| /external/tensorflow/tensorflow/stream_executor/cuda/ |
| cuda_dnn.h | 16 // The CUDA-specific DNN library support, implementing the general DnnSupport
22 #include "tensorflow/stream_executor/dnn.h"
41 // cudnn-library based DNN support. For details on overridden interface
42 // functions, see dnn.h.
43 class CudnnSupport : public dnn::DnnSupport {
50 port::StatusOr<std::unique_ptr<dnn::RnnDescriptor>> createRnnDescriptor(
52 dnn::RnnInputMode input_mode, dnn::RnnDirectionMode direction_mode,
53 dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout
[all...] |
| cuda_dnn.cc | 30 #include "tensorflow/stream_executor/dnn.h"
72 using dnn::BatchDescriptor;
73 using dnn::FilterDescriptor;
74 using dnn::ConvolutionDescriptor;
75 using dnn::PoolingDescriptor;
76 using dnn::NormalizeDescriptor;
294 cudnnConvolutionFwdAlgo_t ToConvForwardAlgo(dnn::AlgorithmDesc algorithm) {
318 dnn::AlgorithmDesc algorithm) {
341 dnn::AlgorithmDesc algorithm) {
442 case dnn::DataLayout::kBatchYXDepth
[all...] |
| /external/tensorflow/tensorflow/stream_executor/ |
| host_buffer.h | 19 #include "tensorflow/stream_executor/dnn.h"
25 // dnn::BatchDescriptor using a device-dependent memory layout.
30 const dnn::BatchDescriptor& descriptor() const { return descriptor_; }
36 // Construct a HostBuffer from the supplied dnn::BatchDescriptor.
37 explicit HostBuffer(const dnn::BatchDescriptor& descriptor)
42 const dnn::BatchDescriptor descriptor_;
|
| dnn.h | 47 namespace dnn { namespace in namespace:perftools::gputools
332 port::ArraySlice<dnn::BatchDescriptor> inputs);
701 // Collects parameters for DNN algorithms
924 const dnn::BatchDescriptor& x_desc
[all...] |
| stream.h | 30 #include "tensorflow/stream_executor/dnn.h"
65 namespace dnn { namespace in namespace:perftools::gputools
74 } // namespace dnn
217 // DNN support
226 const dnn::BatchDescriptor &x_desc,
227 const dnn::BatchDescriptor &scale_offset_desc, const double epsilon,
237 const DeviceMemory<float> &inv_var, const dnn::BatchDescriptor &x_desc,
238 const dnn::BatchDescriptor &scale_offset_desc, const double epsilon,
247 const dnn::BatchDescriptor &x_desc,
248 const dnn::BatchDescriptor &scale_offset_desc, const double epsilon
[all...] |
| plugin.cc | 39 PluginConfig& PluginConfig::SetDnn(PluginId dnn) {
40 dnn_ = dnn;
|
| plugin.h | 51 // described therein will be used to provide BLAS, DNN, FFT, and RNG
73 PluginConfig& SetDnn(PluginId dnn);
78 PluginId dnn() const { return dnn_; } function in class:perftools::gputools::PluginConfig
|
| stream.cc | 38 string ToVlogString(const dnn::BatchDescriptor &descriptor) {
42 string ToVlogString(const dnn::FilterDescriptor &descriptor) {
46 string ToVlogString(const dnn::ConvolutionDescriptor &descriptor) {
50 string ToVlogString(const dnn::PoolingDescriptor &descriptor) {
54 string ToVlogString(const dnn::NormalizeDescriptor &descriptor) {
58 string ToVlogString(dnn::ActivationMode mode) {
59 return dnn::ActivationModeString(mode);
62 string ToVlogString(const dnn::AlgorithmConfig &algo_config) {
66 string ToVlogString(dnn::ElementwiseOperation op) {
67 return dnn::ElementwiseOperationString(op)
[all...] |
| plugin_registry.h | 22 #include "tensorflow/stream_executor/dnn.h"
40 // BLAS, DNN, FFT, and RNG. Each interface is defined in the corresponding
54 typedef dnn::DnnSupport* (*DnnFactory)(internal::StreamExecutorInterface*);
108 std::map<PluginId, DnnFactory> dnn; member in struct:perftools::gputools::PluginRegistry::PluginFactories
117 PluginId blas, dnn, fft, rng; member in struct:perftools::gputools::PluginRegistry::DefaultFactories
|
