triton.experimental.gluon.language.nvidia.rubin.tcgen05_commit

triton.experimental.gluon.language.nvidia.rubin.tcgen05_commit(barrier, pred=True, descs=(), _semantic=None)

This instruction causes the provided mbarrier to be arrived-on with a count of 1 when all async tcgen05 MMA and copy instructions previously issued by the thread are complete.

If descs are provided, the commit will be multicast across the CTA cluster based on the shared layouts of those descriptors. This should be used when the inputs to the tcgen5 MMA come from TMA descriptors using multicast.

Parameters:

• barrier (shared_memory_descriptor) – The barrier to track completion of tcgen05 MMA and copy instructions.

• pred (bool) – Scalar predicate. Operation is skipped if predicate is False. Defaults to True.

• descs (Sequence[shared_memory_descriptor]) – Shared memory descriptors for the preceding multiplication inputs. Defaults to ().