| /external/tensorflow/tensorflow/compiler/xla/client/ |
| client_library.h | 46 LocalClientOptions(perftools::gputools::Platform* platform = nullptr,
51 LocalClientOptions& set_platform(perftools::gputools::Platform* platform);
52 perftools::gputools::Platform* platform() const;
64 perftools::gputools::Platform* platform_;
77 perftools::gputools::Platform* platform = nullptr);
87 static LocalService* GetXlaService(perftools::gputools::Platform* platform);
94 perftools::gputools::Platform* platform = nullptr);
123 std::unordered_map<perftools::gputools::Platform::Id,
127 std::unordered_map<perftools::gputools::Platform::Id,
|
| /external/tensorflow/tensorflow/compiler/xla/service/ |
| device_memory_allocator.cc | 27 const perftools::gputools::Platform* platform,
28 tensorflow::gtl::ArraySlice<perftools::gputools::StreamExecutor*>
33 StatusOr<perftools::gputools::DeviceMemoryBase>
36 TF_ASSIGN_OR_RETURN(perftools::gputools::StreamExecutor * stream_executor,
38 perftools::gputools::DeviceMemoryBase result =
50 int device_ordinal, perftools::gputools::DeviceMemoryBase* mem) {
52 TF_ASSIGN_OR_RETURN(perftools::gputools::StreamExecutor * stream_executor,
58 perftools::gputools::DeviceMemoryBase mem_copy(*mem);
64 StatusOr<perftools::gputools::StreamExecutor*>
|
| backend.h | 47 BackendOptions& set_platform(perftools::gputools::Platform* platform);
48 perftools::gputools::Platform* platform() const;
57 perftools::gputools::Platform* platform_ = nullptr;
69 using StreamPtr = Pool<perftools::gputools::Stream>::SmartPtr;
82 perftools::gputools::Platform* platform() const { return platform_; }
99 const std::vector<perftools::gputools::StreamExecutor*>& stream_executors()
105 StatusOr<perftools::gputools::StreamExecutor*> stream_executor(
111 perftools::gputools::StreamExecutor* default_stream_executor() const {
121 perftools::gputools::StreamExecutor* executor);
160 Backend(perftools::gputools::Platform* platform, Compiler* compiler
[all...] |
| shaped_buffer.h | 44 const perftools::gputools::Platform* platform,
55 const perftools::gputools::Platform* platform() const { return platform_; }
59 const perftools::gputools::DeviceMemoryBase& root_buffer() const {
65 const perftools::gputools::DeviceMemoryBase& buffer(
71 void set_buffer(const perftools::gputools::DeviceMemoryBase& buffer,
78 const ShapeTree<perftools::gputools::DeviceMemoryBase>& buffers() const {
81 ShapeTree<perftools::gputools::DeviceMemoryBase>& buffers() {
104 const perftools::gputools::Platform* platform_;
110 ShapeTree<perftools::gputools::DeviceMemoryBase> buffers_;
|
| compiler.h | 73 virtual perftools::gputools::Platform::Id PlatformId() const = 0;
108 virtual perftools::gputools::Platform::Id PlatformId() const = 0;
120 perftools::gputools::StreamExecutor* executor,
137 perftools::gputools::StreamExecutor* executor,
150 std::vector<std::vector<perftools::gputools::StreamExecutor*>>
171 perftools::gputools::Platform::Id platform_id,
177 const perftools::gputools::Platform* platform);
197 static std::map<perftools::gputools::Platform::Id, CompilerFactory>*
202 static std::map<perftools::gputools::Platform::Id, std::unique_ptr<Compiler>>*
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| buffer_allocations.h | 44 perftools::gputools::DeviceMemoryBase address);
55 std::map<BufferAllocation::Index, perftools::gputools::DeviceMemoryBase>
68 perftools::gputools::DeviceMemoryBase GetDeviceAddress(
73 perftools::gputools::DeviceMemoryBase GetDeviceAddress(
76 perftools::gputools::DeviceMemoryBase GetTempBufferBase() const {
