experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Add XLA_FLAGS=--xla_gpu_gpuasm_extra_flags=...

Browse Source
This commit is contained in:
Frederic Bastien 2020-06-08 13:14:33 -07:00
parent ce806400a4
commit 256cfa1823
3 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tensorflow/compiler/xla/debug_options_flags.cc
View File

@ -37,6 +37,7 @@ DebugOptions DefaultDebugOptionsIgnoringFlags() {
opts.set_xla_gpu_autotune_level(4); opts.set_xla_gpu_autotune_level(4);
opts.set_xla_cpu_multi_thread_eigen(true); opts.set_xla_cpu_multi_thread_eigen(true);
opts.set_xla_gpu_cuda_data_dir("./cuda_sdk_lib"); opts.set_xla_gpu_cuda_data_dir("./cuda_sdk_lib");
opts.set_xla_gpu_gpuasm_extra_flags("");
opts.set_xla_eliminate_hlo_implicit_broadcast(true); opts.set_xla_eliminate_hlo_implicit_broadcast(true);
opts.set_xla_dump_hlo_as_html(false); opts.set_xla_dump_hlo_as_html(false);
opts.set_xla_dump_include_timestamp(true); opts.set_xla_dump_include_timestamp(true);
@ -430,6 +431,10 @@ static void AllocateFlags() {
bool_setter_for(&DebugOptions::set_xla_gpu_disable_gpuasm_optimizations), bool_setter_for(&DebugOptions::set_xla_gpu_disable_gpuasm_optimizations),
flag_values->xla_gpu_disable_gpuasm_optimizations(), flag_values->xla_gpu_disable_gpuasm_optimizations(),
"In XLA:GPU run ptxas in -O0 (default is -O3).")); "In XLA:GPU run ptxas in -O0 (default is -O3)."));
flag_objects->push_back(tensorflow::Flag(
"xla_gpu_gpuasm_extra_flags", string_setter_for(&DebugOptions::set_xla_gpu_gpuasm_extra_flags),
"", //flag_values->xla_gpu_gpuasm_extra_flags(),
"Pass extra parameters to the GPU assembler tool (i.e., ptxas for CUDA)."));
flag_objects->push_back(tensorflow::Flag( flag_objects->push_back(tensorflow::Flag(
"xla_fuel", setter_for_xla_fuel, /*default_value_for_display=*/"", "xla_fuel", setter_for_xla_fuel, /*default_value_for_display=*/"",
"Sets compiler fuel, useful for bisecting bugs in passes. Format " "Sets compiler fuel, useful for bisecting bugs in passes. Format "

6
tensorflow/compiler/xla/service/gpu/stream_executor_util.cc
View File

@ -222,9 +222,13 @@ Status ExecuteKernelOnStream(const se::KernelBase& kernel,
} }
se::GpuAsmOpts PtxOptsFromConfig(const HloModuleConfig& hlo_module_config) { se::GpuAsmOpts PtxOptsFromConfig(const HloModuleConfig& hlo_module_config) {
string extra_string = hlo_module_config.debug_options().xla_gpu_gpuasm_extra_flags();
std::vector<std::string> extra_flags;
extra_flags = absl::StrSplit(extra_string, " ", absl::SkipEmpty());
return se::GpuAsmOpts( return se::GpuAsmOpts(
hlo_module_config.debug_options().xla_gpu_disable_gpuasm_optimizations(), hlo_module_config.debug_options().xla_gpu_disable_gpuasm_optimizations(),
hlo_module_config.debug_options().xla_gpu_cuda_data_dir()); hlo_module_config.debug_options().xla_gpu_cuda_data_dir(),
extra_flags);
} }
// Unimplemented for integers yet. // Unimplemented for integers yet.

5
tensorflow/compiler/xla/xla.proto
View File

@ -287,7 +287,10 @@ message DebugOptions {
// memory, or have bugs. // memory, or have bugs.
bool xla_gpu_unsafe_fallback_to_driver_on_ptxas_not_found = 138; bool xla_gpu_unsafe_fallback_to_driver_on_ptxas_not_found = 138;
// Next id: 141 // Extra parameters to pass the GPU assembler.
string xla_gpu_gpuasm_extra_flags = 141;
// Next id: 142
// Extra options to pass to the compilation backend (e.g. LLVM); specific // Extra options to pass to the compilation backend (e.g. LLVM); specific
// interpretation of these values is left to the backend. // interpretation of these values is left to the backend.