Skip to content

[CI][CPU] Fix worker shutdown crash on CPU backend #41130

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
haosdent wants to merge 1 commit into from
Closed

[CI][CPU] Fix worker shutdown crash on CPU backend #41130

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions vllm/v1/worker/cpu_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -138,6 +138,9 @@ def _init_device_properties(self) -> None:
def _sync_device(self) -> None:
pass

def _empty_cache(self) -> None:
pass

def _zero_block_ids(self, block_ids: list[int]) -> None:
# CPU attention assigns -INF to logits at invalid positions,
# so stale KV cache data never affects computation.
Expand Down
8 changes: 5 additions & 3 deletions vllm/v1/worker/gpu_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1051,6 +1051,9 @@ def _init_device_properties(self) -> None:
def _sync_device(self) -> None:
torch.accelerator.synchronize()

def _empty_cache(self) -> None:
torch.accelerator.empty_cache()

def _get_or_create_async_output_copy_stream(self) -> torch.cuda.Stream:
stream = self.async_output_copy_stream
if stream is None:
Expand Down Expand Up @@ -5903,7 +5906,6 @@ def shutdown(self) -> None:
from vllm.model_executor.layers.rotary_embedding import _ROPE_DICT
from vllm.v1.worker.workspace import reset_workspace_manager

# Calls torch.accelerator.synchronize()
self._cleanup_profiling_kv_cache()
self.compilation_config.static_forward_context.clear()
self.model = None # type: ignore[assignment]
Expand All @@ -5912,7 +5914,7 @@ def shutdown(self) -> None:
reset_workspace_manager()

def _cleanup_profiling_kv_cache(self) -> None:
torch.accelerator.synchronize()
self._sync_device()
if hasattr(self, "kv_caches") and self.kv_caches:
for i in range(len(self.kv_caches)):
self.kv_caches[i] = None # type: ignore
Expand Down Expand Up @@ -5941,7 +5943,7 @@ def _cleanup_profiling_kv_cache(self) -> None:
layer.impl._v_scale_cache = None

gc.collect()
torch.accelerator.empty_cache()
self._empty_cache()

logger.debug("Cleaned up profiling KV cache and CUDA graphs")

Expand Down
Loading