83 const std::set<perftools::gputools::DeviceMemoryBase>& live_addresses,
95 perftools::gputools::DeviceMemoryBase buffer);
100 std::vector<perftools::gputools::DeviceMemoryBase> buffers_;
103 perftools::gputools::DeviceMemoryBase temp_buffer_base_;
|
| fft_thunk.h | 37 class FftScratchAllocator : public perftools::gputools::ScratchAllocator {
44 int64 GetMemoryLimitInBytes(perftools::gputools::Stream* stream) override;
48 perftools::gputools::port::StatusOr<perftools::gputools::DeviceMemory<uint8>>
49 AllocateBytes(perftools::gputools::Stream* stream, int64 byte_size) override;
54 std::vector<perftools::gputools::DeviceMemoryBase> allocated_buffers_;
78 perftools::gputools::Stream* stream) override;
81 const perftools::gputools::fft::Type fft_type_;
86 std::unique_ptr<perftools::gputools::fft::Plan> fft_plan_;
|
| gpu_transfer_manager.h | 39 Status TransferLiteralToInfeed(perftools::gputools::StreamExecutor* executor,
41 Status TransferBufferToInfeed(perftools::gputools::StreamExecutor* executor,
48 perftools::gputools::StreamExecutor* executor, int64 size,
53 Status EnqueueBuffersToInfeed(perftools::gputools::StreamExecutor* executor,
|
| convolution_thunk.h | 69 perftools::gputools::Stream* stream) override;
75 const perftools::gputools::dnn::BatchDescriptor& input_descriptor,
76 perftools::gputools::DeviceMemory<float> input_data,
77 const perftools::gputools::dnn::FilterDescriptor& filter_descriptor,
78 perftools::gputools::DeviceMemory<float> filter_data,
79 const perftools::gputools::dnn::BatchDescriptor& output_descriptor,
80 perftools::gputools::DeviceMemory<float> output_data,
81 const perftools::gputools::dnn::ConvolutionDescriptor&
83 const perftools::gputools::dnn::AlgorithmConfig& algorithm_config,
84 perftools::gputools::Stream* stream, ScratchAllocator* scratch_allocator
[all...] |
| infeed_manager.h | 49 InfeedBuffer(perftools::gputools::StreamExecutor* executor, int64 length)
63 perftools::gputools::DeviceMemoryBase* device_memory() {
68 perftools::gputools::StreamExecutor* executor_; // Not owned.
70 perftools::gputools::DeviceMemoryBase device_memory_;
103 perftools::gputools::Stream* GetStream(
104 perftools::gputools::StreamExecutor* executor);
124 std::unique_ptr<perftools::gputools::Stream> host_to_device_stream_;
127 perftools::gputools::StreamExecutor* host_to_device_executor_;
|
| kernel_thunk.h | 63 perftools::gputools::Stream* stream) override;
79 std::unique_ptr<perftools::gputools::MultiKernelLoaderSpec> loader_spec_
83 std::unordered_map<perftools::gputools::StreamExecutor*,
84 perftools::gputools::KernelBase>
|
| copy_thunk.cc | 34 perftools::gputools::Stream* stream) {
35 perftools::gputools::DeviceMemoryBase destination_data =
52 perftools::gputools::Stream* stream) {
53 perftools::gputools::DeviceMemoryBase destination_data =
55 perftools::gputools::DeviceMemoryBase source_data =
|
| gemm_thunk.h | 53 perftools::gputools::Stream* stream) override;
59 perftools::gputools::Stream* stream) override {
81 StatusOr<::perftools::gputools::blas::AlgorithmType>>
|
| /external/tensorflow/tensorflow/core/util/ |
| stream_executor_util.h | 33 static perftools::gputools::DeviceMemory<T> AsDeviceMemory(const Tensor& t) {
35 return perftools::gputools::DeviceMemory<T>(
36 perftools::gputools::DeviceMemoryBase(ptr, t.TotalBytes()));
43 static Status ConvertStatus(const perftools::gputools::port::Status& s) {
|
| /external/tensorflow/tensorflow/stream_executor/lib/ |
| array_slice.h | 22 namespace gputools { namespace in namespace:perftools
29 } // namespace gputools
|
