OSSForks/tensorflow
1
0
Fork 0
mirror of https://github.com/zebrajr/tensorflow.git synced 2026-01-15 12:15:41 +00:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Use StartDetachedThread instead of SchedClosure to dispatch atom program compilation

Browse Source 
PiperOrigin-RevId: 847528854
This commit is contained in:
Junwhan Ahn
2025-12-21 17:25:22 -08:00
committed by TensorFlower Gardener
parent ce61030c67
commit 7733c4c03d
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
third_party/xla/xla/python/ifrt/ir/transforms/multi_threaded_atom_program_compiler.cc vendored
View File

@@ -180,7 +180,8 @@ absl::StatusOr<CompileFuture> MultiThreadedAtomProgramCompiler::CompileXla(
/*context=*/nullptr, // Shares the same long-living context.
mlir::OwningOpRef<mlir::ModuleOp>(module_op.clone()));
auto [promise, future] = CompileFuture::MakePromise();
tsl::Env::Default()->SchedClosure(
tsl::Env::Default()->StartDetachedThread(
tsl::ThreadOptions(), /*name=*/"MultiThreadedAtomProgramCompiler",
WithCurrentUserContext([this, hlo_program = std::move(hlo_program),
compile_options = std::move(compile_options),
promise = std::move(promise)]() mutable {