Skip to content
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

Jump to bottom

Setting enough cache_size_limit for torch.compile warmup #238

Merged
merged 10 commits into from
Sep 25, 2024
Merged

Setting enough cache_size_limit for torch.compile warmup #238

Changes from 2 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
c381661
Setting enough cache size limit for torch.compile
zehao-intel Sep 4, 2024
704c2b6
use self.enforce_eager
zehao-intel Sep 5, 2024
401e263
Update vllm/worker/habana_model_runner.py
zehao-intel Sep 9, 2024
08ac4ec
Update vllm/worker/habana_model_runner.py
zehao-intel Sep 9, 2024
9f81efb
Merge branch 'habana_main' into fix_recompile
zehao-intel Sep 11, 2024
7a82b73
select bigger value
zehao-intel Sep 11, 2024
fae0ffc
add explanatory note on why multiply 8
zehao-intel Sep 11, 2024
2d8f582
fix format
zehao-intel Sep 12, 2024
16ab8c4
plus one
zehao-intel Sep 24, 2024
cec53ee
fix yapf
zehao-intel Sep 24, 2024
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
5 changes: 5 additions & 0 deletions vllm/worker/habana_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -596,6 +596,11 @@ def _setup_buckets(self) -> None:
bucket for bucket in self.decode_buckets
if self._is_valid_bucket(bucket)
]
if not htorch.utils.internal.is_lazy() and not self.enforce_eager:
cache_size_limit = len(self.prompt_buckets)+len(self.decode_buckets)
zehao-intel marked this conversation as resolved.
Show resolved Hide resolved
torch._dynamo.config.cache_size_limit = cache_size_limit
torch._dynamo.config.accumulated_cache_size_limit = cache_size_limit*8
zehao-intel marked this conversation as resolved.
Show resolved Hide resolved

msg = (f"Generated {len(self.decode_buckets)} decode buckets: "
f"{list(sorted(self.decode_buckets))}")
logger.info(msg)
Expand Down