| strcat.h | 16 // IWYU pragma: private, include "perftools/gputools/executor/stream_executor.h"
24 namespace gputools { namespace in namespace:perftools
31 } // namespace gputools
|
| stringprintf.h | 22 namespace gputools { namespace in namespace:perftools
29 } // namespace gputools
|
| /external/tensorflow/tensorflow/compiler/xla/service/cpu/ |
| cpu_transfer_manager.h | 40 Status TransferLiteralToInfeed(perftools::gputools::StreamExecutor* executor,
42 Status TransferBufferToInfeed(perftools::gputools::StreamExecutor* executor,
45 perftools::gputools::StreamExecutor* executor, const Shape& literal_shape,
52 perftools::gputools::StreamExecutor* executor, int64 size,
57 perftools::gputools::StreamExecutor* executor,
62 perftools::gputools::StreamExecutor* executor, void* destination,
69 perftools::gputools::StreamExecutor* executor,
|
| /external/tensorflow/tensorflow/compiler/xla/ |
| device_util.h | 32 string DeviceIdentifier(perftools::gputools::StreamExecutor* stream_exec) {
|
| /external/tensorflow/tensorflow/stream_executor/ |
| stream_executor_internal.cc | 19 namespace gputools { namespace in namespace:perftools
41 } // namespace gputools
|
| /external/tensorflow/tensorflow/contrib/nccl/kernels/ |
| nccl_manager.h | 59 perftools::gputools::StreamExecutor* executor,
61 perftools::gputools::Stream* tensor_stream,
68 perftools::gputools::StreamExecutor* executor,
70 perftools::gputools::Stream* tensor_stream,
73 perftools::gputools::StreamExecutor* executor,
75 perftools::gputools::Stream* tensor_stream,
82 perftools::gputools::StreamExecutor* executor,
84 perftools::gputools::Stream* tensor_stream,
88 perftools::gputools::StreamExecutor* executor,
90 perftools::gputools::Stream* tensor_stream
[all...] |
| /external/tensorflow/tensorflow/contrib/rnn/kernels/ |
| blas_gemm.cc | 29 perftools::gputools::DeviceMemory<T> AsDeviceMemory(const T* cuda_memory) {
30 perftools::gputools::DeviceMemoryBase wrapped(const_cast<T*>(cuda_memory));
31 perftools::gputools::DeviceMemory<T> typed(wrapped);
44 perftools::gputools::blas::Transpose trans[] = {
45 perftools::gputools::blas::Transpose::kNoTranspose,
46 perftools::gputools::blas::Transpose::kTranspose};
|
| /external/tensorflow/tensorflow/core/common_runtime/gpu/ |
| gpu_event_mgr.h | 33 namespace gputools { namespace in namespace:perftools
37 } // namespace gputools
49 EventMgr(perftools::gputools::StreamExecutor* se,
56 void ThenDeleteTensors(perftools::gputools::Stream* stream,
70 inline void ThenDeleteBuffer(perftools::gputools::Stream* stream,
81 inline void ThenExecute(perftools::gputools::Stream* stream,
94 perftools::gputools::StreamExecutor* const exec_;
104 perftools::gputools::Event* event;
136 void QueueInUse(perftools::gputools::Stream* stream, InUse in_use)
139 void QueueTensors(perftools::gputools::Stream* stream
[all...] |
| /external/tensorflow/tensorflow/compiler/jit/ |
| xla_device_context.h | 52 explicit XlaTransferManager(perftools::gputools::Stream* stream);
59 perftools::gputools::Stream* stream() const { return stream_; }
64 perftools::gputools::Stream* stream_;
72 explicit XlaDeviceContext(perftools::gputools::Stream* stream);
80 perftools::gputools::Stream* stream() const override {
|
| /external/tensorflow/tensorflow/core/grappler/ |
| devices.cc | 33 perftools::gputools::Platform* gpu_manager = GPUMachineManager();
39 perftools::gputools::StreamExecutor* se = exec_status.ValueOrDie();
40 const perftools::gputools::DeviceDescription& desc =
59 perftools::gputools::Platform* gpu_platform = GPUMachineManager();
61 perftools::gputools::StreamExecutor* se =
|
