DeviceMemory<T> &mem) { in CUDAMemory()
49 T *CUDAMemoryMutable(DeviceMemory<T> *mem) { in CUDAMemoryMutable()
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | cudnn_batchnorm_thunk.cc | 109 se::DeviceMemory<float> output(buffer_allocations.GetDeviceAddress(output_)); in ExecuteOnStream()
111 se::DeviceMemory<float>(buffer_allocations.GetDeviceAddress(operand_)), in ExecuteOnStream()
112 se::DeviceMemory<float>(buffer_allocations.GetDeviceAddress(scale_)), in ExecuteOnStream()
113 se::DeviceMemory<float>(buffer_allocations.GetDeviceAddress(offset_)), in ExecuteOnStream()
114 se::DeviceMemory<float>(buffer_allocations.GetDeviceAddress(mean_)), in ExecuteOnStream()
115 se::DeviceMemory<float>(buffer_allocations.GetDeviceAddress(variance_)), in ExecuteOnStream()
171 se::DeviceMemory<float> output_data( in ExecuteOnStream()
173 se::DeviceMemory<float> output_mean( in ExecuteOnStream()
175 se::DeviceMemory<float> output_inv_stddev( in ExecuteOnStream()
178 se::DeviceMemory<float> null_device_ptr(nullptr); in ExecuteOnStream()
[all …]
|
| D | cudnn_convolution_runner.cc | 27 using se::DeviceMemory;
52 se::port::StatusOr<DeviceMemory<uint8>> AllocateBytes( in AllocateBytes()
65 return se::DeviceMemory<uint8>(scratch_); in AllocateBytes()
76 const Shape& output_shape, DeviceMemory<T> input_buf, in RunCudnnConvolution()
77 DeviceMemory<T> filter_buf, DeviceMemory<T> output_buf, in RunCudnnConvolution()
249 se::DeviceMemory<float>(input_buf), se::DeviceMemory<float>(filter_buf), in RunCudnnConvolution()
250 se::DeviceMemory<float>(output_buf), scratch_allocator, window, dnums, in RunCudnnConvolution()
254 se::DeviceMemory<Eigen::half>(input_buf), in RunCudnnConvolution()
255 se::DeviceMemory<Eigen::half>(filter_buf), in RunCudnnConvolution()
256 se::DeviceMemory<Eigen::half>(output_buf), in RunCudnnConvolution()
|
| D | fft_thunk.cc | 52 se::port::StatusOr<se::DeviceMemory<uint8>> FftScratchAllocator::AllocateBytes( in AllocateBytes()
74 return se::DeviceMemory<uint8>(allocated_buffer); in AllocateBytes()
175 se::DeviceMemory<complex64> input_data( in ExecuteOnStream()
177 se::DeviceMemory<complex64> output_data( in ExecuteOnStream()
184 se::DeviceMemory<complex64> input_data( in ExecuteOnStream()
186 se::DeviceMemory<complex64> output_data( in ExecuteOnStream()
200 se::DeviceMemory<float> input_data( in ExecuteOnStream()
202 se::DeviceMemory<complex64> output_data( in ExecuteOnStream()
209 se::DeviceMemory<complex64> input_data( in ExecuteOnStream()
211 se::DeviceMemory<float> output_data( in ExecuteOnStream()
|
| D | gemm_thunk.cc | 55 se::DeviceMemory<Element> lhs_data(lhs_matrix.data); in DoGemm()
56 se::DeviceMemory<Element> rhs_data(rhs_matrix.data); in DoGemm()
57 se::DeviceMemory<Element> output_data(output_matrix.data); in DoGemm()
98 se::DeviceMemory<Element> lhs_data(lhs_matrix.data); in DoGemmWithAlgorithm()
99 se::DeviceMemory<Element> rhs_data(rhs_matrix.data); in DoGemmWithAlgorithm()
100 se::DeviceMemory<Element> output_data(output_matrix.data); in DoGemmWithAlgorithm()
|
| D | convolution_thunk.h | 76 perftools::gputools::DeviceMemory<float> input_data,
78 perftools::gputools::DeviceMemory<float> filter_data,
80 perftools::gputools::DeviceMemory<float> output_data,
|
| /external/tensorflow/tensorflow/core/common_runtime/gpu/ |
| D | gpu_debug_allocator_test.cc | 46 gpu::DeviceMemory<int64> gpu_array_ptr{gpu::DeviceMemoryBase{gpu_array}}; in TEST()
71 gpu::DeviceMemory<int64> gpu_array_ptr{ in TEST()
76 gpu::DeviceMemory<int64> gpu_hdr_ptr{ in TEST()
104 gpu::DeviceMemory<int64> gpu_array_ptr{ in TEST()
110 gpu::DeviceMemory<int64> gpu_ftr_ptr{ in TEST()
134 gpu::DeviceMemory<float> gpu_array_ptr{gpu::DeviceMemoryBase{gpu_array}}; in TEST()
177 gpu::DeviceMemory<float> gpu_array_ptr{gpu::DeviceMemoryBase{gpu_array}}; in TEST()
|
| /external/tensorflow/tensorflow/contrib/cudnn_rnn/kernels/ |
| D | cudnn_rnn_ops.cc | 104 using perftools::gputools::DeviceMemory;
178 const DeviceMemory<T> AsDeviceMemory(const Tensor* tensor) { in AsDeviceMemory()
179 return DeviceMemory<T>::MakeFromByteSize( in AsDeviceMemory()
185 DeviceMemory<T> AsDeviceMemory(Tensor* tensor) { in AsDeviceMemory()
186 return DeviceMemory<T>::MakeFromByteSize( in AsDeviceMemory()
192 DeviceMemory<U> CastDeviceMemory(Tensor* tensor) { in CastDeviceMemory()
193 return DeviceMemory<U>::MakeFromByteSize( in CastDeviceMemory()
240 StatusOr<DeviceMemory<uint8>> AllocateBytes( in AllocateBytes()
252 return StatusOr<DeviceMemory<uint8>>( in AllocateBytes()
276 StatusOr<DeviceMemory<uint8>> AllocateBytes( in AllocateBytes()
[all …]
|