| /external/tensorflow/tensorflow/stream_executor/ |
| D | stream_executor_pimpl.cc | 75 ScopedTracer(StreamExecutor *stream_exec, BeginCallT begin_call, in ScopedTracer()
107 StreamExecutor *stream_exec_;
116 MakeScopedTracer(StreamExecutor *stream_exec, BeginCallT begin_call, in MakeScopedTracer()
128 /* static */ absl::Mutex StreamExecutor::static_mu_{absl::kConstInit};
139 StreamExecutor::StreamExecutor( in StreamExecutor() function in stream_executor::StreamExecutor
167 StreamExecutor::~StreamExecutor() { in ~StreamExecutor()
186 port::Status StreamExecutor::Init(DeviceOptions device_options) { in Init()
190 port::Status StreamExecutor::Init() { return Init(DeviceOptions::Default()); } in Init()
192 port::Status StreamExecutor::GetKernel(const MultiKernelLoaderSpec &spec, in GetKernel()
197 void StreamExecutor::UnloadKernel(const KernelBase *kernel) { in UnloadKernel()
[all …]
|
| D | stream_test.cc | 25 std::unique_ptr<StreamExecutor> NewStreamExecutor() { in NewStreamExecutor()
34 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
40 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
47 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
68 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
99 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
138 std::unique_ptr<StreamExecutor> executor = NewStreamExecutor(); in TEST_F()
|
| D | executor_cache.h | 38 typedef port::StatusOr<std::unique_ptr<StreamExecutor>> ExecutorFactory();
39 port::StatusOr<StreamExecutor*> GetOrCreate(
45 port::StatusOr<StreamExecutor*> Get(const StreamExecutorConfig& config);
65 std::pair<StreamExecutorConfig, std::unique_ptr<StreamExecutor>>>
|
| D | platform.h | 35 class StreamExecutor; variable
154 virtual port::StatusOr<StreamExecutor*> ExecutorForDevice(int ordinal) = 0;
159 virtual port::StatusOr<StreamExecutor*> ExecutorForDeviceWithPluginConfig(
164 virtual port::StatusOr<StreamExecutor*> GetExecutor(
170 virtual port::StatusOr<std::unique_ptr<StreamExecutor>> GetUncachedExecutor(
|
| D | stream_executor_pimpl.h | 73 class StreamExecutor {
75 StreamExecutor(
80 ~StreamExecutor();
315 port::Status EnablePeerAccessTo(StreamExecutor *other);
323 bool CanEnablePeerAccessTo(StreamExecutor *other);
750 SE_DISALLOW_COPY_AND_ASSIGN(StreamExecutor);
756 explicit ScopedModuleHandle(StreamExecutor *executor, in ScopedModuleHandle()
782 StreamExecutor *executor_;
793 StreamExecutor::CreateTypedKernel(absl::string_view kernel_name, in CreateTypedKernel()
810 inline DeviceMemory<T> StreamExecutor::AllocateArray(uint64 element_count, in AllocateArray()
[all …]
|
| D | timer.h | 29 class StreamExecutor; variable
41 explicit Timer(StreamExecutor *parent);
62 StreamExecutor *parent_;
|
| /external/tensorflow/tensorflow/compiler/xla/service/cpu/ |
| D | cpu_transfer_manager.h | 42 Status TransferLiteralToInfeed(se::StreamExecutor* executor,
44 Status TransferLiteralFromOutfeed(se::StreamExecutor* executor,
48 se::StreamExecutor* executor, in CanShapedBufferBeAccessedNow()
54 se::StreamExecutor* executor, in CanBufferBeAccessedNow()
60 Status TransferBufferToInfeed(se::StreamExecutor* executor, int64 size,
66 se::StreamExecutor* executor, int64 size, const void* source);
70 se::StreamExecutor* executor,
74 StatusOr<Shape> TransferArrayBufferFromOutfeed(se::StreamExecutor* executor,
82 se::StreamExecutor* executor,
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | gpu_compiler.h | 57 std::unique_ptr<HloModule> module, se::StreamExecutor* stream_exec,
63 se::StreamExecutor* executor, bool optimize,
67 se::StreamExecutor* stream_exec,
71 HloModule* hlo_module, se::StreamExecutor* stream_exec,
75 HloModule* hlo_module, se::StreamExecutor* stream_exec,
84 virtual GpuVersion GetGpuVersion(se::StreamExecutor* stream_exec) = 0;
91 se::StreamExecutor* stream_exec, bool relocatable,
97 std::unique_ptr<HloModule> module, se::StreamExecutor* stream_exec,
107 se::StreamExecutor* stream_exec, const CompileOptions& options,
130 se::StreamExecutor* stream_exec, in LinkModules()
[all …]
|
| D | gpu_transfer_manager.h | 41 Status TransferLiteralToInfeed(se::StreamExecutor* executor,
