Home
last modified time | relevance | path

Searched refs:cubin (Results 1 – 11 of 11) sorted by relevance

/external/tensorflow/tensorflow/compiler/xla/service/gpu/
Dgpu_executable.h54 GpuExecutable(const string& ptx, const std::vector<uint8>& cubin,
75 const std::vector<uint8>& cubin() const { return cubin_; } in cubin() function
Dgpu_executable.cc51 const string& ptx, const std::vector<uint8>& cubin, in GpuExecutable() argument
61 cubin_(cubin), in GpuExecutable()
196 if (!cubin().empty()) { in ResolveConstantGlobals()
197 module_spec.AddCudaCubinInMemory(cubin()); in ResolveConstantGlobals()
Dkernel_thunk.cc46 if (!executable.cubin().empty()) { in Initialize()
48 reinterpret_cast<const char*>(executable.cubin().data()), in Initialize()
Dnvptx_compiler.cc605 string cubin; in CompilePtx() local
607 cubin_path, &cubin)); in CompilePtx()
608 std::vector<uint8> cubin_vector(cubin.begin(), cubin.end()); in CompilePtx()
763 const std::vector<uint8> cubin = in RunBackend() local
793 ptx, cubin, {cc_major, cc_minor}, std::move(thunk_schedule), in RunBackend()
/external/tensorflow/tensorflow/stream_executor/cuda/
Dcuda_gpu_executor.cc220 bool GpuExecutor::LoadModuleFromCuBin(const char* cubin, CUmodule* module) { in LoadModuleFromCuBin() argument
222 std::tie(*module, module_refcount) = gpu_binary_to_module_[cubin]; in LoadModuleFromCuBin()
225 auto load_status = GpuDriver::LoadCubin(context_, cubin, module); in LoadModuleFromCuBin()
231 VLOG(3) << "Loaded CUBIN " << static_cast<const void *>(cubin) in LoadModuleFromCuBin()
235 VLOG(3) << "CUBIN " << static_cast<const void *>(cubin) in LoadModuleFromCuBin()
238 gpu_binary_to_module_[cubin] = {*module, module_refcount}; in LoadModuleFromCuBin()
278 const char *cubin = spec.cuda_cubin_in_memory().bytes(); in GetKernel() local
279 if (!LoadModuleFromCuBin(cubin, &module)) { in GetKernel()
282 kernel_to_gpu_binary_[kernel] = cubin; in GetKernel()
/external/skqp/src/compute/hs/cuda/sm_35/u64/
Dgen.bat39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
/external/skia/src/compute/hs/cuda/sm_35/u64/
Dgen.bat39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
/external/skqp/src/compute/hs/cuda/sm_35/u32/
Dgen.bat39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
/external/skia/src/compute/hs/cuda/sm_35/u32/
Dgen.bat39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
/external/tensorflow/tensorflow/stream_executor/gpu/
Dgpu_executor.h273 bool LoadModuleFromCuBin(const char* cubin, GpuModuleHandle* module)
/external/tensorflow/tensorflow/stream_executor/rocm/
Drocm_gpu_executor.cc385 bool GpuExecutor::LoadModuleFromCuBin(const char* cubin, hipModule_t* module) { in LoadModuleFromCuBin() argument