/aosp_15_r20/external/eigen/unsupported/test/ |
H A D | cxx11_tensor_chipping_sycl.cpp | 50 TensorMap<Tensor<DataType, 5, DataLayout,IndexType>> gpu_tensor(gpu_data_tensor, tensorRange); in test_static_chip_sycl() local 54 gpu_chip1.device(sycl_device)=gpu_tensor.template chip<0l>(1l); in test_static_chip_sycl() 78 gpu_chip2.device(sycl_device)=gpu_tensor.template chip<1l>(1l); in test_static_chip_sycl() 102 gpu_chip3.device(sycl_device)=gpu_tensor.template chip<2l>(2l); in test_static_chip_sycl() 126 gpu_chip4.device(sycl_device)=gpu_tensor.template chip<3l>(5l); in test_static_chip_sycl() 151 gpu_chip5.device(sycl_device)=gpu_tensor.template chip<4l>(7l); in test_static_chip_sycl() 199 TensorMap<Tensor<DataType, 5, DataLayout,IndexType>> gpu_tensor(gpu_data_tensor, tensorRange); in test_dynamic_chip_sycl() local 203 gpu_chip1.device(sycl_device)=gpu_tensor.chip(1l,0l); in test_dynamic_chip_sycl() 227 gpu_chip2.device(sycl_device)=gpu_tensor.chip(1l,1l); in test_dynamic_chip_sycl() 251 gpu_chip3.device(sycl_device)=gpu_tensor.chip(2l,2l); in test_dynamic_chip_sycl() [all …]
|
H A D | cxx11_tensor_striding_sycl.cpp | 53 …Eigen::TensorMap<Eigen::Tensor<DataType, 4, DataLayout, IndexType> > gpu_tensor(d_tensor, tensor_d… in test_simple_striding() local 65 gpu_no_stride.device(sycl_device)=gpu_tensor.stride(strides); in test_simple_striding() 92 gpu_stride.device(sycl_device)=gpu_tensor.stride(strides); in test_simple_striding() 136 …Eigen::TensorMap<Eigen::Tensor<DataType, 4, DataLayout, IndexType> > gpu_tensor(d_tensor, tensor_d… in test_striding_as_lvalue() local 151 gpu_stride.stride(strides).device(sycl_device)=gpu_tensor; in test_striding_as_lvalue() 172 gpu_no_stride.stride(strides).device(sycl_device)=gpu_tensor.stride(no_strides); in test_striding_as_lvalue()
|
H A D | cxx11_tensor_patch_sycl.cpp | 58 TensorMap<Tensor<DataType, 4, DataLayout,IndexType>> gpu_tensor(gpu_data_tensor, tensorRange); in test_simple_patch_sycl() local 62 gpu_no_patch.device(sycl_device)=gpu_tensor.extract_patches(patch_dims); in test_simple_patch_sycl() 98 gpu_single_patch.device(sycl_device)=gpu_tensor.extract_patches(patch_dims); in test_simple_patch_sycl() 133 gpu_twod_patch.device(sycl_device)=gpu_tensor.extract_patches(patch_dims); in test_simple_patch_sycl() 189 gpu_threed_patch.device(sycl_device)=gpu_tensor.extract_patches(patch_dims); in test_simple_patch_sycl()
|
H A D | cxx11_tensor_inflation_sycl.cpp | 56 TensorMap<Tensor<DataType, 4, DataLayout,IndexType>> gpu_tensor(gpu_data_tensor, tensorRange); in test_simple_inflation_sycl() local 60 gpu_no_stride.device(sycl_device)=gpu_tensor.inflate(strides); in test_simple_inflation_sycl() 95 gpu_inflated.device(sycl_device)=gpu_tensor.inflate(strides); in test_simple_inflation_sycl()
|
/aosp_15_r20/external/tensorflow/tensorflow/core/runtime_fallback/runtime/gpu/ |
H A D | conversion_function.cc | 46 const tfrt::gpu::DenseGpuTensor& gpu_tensor, Device* device, in CopyRefGpuTensorToRuntimeFallbackTensor() argument 50 gpu_tensor.CopyBufferRef(); in CopyRefGpuTensorToRuntimeFallbackTensor() 52 std::move(gpu_buffer), gpu_tensor.dtype(), gpu_tensor.shape()); in CopyRefGpuTensorToRuntimeFallbackTensor() 57 return RuntimeFallbackTensor(gpu_tensor.shape(), gpu_tensor.dtype(), in CopyRefGpuTensorToRuntimeFallbackTensor() 146 tfrt::gpu::DenseGpuTensor gpu_tensor{ in ConvertRuntimeFallbackTensorToDenseGpuTensor() local 154 exec_ctx.host(), std::move(gpu_tensor)); in ConvertRuntimeFallbackTensorToDenseGpuTensor()