43 Status TransferLiteralFromOutfeed(se::StreamExecutor* executor,
50 se::StreamExecutor* executor, int64 size, const void* source);
54 Status EnqueueBuffersToInfeed(se::StreamExecutor* executor,
|
| D | nvptx_compiler.h | 42 HloModule* hlo_module, se::StreamExecutor* stream_exec,
46 HloModule* hlo_module, se::StreamExecutor* stream_exec,
51 GpuVersion GetGpuVersion(se::StreamExecutor* stream_exec) override;
55 GpuVersion gpu_version, se::StreamExecutor* stream_exec, bool relocatable,
60 se::StreamExecutor* stream_exec,
78 se::StreamExecutor* stream_exec, const string& ptx, int cc_major,
|
| /external/tensorflow/tensorflow/stream_executor/tpu/ |
| D | c_api_defn.h | 31 class StreamExecutor; variable
39 stream_executor::StreamExecutor* executor;
43 explicit SE_Stream(stream_executor::StreamExecutor* parent) in SE_Stream()
49 explicit SE_Event(stream_executor::StreamExecutor* parent) : event(parent) {} in SE_Event()
54 explicit SE_Timer(stream_executor::StreamExecutor* parent) : timer(parent) {} in SE_Timer()
|
| D | tpu_transfer_manager.h | 54 Status TransferLiteralToInfeed(stream_executor::StreamExecutor* executor,
58 stream_executor::StreamExecutor* executor,
62 se::StreamExecutor* executor,
66 absl::Span<stream_executor::StreamExecutor* const> executor) override;
74 stream_executor::StreamExecutor* executor,
78 se::StreamExecutor* executor,
|
| /external/tensorflow/tensorflow/compiler/xla/service/ |
| D | backend.h | 106 const std::vector<se::StreamExecutor*>& stream_executors() const { in stream_executors()
111 StatusOr<se::StreamExecutor*> stream_executor(int device_ordinal) const;
116 se::StreamExecutor* default_stream_executor() const { in default_stream_executor()
125 StatusOr<StreamPool::Ptr> BorrowStream(se::StreamExecutor* executor);
160 absl::Span<se::StreamExecutor* const> stream_executors,
173 std::vector<se::StreamExecutor*> stream_executors_;
178 absl::flat_hash_map<se::StreamExecutor*, std::unique_ptr<StreamPool>>
|
| D | compiler.h | 183 std::unique_ptr<HloModule> module, se::StreamExecutor* executor,
186 std::unique_ptr<HloModule> module, se::StreamExecutor* executor, in RunHloPasses()
198 se::StreamExecutor* executor, bool optimize, in RunHloPassesAndBufferAssignement()
212 std::unique_ptr<HloModule> module, se::StreamExecutor* executor,
215 std::unique_ptr<HloModule> module, se::StreamExecutor* executor, in RunBackend()
229 std::vector<std::vector<se::StreamExecutor*>> stream_exec,
233 std::vector<std::vector<se::StreamExecutor*>> stream_exec, in Compile()
247 se::StreamExecutor* executor) const;
257 se::StreamExecutor* executor) const;
|
| D | platform_util.cc | 130 static bool IsDeviceSupported(se::StreamExecutor* executor) { in IsDeviceSupported()
164 /* static */ StatusOr<std::vector<se::StreamExecutor*>>
182 std::vector<se::StreamExecutor*> stream_executors(device_count, nullptr); in GetStreamExecutors()
205 se::StreamExecutor* executor = executor_status.ValueOrDie(); in GetStreamExecutors()
221 std::vector<se::StreamExecutor*> out; in GetStreamExecutors()
222 for (se::StreamExecutor* executor : stream_executors) { in GetStreamExecutors()
|
| /external/tensorflow/tensorflow/stream_executor/rocm/ |
| D | rocm_platform.cc | 47 StreamExecutor* exec = GetExecutor(config).ValueOrDie(); in InspectNumaNodes()
71 StreamExecutor* exec = GetExecutor(config).ValueOrDie(); in DeviceToBus()
75 port::StatusOr<StreamExecutor*> ROCmPlatform::FirstExecutorForBus( in FirstExecutorForBus()
112 port::StatusOr<StreamExecutor*> ROCmPlatform::ExecutorForDevice(int ordinal) { in ExecutorForDevice()
120 port::StatusOr<StreamExecutor*> ROCmPlatform::ExecutorForDeviceWithPluginConfig( in ExecutorForDeviceWithPluginConfig()
129 port::StatusOr<StreamExecutor*> ROCmPlatform::GetExecutor( in GetExecutor()
135 port::StatusOr<std::unique_ptr<StreamExecutor>>
137 auto executor = absl::make_unique<StreamExecutor>( in GetUncachedExecutor()
|
| D | rocm_platform.h | 56 port::StatusOr<StreamExecutor*> FirstExecutorForBus(int bus_ordinal);
70 port::StatusOr<StreamExecutor*> ExecutorForDevice(int ordinal) override;
