Searched refs:compute_capability (Results 1 – 10 of 10) sorted by relevance
| /external/tensorflow/tensorflow/core/profiler/utils/ |
| D | hardware_type_utils.cc | 33 switch (device_cap.compute_capability().major()) { in GetFmaMaxThroughputPerSMPerCycle()
48 if (device_cap.compute_capability().minor() > 0) { in GetFmaMaxThroughputPerSMPerCycle()
63 if (device_cap.compute_capability().minor() >= 6) { in GetFmaMaxThroughputPerSMPerCycle()
90 switch (device_cap.compute_capability().major()) { in GpuModelName()
100 if (device_cap.compute_capability().minor() < 5) { in GpuModelName()
|
| D | device_caps_utils.cc | 44 caps.compute_capability().major()); in SetDeviceCaps()
47 caps.compute_capability().minor()); in SetDeviceCaps()
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/llvm_gpu_backend/ |
| D | gpu_backend_lib.cc | 89 static string GetSmName(std::pair<int, int> compute_capability) { in GetSmName() argument
91 compute_capability.first * 10 + compute_capability.second; in GetSmName()
110 LOG(WARNING) << "Unknown compute capability (" << compute_capability.first in GetSmName()
111 << ", " << compute_capability.second << ") ." in GetSmName()
317 std::pair<int, int> compute_capability, in LinkLibdeviceIfNecessary() argument
343 auto compute_capability = absl::get_if<std::pair<int, int>>(&gpu_version); in NVPTXTargetModuleLinker() local
344 if (!compute_capability) { in NVPTXTargetModuleLinker()
347 TF_RETURN_IF_ERROR(LinkLibdeviceIfNecessary(module, *compute_capability, in NVPTXTargetModuleLinker()
366 llvm::Triple target_triple, std::pair<int, int> compute_capability, in NVPTXGetTargetMachine() argument
370 return GetTargetMachine(target_triple, GetSmName(compute_capability), in NVPTXGetTargetMachine()
[all …]
|
| /external/tensorflow/tensorflow/python/keras/mixed_precision/ |
| D | device_compatibility_check_test.py | 29 def device_details(device_name, compute_capability=None): argument
33 if compute_capability:
34 details['compute_capability'] = compute_capability
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | nvptx_compiler.cc | 307 std::pair<int, int> compute_capability = in CompileTargetBinary() local
334 stream_exec, ptx, compute_capability.first, compute_capability.second, in CompileTargetBinary()
|
| /external/tensorflow/tensorflow/core/profiler/protobuf/ |
| D | hardware_types.proto | 29 CudaComputeCapability compute_capability = 5; field
|
| /external/tensorflow/tensorflow/core/grappler/optimizers/ |
| D | generic_layout_optimizer.cc | 63 double compute_capability = 0.0; in GetNumGPUs() local
64 if (absl::SimpleAtod(compute_capability_it->second, &compute_capability) && in GetNumGPUs()
65 compute_capability >= 7.0) { in GetNumGPUs()
|
| /external/tensorflow/tensorflow/core/protobuf/ |
| D | autotuning.proto | 80 ComputeCapability compute_capability = 4; field
|
| /external/tensorflow/ |
| D | configure.py | 1033 for compute_capability in tf_cuda_compute_capabilities.split(','):
1034 m = re.match('[0-9]+.[0-9]+', compute_capability)
1038 compute_capability)
1040 print('Invalid compute capability: %s' % compute_capability)
|
| /external/tensorflow/tensorflow/python/framework/ |
| D | test_util.py | 1590 cc = gpu_info.compute_capability or (0, 0)
|