Searched refs:cubin (Results 1 – 11 of 11) sorted by relevance
| /external/llvm-project/mlir/test/Conversion/GPUToCUDA/ |
| D | lower-nvvm-kernel-to-cubin.mlir | 1 // RUN: mlir-opt %s --test-kernel-to-cubin -split-input-file | FileCheck %s
3 // CHECK: attributes {nvvm.cubin = "CUBIN"}
|
| /external/tensorflow/tensorflow/stream_executor/cuda/ |
| D | cuda_asm_compiler.cc | 49 std::vector<uint8> cubin(static_cast<uint8*>(cubin_out), in LinkGpuAsm() local
52 return std::move(cubin); in LinkGpuAsm()
|
| D | cuda_gpu_executor.cc | 220 port::Status GpuExecutor::LoadModuleFromCuBin(const char* cubin, in LoadModuleFromCuBin() argument
223 std::tie(*module, module_refcount) = gpu_binary_to_module_[cubin]; in LoadModuleFromCuBin()
226 TF_RETURN_IF_ERROR(GpuDriver::LoadCubin(context_, cubin, module)); in LoadModuleFromCuBin()
228 VLOG(3) << "Loaded CUBIN " << static_cast<const void*>(cubin) in LoadModuleFromCuBin()
232 VLOG(3) << "CUBIN " << static_cast<const void*>(cubin) in LoadModuleFromCuBin()
235 gpu_binary_to_module_[cubin] = {*module, module_refcount}; in LoadModuleFromCuBin()
274 const char* cubin = spec.cuda_cubin_in_memory().bytes(); in GetKernel() local
275 TF_RETURN_IF_ERROR(LoadModuleFromCuBin(cubin, &module)); in GetKernel()
276 kernel_to_gpu_binary_[kernel] = cubin; in GetKernel()
|
| /external/tensorflow/third_party/nccl/ |
| D | build_defs.bzl.tpl | 91 cubin = ctx.actions.declare_file("%s_%s.cubin" % (name, arch))
94 outputs = [register_h, cubin],
100 "--output-file=%s" % cubin.path,
104 cubins.append(cubin)
105 images.append("--image=profile=%s,file=%s" % (arch, cubin.path))
277 into cubin files, one per GPU architecture. It also produces a header file
290 register.h xy.cubin
|
| /external/llvm-project/mlir/test/Conversion/GPUCommon/ |
| D | lower-launch-func-to-gpu-runtime-calls.mlir | 1 // RUN: mlir-opt -allow-unregistered-dialect %s --gpu-to-llvm="gpu-binary-annotation=nvvm.cubin" | …
11 nvvm.cubin = "CUBIN", rocdl.hsaco = "HSACO"
|
| /external/tensorflow/tensorflow/stream_executor/gpu/ |
| D | asm_compiler.cc | 278 std::string cubin; in CompileGpuAsm() local
280 cubin_path, &cubin)); in CompileGpuAsm()
281 std::vector<uint8> cubin_vector(cubin.begin(), cubin.end()); in CompileGpuAsm()
|
| /external/skqp/src/compute/hs/cuda/sm_35/u64/ |
| D | gen.bat | 39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
|
| /external/skqp/src/compute/hs/cuda/sm_35/u32/ |
| D | gen.bat | 39 nvcc -I ../../../.. -use_fast_math -res-usage -cubin -arch sm_35 %HS_SRC%
|
| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | nvptx_compiler.cc | 333 std::vector<uint8> cubin = CompileGpuAsmOrGetCachedResult( in CompileTargetBinary() local
338 std::move(cubin)); in CompileTargetBinary()
|
| /external/tensorflow/tensorflow/core/kernels/mlir_generated/ |
| D | build_defs.bzl | 1 """Generates cubin headers for TF dialect ops."""
29 "GPU binaries in either cubin format or hsaco format",
|
| /external/tensorflow/tensorflow/stream_executor/rocm/ |
| D | rocm_gpu_executor.cc | 386 port::Status GpuExecutor::LoadModuleFromCuBin(const char* cubin, in LoadModuleFromCuBin() argument
|