72 port::StatusOr<StreamExecutor*> ExecutorForDeviceWithPluginConfig(
75 port::StatusOr<StreamExecutor*> GetExecutor(
78 port::StatusOr<std::unique_ptr<StreamExecutor>> GetUncachedExecutor(
|
| /external/tensorflow/tensorflow/stream_executor/cuda/ |
| D | cuda_platform.cc | 85 StreamExecutor* exec = GetExecutor(config).ValueOrDie(); in InspectNumaNodes()
109 StreamExecutor* exec = GetExecutor(config).ValueOrDie(); in DeviceToBus()
113 port::StatusOr<StreamExecutor*> CudaPlatform::FirstExecutorForBus( in FirstExecutorForBus()
149 port::StatusOr<StreamExecutor*> CudaPlatform::ExecutorForDevice(int ordinal) { in ExecutorForDevice()
157 port::StatusOr<StreamExecutor*> CudaPlatform::ExecutorForDeviceWithPluginConfig( in ExecutorForDeviceWithPluginConfig()
166 port::StatusOr<StreamExecutor*> CudaPlatform::GetExecutor( in GetExecutor()
172 port::StatusOr<std::unique_ptr<StreamExecutor>>
174 auto executor = absl::make_unique<StreamExecutor>( in GetUncachedExecutor()
|
| D | cuda_platform.h | 56 port::StatusOr<StreamExecutor*> FirstExecutorForBus(int bus_ordinal);
70 port::StatusOr<StreamExecutor*> ExecutorForDevice(int ordinal) override;
72 port::StatusOr<StreamExecutor*> ExecutorForDeviceWithPluginConfig(
75 port::StatusOr<StreamExecutor*> GetExecutor(
78 port::StatusOr<std::unique_ptr<StreamExecutor>> GetUncachedExecutor(
|
| /external/tensorflow/tensorflow/c/experimental/stream_executor/ |
| D | stream_executor_test.cc | 210 TEST(StreamExecutor, SuccessfulRegistration) { in TEST() argument
227 port::StatusOr<StreamExecutor*> maybe_executor = in TEST()
232 TEST(StreamExecutor, NameNotSet) { in TEST() argument
247 TEST(StreamExecutor, InvalidNameWithSemicolon) { in TEST() argument
264 TEST(StreamExecutor, InvalidNameWithSlash) { in TEST() argument
280 TEST(StreamExecutor, CreateDeviceNotSet) { in TEST() argument
296 TEST(StreamExecutor, UnifiedMemoryAllocateNotSet) { in TEST() argument
325 StreamExecutor* GetExecutor(int ordinal) { in GetExecutor()
331 port::StatusOr<StreamExecutor*> maybe_executor = in GetExecutor()
358 StreamExecutor* executor = GetExecutor(0); in TEST_F()
[all …]
|
| /external/tensorflow/tensorflow/stream_executor/host/ |
| D | host_platform.cc | 49 port::StatusOr<StreamExecutor*> HostPlatform::ExecutorForDevice(int ordinal) { in ExecutorForDevice()
57 port::StatusOr<StreamExecutor*> HostPlatform::ExecutorForDeviceWithPluginConfig( in ExecutorForDeviceWithPluginConfig()
66 port::StatusOr<StreamExecutor*> HostPlatform::GetExecutor( in GetExecutor()
72 port::StatusOr<std::unique_ptr<StreamExecutor>>
74 auto executor = absl::make_unique<StreamExecutor>( in GetUncachedExecutor()
|
| D | host_platform.h | 57 port::StatusOr<StreamExecutor*> ExecutorForDevice(int ordinal) override;
59 port::StatusOr<StreamExecutor*> ExecutorForDeviceWithPluginConfig(
62 port::StatusOr<StreamExecutor*> GetExecutor(
65 port::StatusOr<std::unique_ptr<StreamExecutor>> GetUncachedExecutor(
|
| /external/tensorflow/tensorflow/compiler/xla/service/interpreter/ |
| D | platform.h | 46 port::StatusOr<StreamExecutor*> ExecutorForDevice(int ordinal) override;
48 port::StatusOr<StreamExecutor*> ExecutorForDeviceWithPluginConfig(
51 port::StatusOr<StreamExecutor*> GetExecutor(
54 port::StatusOr<std::unique_ptr<StreamExecutor>> GetUncachedExecutor(
|
| D | platform.cc | 50 port::StatusOr<StreamExecutor*> XlaInterpreterPlatform::ExecutorForDevice( in ExecutorForDevice()
59 port::StatusOr<StreamExecutor*>
69 port::StatusOr<StreamExecutor*> XlaInterpreterPlatform::GetExecutor( in GetExecutor()
75 port::StatusOr<std::unique_ptr<StreamExecutor>>
78 auto executor = absl::make_unique<StreamExecutor>( in GetUncachedExecutor()
|
| /external/tensorflow/tensorflow/core/common_runtime/device/ |
| D | device_event_mgr.h | 35 class StreamExecutor; variable
87 se::StreamExecutor* const exec_;
99 EventMgr(se::StreamExecutor* se, const GPUOptions& gpu_options);
153 EventMgr* GetEventMgr(se::StreamExecutor* se, const GPUOptions& gpu_options);
160 std::map<se::StreamExecutor*, EventMgr*> event_mgr_map_ TF_GUARDED_BY(mu_);
|