We do not generate _serialized_parallel calls in device mode, no
need for an external API.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Comment Actions
I thought these were used for nested parallel clauses, but if clang doesn't emit calls to them, better deleted
Comment Actions
Yep, dead - looks like the calls from the gpu were removed when __kmpc_parallel_51 was introduced