/external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
D | nvptx_compiler.h | 101 const string& ptx, int cc_major, int cc_minor, 117 CompilationCacheKey(std::string ptx, int cc_major, int cc_minor) in CompilationCacheKey() 118 : ptx(std::move(ptx)), cc_major(cc_major), cc_minor(cc_minor) {} in CompilationCacheKey() 120 int cc_major; member 126 tensorflow::Hash64Combine(tensorflow::Hash64(key.ptx), key.cc_major), in operator() 133 return a.cc_major == b.cc_major && a.cc_minor == b.cc_minor && in operator()
|
D | nvptx_compiler.cc | 540 const string& ptx, int cc_major, int cc_minor, in CompilePtx() argument 582 absl::StrCat("-arch=sm_", cc_major, cc_minor)}; in CompilePtx() 737 int cc_major, cc_minor; in RunBackend() local 738 if (!stream_exec->GetDeviceDescription().cuda_compute_capability(&cc_major, in RunBackend() 742 cc_major = 2; in RunBackend() 749 TF_ASSIGN_OR_RETURN(ptx, CompileToPtx(&llvm_module, {cc_major, cc_minor}, in RunBackend() 764 CompilePtxOrGetCachedResult(ptx, cc_major, cc_minor, module->config()); in RunBackend() 793 ptx, cubin, {cc_major, cc_minor}, std::move(thunk_schedule), in RunBackend() 804 const string& ptx, int cc_major, int cc_minor, in CompilePtxOrGetCachedResult() argument 819 std::forward_as_tuple(ptx, cc_major, cc_minor), in CompilePtxOrGetCachedResult() [all …]
|
D | cudnn_conv_algorithm_picker.cc | 117 int cc_major, cc_minor; in GetComputeCapability() local 118 stream_executor->GetDeviceDescription().cuda_compute_capability(&cc_major, in GetComputeCapability() 120 cc.set_major(cc_major); in GetComputeCapability()
|
/external/tensorflow/tensorflow/core/common_runtime/gpu/ |
D | gpu_device_test.cc | 44 Status GetComputeCapability(PlatformGpuId gpu_id, int* cc_major, in GetComputeCapability() argument 49 if (!se->GetDeviceDescription().cuda_compute_capability(cc_major, cc_minor)) { in GetComputeCapability() 50 *cc_major = 0; in GetComputeCapability() 225 int cc_major, cc_minor; in TEST_F() local 226 TF_ASSERT_OK(GetComputeCapability(PlatformGpuId(0), &cc_major, &cc_minor)); in TEST_F() 228 if (cc_major >= 6) { in TEST_F() 249 int cc_major, cc_minor; in TEST_F() local 250 TF_ASSERT_OK(GetComputeCapability(kPlatformGpuId, &cc_major, &cc_minor)); in TEST_F() 252 if (cc_major < 6) { in TEST_F()
|
D | gpu_device.cc | 904 int cc_major = 0, cc_minor = 0; in SingleVirtualDeviceMemoryLimit() local 905 if (!se->GetDeviceDescription().cuda_compute_capability(&cc_major, in SingleVirtualDeviceMemoryLimit() 909 if (cc_major < 6) { in SingleVirtualDeviceMemoryLimit() 1164 int cc_major; in GetShortDeviceDescription() local 1166 if (!desc.cuda_compute_capability(&cc_major, &cc_minor)) { in GetShortDeviceDescription() 1167 cc_major = 0; in GetShortDeviceDescription() 1173 ", compute capability: ", cc_major, ".", cc_minor); in GetShortDeviceDescription() 1544 int cc_major; in GetValidDeviceIds() local 1546 if (!description.cuda_compute_capability(&cc_major, &cc_minor)) { in GetValidDeviceIds() 1548 cc_major = 0; in GetValidDeviceIds() [all …]
|
/external/tensorflow/tensorflow/stream_executor/ |
D | stream_executor_pimpl.cc | 330 int cc_major, cc_minor; in GetConvolveAlgorithms() local 331 GetDeviceDescription().cuda_compute_capability(&cc_major, &cc_minor); in GetConvolveAlgorithms() 332 return dnn_support->GetConvolveAlgorithms(with_winograd_nonfused, cc_major, in GetConvolveAlgorithms() 352 int cc_major, cc_minor; in GetConvolveBackwardDataAlgorithms() local 353 GetDeviceDescription().cuda_compute_capability(&cc_major, &cc_minor); in GetConvolveBackwardDataAlgorithms() 355 with_winograd_nonfused, cc_major, cc_minor, out_algorithms); in GetConvolveBackwardDataAlgorithms() 365 int cc_major, cc_minor; in GetConvolveBackwardFilterAlgorithms() local 366 GetDeviceDescription().cuda_compute_capability(&cc_major, &cc_minor); in GetConvolveBackwardFilterAlgorithms() 368 with_winograd_nonfused, cc_major, cc_minor, out_algorithms); in GetConvolveBackwardFilterAlgorithms()
|
D | dnn.cc | 30 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveAlgorithms() argument 40 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardDataAlgorithms() argument 46 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardFilterAlgorithms() argument