|
/aosp_15_r20/external/tensorflow/tensorflow/core/common_runtime/gpu/ |
H A D | gpu_util.cc | 268 const Tensor* gpu_tensor, Tensor* cpu_tensor, in CopyGPUTensorToCPU() argument 273 Status s = PrepareCopy(gpu_device, device_context, *gpu_tensor, cpu_tensor, in CopyGPUTensorToCPU() 290 const int64_t total_bytes = gpu_tensor->TotalBytes(); in CopyGPUTensorToCPU() 292 void* src_ptr = GetBase(gpu_tensor); in CopyGPUTensorToCPU() 298 TensorReference input_ref(*gpu_tensor); in CopyGPUTensorToCPU() 313 Device* gpu_device, Tensor* gpu_tensor, in CopyCPUTensorToGPU() argument 318 Status s = PrepareCopy(gpu_device, device_context, *cpu_tensor, gpu_tensor, in CopyCPUTensorToGPU() 349 void* dst_ptr = GetBase(gpu_tensor); in CopyCPUTensorToGPU()
|
H A D | gpu_util.h | 49 const Tensor* gpu_tensor, Tensor* cpu_tensor, 90 Device* gpu_device, Tensor* gpu_tensor,
|
H A D | gpu_device_test.cc | 114 void CopyCPUToGPU(Tensor* cpu_tensor, Tensor* gpu_tensor, Device* device, in CopyCPUToGPU() argument 117 gpu_tensor)); in CopyCPUToGPU() 120 void CopyGPUToCPU(Tensor* gpu_tensor, Tensor* cpu_tensor, Device* device, in CopyGPUToCPU() argument 123 gpu_tensor, /*tensor_name=*/"", device, cpu_tensor)); in CopyGPUToCPU()
|
H A D | gpu_device.cc | 589 Tensor BaseGPUDevice::CopyGpuTensorToHostDebugOnly(const Tensor& gpu_tensor) { in CopyGpuTensorToHostDebugOnly() argument 590 Tensor host_tensor(gpu_tensor.dtype(), gpu_tensor.shape()); in CopyGpuTensorToHostDebugOnly() 593 se::DeviceMemoryBase(gpu_tensor.data(), in CopyGpuTensorToHostDebugOnly() 594 gpu_tensor.TotalBytes()), in CopyGpuTensorToHostDebugOnly() 595 gpu_tensor.TotalBytes()) in CopyGpuTensorToHostDebugOnly()
|
H A D | gpu_device.h | 205 Tensor CopyGpuTensorToHostDebugOnly(const Tensor& gpu_tensor);
|
/aosp_15_r20/external/tensorflow/tensorflow/lite/delegates/gpu/metal/ |
H A D | inference_context.cc | 347 MetalSpatialTensor* gpu_tensor = GetTensor(id); in SetInputTensor() local 348 TensorDescriptor descriptor_with_data = gpu_tensor->GetDescriptor(); in SetInputTensor() 350 return gpu_tensor->UploadDescriptorData(descriptor_with_data, device_); in SetInputTensor() 355 const MetalSpatialTensor* gpu_tensor = GetTensor(id); in GetOutputTensor() local 356 const auto dst_shape = BHWC(gpu_tensor->Batch(), gpu_tensor->Height(), in GetOutputTensor() 357 gpu_tensor->Width(), gpu_tensor->Channels()); in GetOutputTensor() 363 RETURN_IF_ERROR(gpu_tensor->ToDescriptor(&desc, device_)); in GetOutputTensor()
|
H A D | metal_arguments.cc | 76 MetalSpatialTensor gpu_tensor; in CreateMetalObject() local 77 RETURN_IF_ERROR(gpu_tensor.CreateFromDescriptor(*tensor_desc, device)); in CreateMetalObject() 78 *result = std::make_unique<MetalSpatialTensor>(std::move(gpu_tensor)); in CreateMetalObject()
|
H A D | BUILD | 235 "//tensorflow/lite/delegates/gpu/common/task:gpu_tensor",
|
/aosp_15_r20/external/tensorflow/tensorflow/lite/delegates/gpu/cl/ |
H A D | inference_context.cc | 905 Tensor* gpu_tensor = GetTensor(id); in SetInputTensor() local 906 TensorDescriptor descriptor_with_data = gpu_tensor->GetDescriptor(); in SetInputTensor() 908 return gpu_tensor->UploadDescriptorData(descriptor_with_data, queue); in SetInputTensor() 914 const Tensor* gpu_tensor = GetTensor(id); in GetOutputTensor() local 915 const auto dst_shape = BHWC(gpu_tensor->Batch(), gpu_tensor->Height(), in GetOutputTensor() 916 gpu_tensor->Width(), gpu_tensor->Channels()); in GetOutputTensor() 922 RETURN_IF_ERROR(gpu_tensor->ToDescriptor(&desc, queue)); in GetOutputTensor()
