triton.experimental.gluon.language.amd.cdna5.async_copy

Functions

global_to_shared

Asynchronously copy elements from global memory to shared memory.

shared_to_global

Asynchronously copy elements from shared memory to global memory.

commit_group

Commit oustanding async operations.

wait_group

Wait for outstanding commit groups.

mbarrier_arrive

Arrive on the mbarrier once all outstanding async copies are complete.