|
D | dnn.h | 1272 bool with_winograd_nonfused, int cc_major, int cc_minor, 1365 bool with_winograd_nonfused, int cc_major, int cc_minor, 1412 bool with_winograd_nonfused, int cc_major, int cc_minor,
|
/external/tensorflow/tensorflow/core/kernels/ |
D | gpu_utils.cc | 48 int cc_major, cc_minor; in GetComputeCapability() local 49 stream_executor->GetDeviceDescription().cuda_compute_capability(&cc_major, in GetComputeCapability() 51 cc.set_major(cc_major); in GetComputeCapability()
|
/external/tensorflow/tensorflow/contrib/fused_conv/kernels/ |
D | fused_conv2d_bias_activation_op.cc | 304 int cc_major, cc_minor; in GetComputeCapability() local 305 stream_executor->GetDeviceDescription().cuda_compute_capability(&cc_major, in GetComputeCapability() 307 cc.set_major(cc_major); in GetComputeCapability() 460 int cc_major, cc_minor; in launch() local 461 stream->parent()->GetDeviceDescription().cuda_compute_capability(&cc_major, in launch() 464 ctx, ((cc_major == 6 && cc_minor >= 1) || cc_major > 6), in launch()
|
/external/tensorflow/tensorflow/stream_executor/cuda/ |
D | cuda_blas.cc | 1598 int cc_major, cc_minor; in DoBlasGemm() local 1599 stream->parent()->GetDeviceDescription().cuda_compute_capability(&cc_major, in DoBlasGemm() 1603 if (cc_major >= 7 && TensorOpMathEnabled()) { in DoBlasGemm() 1872 static bool TensorOpsAvailable(int cc_major) { in TensorOpsAvailable() argument 1878 if (cc_major >= 7 && TensorOpMathEnabled() && in TensorOpsAvailable() 1895 int cc_major, cc_minor; in DoBlasGemmWithAlgorithmImpl() local 1897 &cc_major, &cc_minor) && in DoBlasGemmWithAlgorithmImpl() 1898 cc_major < 5) { in DoBlasGemmWithAlgorithmImpl() 1899 VLOG(2) << "DoBlasGemmWithAlgorithm returning false because sm" << cc_major in DoBlasGemmWithAlgorithmImpl() 1904 if (UsesTensorOps(algorithm) && !TensorOpsAvailable<InT>(cc_major)) { in DoBlasGemmWithAlgorithmImpl() [all …]
|
D | cuda_dnn.h | 203 bool with_winograd_nonfused, int cc_major, int cc_minor, 210 bool with_winograd_nonfused, int cc_major, int cc_minor, 214 bool with_winograd_nonfused, int cc_major, int cc_minor,
|
D | cuda_dnn.cc | 3038 inline bool TensorOpMathAvailable(int cc_major) { in TensorOpMathAvailable() argument 3039 return cc_major >= 7 && CUDNN_VERSION >= 7000 && TensorOpMathEnabled(); in TensorOpMathAvailable() 3043 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveAlgorithms() argument 3045 bool tensor_op_math_available = TensorOpMathAvailable(cc_major); in GetConvolveAlgorithms() 3104 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardDataAlgorithms() argument 3106 bool tensor_op_math_available = TensorOpMathAvailable(cc_major); in GetConvolveBackwardDataAlgorithms() 3139 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardFilterAlgorithms() argument 3141 bool tensor_op_math_available = TensorOpMathAvailable(cc_major); in GetConvolveBackwardFilterAlgorithms() 3429 int cc_major, cc_minor; in DoFusedConvolve() local 3430 stream->parent()->GetDeviceDescription().cuda_compute_capability(&cc_major, in DoFusedConvolve() [all …]
|
D | cuda_driver.cc | 1350 /* static */ port::Status GpuDriver::GetComputeCapability(int* cc_major, in GetComputeCapability() argument 1353 *cc_major = 0; in GetComputeCapability() 1357 cc_major, CU_DEVICE_ATTRIBUTE_COMPUTE_CAPABILITY_MAJOR, device); in GetComputeCapability()
|
/external/tensorflow/tensorflow/stream_executor/rocm/ |
D | rocm_dnn.h | 194 bool with_winograd_nonfused, int cc_major, int cc_minor, 201 bool with_winograd_nonfused, int cc_major, int cc_minor, 205 bool with_winograd_nonfused, int cc_major, int cc_minor,
|
D | rocm_driver.cc | 1055 /* static */ port::Status GpuDriver::GetComputeCapability(int* cc_major, in GetComputeCapability() argument
|
D | rocm_dnn.cc | 2949 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveAlgorithms() argument 2970 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardDataAlgorithms() argument 2985 bool with_winograd_nonfused, int cc_major, int cc_minor, in GetConvolveBackwardFilterAlgorithms() argument
|
/external/tensorflow/tensorflow/stream_executor/gpu/ |
D | gpu_driver.h | 398 static port::Status GetComputeCapability(int* cc_major, int* cc_minor,
|