|
H A D | cl_arguments.cc | 115 Tensor gpu_tensor; in CreateCLObject() local 116 RETURN_IF_ERROR(gpu_tensor.CreateFromDescriptor(*tensor_desc, context)); in CreateCLObject() 117 *result = std::make_unique<Tensor>(std::move(gpu_tensor)); in CreateCLObject()
|
/aosp_15_r20/external/tensorflow/tensorflow/core/tfrt/eager/cpp_tests/core_runtime/ |
H A D | op_handler_selector_test.cc | 403 tensorflow::core::RefCountPtr<FakeTensorHandle> gpu_tensor( in TEST_F() local 417 TF_ASSERT_OK(op->AddInput(gpu_tensor.get())); in TEST_F() 432 tensorflow::core::RefCountPtr<FakeTensorHandle> gpu_tensor( in TEST_F() local 446 TF_ASSERT_OK(op->AddInput(gpu_tensor.get())); in TEST_F()
|
/aosp_15_r20/external/tensorflow/tensorflow/lite/delegates/gpu/common/task/ |
H A D | BUILD | 75 "//tensorflow/lite/delegates/gpu/common/task:gpu_tensor", 85 name = "gpu_tensor", 86 hdrs = ["gpu_tensor.h"],
|
/aosp_15_r20/external/tensorflow/tensorflow/compiler/tf2tensorrt/kernels/ |
H A D | trt_engine_op.cc | 549 const Tensor& gpu_tensor = ctx->input(i); in CopyToHostAsync() local 551 t.flat<int32>().data(), gpu_tensor.flat<int32>().data(), in CopyToHostAsync() 563 Tensor* gpu_tensor; in CopyToDeviceAsync() local 565 ctx->allocate_output(t, native_tensor.shape(), &gpu_tensor)); in CopyToDeviceAsync() 566 auto ret = cudaMemcpyAsync(gpu_tensor->flat<int32>().data(), in CopyToDeviceAsync()
|
/aosp_15_r20/external/tensorflow/tensorflow/core/runtime_fallback/runtime/ |
H A D | BUILD | 133 "@tf_runtime//backends/gpu:gpu_tensor", 231 "@tf_runtime//backends/gpu:gpu_tensor",
|
/aosp_15_r20/external/tensorflow/tensorflow/lite/delegates/gpu/metal/benchmarking/ |
H A D | main.mm | 82 tflite::gpu::TensorFloat32 gpu_tensor; 83 RETURN_IF_ERROR(gpu_context->GetOutputTensor(outputs[i]->id, &gpu_tensor)); 93 const float* gpu_out = gpu_tensor.data.data();
|
/aosp_15_r20/external/tensorflow/tensorflow/core/common_runtime/ |
H A D | direct_session_test.cc | 2401 Tensor gpu_tensor; in TEST() local 2415 gpu_tensor = outputs[0]; in TEST() 2416 ASSERT_TRUE(IsCUDATensor(gpu_tensor)); in TEST() 2425 TF_ASSERT_OK(session->RunCallable(feed_gpu_fetch_cpu, {gpu_tensor}, in TEST() 2474 Tensor gpu_tensor; in TestFeedAndFetchTensorsInDeviceMemory() local 2489 gpu_tensor = outputs[0]; in TestFeedAndFetchTensorsInDeviceMemory() 2490 ASSERT_TRUE(IsCUDATensor(gpu_tensor)) << DataType_Name(dtype); in TestFeedAndFetchTensorsInDeviceMemory() 2501 TF_ASSERT_OK(session->RunCallable(handle, {gpu_tensor}, &outputs, nullptr)) in TestFeedAndFetchTensorsInDeviceMemory()
|
/aosp_15_r20/external/tensorflow/tensorflow/core/tfrt/runtime/ |
H A D | BUILD | 49 "@tf_runtime//backends/gpu:gpu_tensor",
|
/aosp_15_r20/external/tensorflow/tensorflow/python/eager/ |
H A D | ops_test.py | 287 gpu_tensor = cpu_tensor.gpu() 288 self.assertAllEqual(cpu_tensor + gpu_tensor, 2.0)
|
/aosp_15_r20/external/tensorflow/tensorflow/python/framework/ |
H A D | config_test.py | 74 gpu_tensor = cpu_tensor.gpu() 75 self.assertAllEqual(cpu_tensor + gpu_tensor, 2.0)
|
/aosp_15_r20/external/pytorch/docs/cpp/source/notes/ |
H A D | tensor_creation.rst | 326 torch::Tensor gpu_tensor = float_tensor.to(torch::kCUDA); 344 torch::Tensor async_cpu_tensor = gpu_tensor.to(torch::kCPU, /*non_blocking=*/true);
|