| stream_executor_pimpl.h | 354 std::vector<dnn::AlgorithmDesc> *out_algorithms);
359 std::vector<dnn::AlgorithmDesc> *out_algorithms);
365 std::vector<dnn::AlgorithmDesc> *out_algorithms);
372 port::StatusOr<std::unique_ptr<dnn::RnnDescriptor>> createRnnDescriptor(
374 dnn::RnnInputMode input_mode, dnn::RnnDirectionMode direction_mode,
375 dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout,
380 port::StatusOr<std::unique_ptr<dnn::RnnSequenceTensorDescriptor>>
382 int data_size, dnn::DataType data_type)
[all...] |
| plugin_registry.cc | 33 return "DNN";
45 blas(kNullPlugin), dnn(kNullPlugin), fft(kNullPlugin), rng(kNullPlugin) { }
129 default_factories_[platform_id].dnn = plugin_id;
153 return factories.dnn.find(plugin_id) != factories.dnn.end();
243 EMIT_PLUGIN_SPECIALIZATIONS(DnnFactory, dnn, "DNN");
|
| stream_executor_pimpl.cc | 297 std::vector<dnn::AlgorithmDesc> *out_algorithms) {
298 dnn::DnnSupport *dnn_support = AsDnn();
310 std::vector<dnn::AlgorithmDesc> *out_algorithms) {
311 dnn::DnnSupport *dnn_support = AsDnn();
323 std::vector<dnn::AlgorithmDesc> *out_algorithms) {
324 dnn::DnnSupport *dnn_support = AsDnn();
343 port::StatusOr<std::unique_ptr<dnn::RnnDescriptor>>
346 dnn::RnnInputMode input_mode, dnn::RnnDirectionMode direction_mode,
347 dnn::RnnMode rnn_mode, dnn::DataType data_type, float dropout, uint64 seed
[all...] |
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| convolution_thunk.h | 40 // Constructs a thunk for launching a DNN convolution. When run, it will
75 const perftools::gputools::dnn::BatchDescriptor& input_descriptor,
77 const perftools::gputools::dnn::FilterDescriptor& filter_descriptor,
79 const perftools::gputools::dnn::BatchDescriptor& output_descriptor,
81 const perftools::gputools::dnn::ConvolutionDescriptor&
83 const perftools::gputools::dnn::AlgorithmConfig& algorithm_config,
85 perftools::gputools::dnn::ProfileResult* profile_result);
|
| convolution_thunk.cc | 33 using se::dnn::AlgorithmDesc;
34 using se::dnn::BatchDescriptor;
35 using se::dnn::ConvolutionDescriptor;
36 using se::dnn::DataLayout;
37 using se::dnn::FilterDescriptor;
38 using se::dnn::FilterLayout;
75 se::dnn::AlgorithmConfig algorithm_config(
76 se::dnn::AlgorithmDesc(algorithm_, tensor_ops_enabled_));
|
| cudnn_convolution_runner.h | 80 perftools::gputools::dnn::AlgorithmConfig algorithm,
82 perftools::gputools::dnn::ProfileResult* profile_result = nullptr);
91 perftools::gputools::dnn::AlgorithmConfig algorithm,
93 perftools::gputools::dnn::ProfileResult* profile_result = nullptr);
|
| cudnn_batchnorm_thunk.cc | 32 namespace dnn = se::dnn;
34 static std::pair<dnn::BatchDescriptor /*input_desc*/,
35 dnn::BatchDescriptor /*scale_offset_desc*/>
62 dnn::BatchDescriptor input_desc;
63 input_desc.set_layout(dnn::DataLayout::kBatchDepthYX)
69 dnn::BatchDescriptor scale_offset_desc;
70 scale_offset_desc.set_layout(dnn::DataLayout::kBatchDepthYX)
104 dnn::BatchDescriptor operand_desc;
105 dnn::BatchDescriptor scale_offset_desc
[all...] |
| cudnn_convolution_runner.cc | 30 using se::dnn::AlgorithmConfig;
31 using se::dnn::BatchDescriptor;
32 using se::dnn::ConvolutionDescriptor;
33 using se::dnn::DataLayout;
34 using se::dnn::DimIndex;
35 using se::dnn::FilterDescriptor;
36 using se::dnn::FilterLayout;
37 using se::dnn::ProfileResult;
223 perftools::gputools::dnn::AlgorithmConfig algorithm,
225 perftools::gputools::dnn::ProfileResult* profile_result)
[all...] |
| /external/tensorflow/tensorflow/core/kernels/ |
| cudnn_pooling_gpu.h | 41 perftools::gputools::dnn::PoolingMode pooling_mode,
