[ROCm][Perf] Enable layer-aware CSA2 multi-stream overlap for DeepSeek-V4.1-Flash (#57407)

Signed-off-by: Shanshan Shen <87969357+shen-shanshan@users.noreply.github.com>
This commit is contained in:
Shanshan Shen
2026-09-27 23:48:01 +08:00
committed by GitHub
parent 924707f1bf
commit 44af287ebe
4 changed files with 496 additions and 20 deletions
+9 -2
View File
@@ -70,6 +70,7 @@ def execute_in_parallel(
done_events: list[torch.cuda.Event],
aux_streams: list[torch.cuda.Stream] | None = None,
enable: bool = False,
default_first: bool = False,
) -> tuple[Any, list[Any]]:
"""Run default_fn on the current stream and aux_fns concurrently on
aux_streams.
@@ -97,6 +98,10 @@ def execute_in_parallel(
so callers that pass aux_streams must also pass enable=True
(typically gated by an env var) to actually overlap. When False,
execution falls back to sequential on the current stream.
default_first: When True, enqueue the default chain before the aux
chains. The default chain is typically the longest, so it starts
first; the aux streams still only wait on start_event, so they
overlap the default chain. Defaults to False.
Returns:
Tuple of (default_result, aux_results) where aux_results[i] is the
@@ -117,6 +122,8 @@ def execute_in_parallel(
pending: list[torch.cuda.Event] = []
start_event.record()
if default_first:
default_result = default_fn()
for i, fn in enumerate(aux_fns):
if fn is None:
continue
@@ -125,8 +132,8 @@ def execute_in_parallel(
aux_results[i] = fn()
done_events[i].record()
pending.append(done_events[i])
default_result = default_fn()
if not default_first:
default_result = default_fn()
for ev in pending:
ev.wait()