[CI][AMD]BugFix] Fix deadlock occuring in test_moe_layer (#40767)
Signed-off-by: Randall Smith <Randall.Smith@amd.com>
This commit is contained in:
@@ -1632,6 +1632,7 @@ def _parallel_worker(
|
||||
if all2all_manager is not None:
|
||||
all2all_manager.destroy()
|
||||
total = total + 1
|
||||
torch.distributed.barrier()
|
||||
|
||||
skipped = total - (passed + failed)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user