Skip to content

[BUG] MM LRU cache fixing #16599

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
wants to merge 1 commit into from
Closed

[BUG] MM LRU cache fixing #16599

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
7 changes: 7 additions & 0 deletions vllm/v1/engine/mm_input_cache.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,13 @@ def get_and_update_p0(
full_mm_inputs = list[Optional[MultiModalKwargs]]()
for mm_input, mm_hash in zip(mm_inputs, mm_hashes):
if mm_hash in self.mm_cache:
# Client and Server must be exactly the same (see description
# in the top of this file).
# `in` in above statement don't update access time by design.
# But server side make a direct access and update access time.
# Have to make a dummy access to update access time to keep
# LRU order of caches consistent.
_ = self.mm_cache[mm_hash]
mm_input = None
else:
self.mm_cache[mm_hash] = mm_input
Expand Down