55 perftools::gputools::dnn::PoolingMode pooling_mode,
|
| pooling_ops_common_gpu.h | 43 perftools::gputools::dnn::PoolingMode pooling_mode,
58 perftools::gputools::dnn::PoolingMode pooling_mode,
|
| cudnn_pooling_gpu.cc | 36 perftools::gputools::dnn::PoolingMode pooling_mode,
70 perftools::gputools::dnn::PoolingDescriptor pooling_desc(3);
72 perftools::gputools::dnn::BatchDescriptor input_desc(3);
75 .set_layout(perftools::gputools::dnn::DataLayout::kBatchDepthYX);
76 perftools::gputools::dnn::BatchDescriptor output_desc(3);
79 .set_layout(perftools::gputools::dnn::DataLayout::kBatchDepthYX);
81 const auto dim_i = static_cast<perftools::gputools::dnn::DimIndex>(i);
119 perftools::gputools::dnn::PoolingMode pooling_mode,
125 CHECK((pooling_mode != perftools::gputools::dnn::PoolingMode::kMaximum) ||
189 perftools::gputools::dnn::PoolingDescriptor pooling_desc(3)
[all...] |
| pooling_ops_common.cc | 143 perftools::gputools::dnn::PoolingMode pooling_mode,
187 perftools::gputools::dnn::PoolingDescriptor pooling_desc;
197 perftools::gputools::dnn::BatchDescriptor input_desc;
202 .set_layout(perftools::gputools::dnn::DataLayout::kBatchDepthYX);
204 perftools::gputools::dnn::BatchDescriptor output_desc;
209 .set_layout(perftools::gputools::dnn::DataLayout::kBatchDepthYX);
240 perftools::gputools::dnn::PoolingMode pooling_mode,
245 CHECK((pooling_mode != perftools::gputools::dnn::PoolingMode::kMaximum) ||
330 perftools::gputools::dnn::PoolingDescriptor pooling_desc;
340 perftools::gputools::dnn::BatchDescriptor orig_output_desc
[all...] |
| /external/tensorflow/tensorflow/contrib/learn/python/learn/estimators/ |
| nonlinear_test.py | 25 from tensorflow.contrib.learn.python.learn.estimators import dnn
41 classifier = dnn.DNNClassifier(
49 classifier.get_variable_value("dnn/hiddenlayer_0/weights").shape,
52 classifier.get_variable_value("dnn/hiddenlayer_1/weights").shape,
55 classifier.get_variable_value("dnn/hiddenlayer_2/weights").shape,
58 classifier.get_variable_value("dnn/logits/weights").shape, (10, 3))
59 self.assertIn("dnn/hiddenlayer_0/biases", variable_names)
60 self.assertIn("dnn/hiddenlayer_1/biases", variable_names)
61 self.assertIn("dnn/hiddenlayer_2/biases", variable_names)
62 self.assertIn("dnn/logits/biases", variable_names
[all...] |
| dnn_test.py | 31 from tensorflow.contrib.learn.python.learn.estimators import dnn
80 dnn._dnn_model_fn(features, labels, model_fn.ModeKeys.TRAIN, params)
114 model_ops = dnn._dnn_model_fn(features, labels, model_fn.ModeKeys.TRAIN,
118 embedding_language, 'dnn', 'dnn/input_from_feature_columns')
120 embedding_wire, 'dnn', 'dnn/input_from_feature_columns')
150 return dnn._dnn_model_fn(features, labels, model_fn.ModeKeys.TRAIN, params)
169 estimator=dnn.DNNClassifier(
181 estimator_test_utils.assert_estimator_contract(self, dnn.DNNEstimator
[all...] |
| /external/tensorflow/tensorflow/contrib/fused_conv/kernels/ |
| fused_conv2d_bias_activation_op.cc | 250 namespace dnn = ::perftools::gputools::dnn;
257 dnn::AlgorithmConfig>
439 constexpr auto data_layout = is_int8x4 ? dnn::DataLayout::kBatchDepthYX4
440 : dnn::DataLayout::kBatchDepthYX;
441 constexpr auto filter_layout = is_int8x4 ? dnn::FilterLayout::kOutputInputYX4
442 : dnn::FilterLayout::kOutputInputYX;
444 dnn::BatchDescriptor conv_input_desc;
450 dnn::FilterDescriptor filter_desc;
456 dnn::BatchDescriptor side_input_desc
[all...] |
| /external/tensorflow/tensorflow/python/estimator/ |
| estimator_lib.py | 24 from tensorflow.python.estimator.canned.dnn import DNNClassifier
25 from tensorflow.python.estimator.canned.dnn import DNNRegressor
|
