Skip to content

[Bugfix] Fix size calculation of processing cache #15114

New issue

Have a question about this project? No Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

No Sign up for GitHub

By clicking “No 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? No Sign in to your account

Jump to bottom
Merged
merged 3 commits into from
Mar 19, 2025
Merged

[Bugfix] Fix size calculation of processing cache #15114

merged 3 commits into from
Mar 19, 2025

Conversation

DarkLight1337
Copy link
Member

@DarkLight1337 DarkLight1337 commented Mar 19, 2025 •
edited by github-actions bot
Loading

The processing cache incorrectly calculated the item size as basically zero because it failed to recurse into the individual tensors, resulting in effectively infinite capacity.

FIX #15105
FIX #15085

@DarkLight1337
[Bugfix] Fix size calculation of processing cache
f82f326 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
@DarkLight1337 DarkLight1337 added the ready ONLY add when PR is ready to merge/full CI is needed label Mar 19, 2025
@DarkLight1337 DarkLight1337 requested a review from ywang96 as a code owner March 19, 2025 09:00
@github-actions GitHub Actions
Copy link

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

@mergify mergify bot added the multi-modality Related to multi-modality (#4194) label Mar 19, 2025
@DarkLight1337 DarkLight1337 mentioned this pull request Mar 19, 2025
Closed
1 task
ywang96
ywang96 reviewed Mar 19, 2025
View reviewed changes
Copy link
Member

@ywang96 ywang96 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Given this is now actually an important/potentially breaking feature, can we add some test on this?

@DarkLight1337 DarkLight1337 mentioned this pull request Mar 19, 2025
Closed
1 task
@DarkLight1337
Add test
f664339 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
@DarkLight1337
Copy link
Member Author

DarkLight1337 commented Mar 19, 2025 •
edited
Loading

Given this is now actually an important/potentially breaking feature, can we add some test on this?

Added in f664339

ywang96
ywang96 approved these changes Mar 19, 2025
View reviewed changes
Copy link
Member

@ywang96 ywang96 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Verified with benchmark script - Thanks for fixing this!

@DarkLight1337
Fix doc
3a7bb12 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
@DarkLight1337 DarkLight1337 mentioned this pull request Mar 19, 2025
Open
@vllm-bot vllm-bot merged commit 3d44643 into vllm-project:main Mar 19, 2025
28 of 30 checks passed
@DarkLight1337 DarkLight1337 deleted the fix-mm-processing-cache branch March 19, 2025 12:53
@engchina engchina mentioned this pull request Mar 19, 2025
Closed
1 task
gmarinho2 pushed a commit to gmarinho2/vllm that referenced this pull request Apr 1, 2025
@DarkLight1337 @gmarinho2
[Bugfix] Fix size calculation of processing cache (vllm-project#15114)
862b49d 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
lulmer pushed a commit to lulmer/vllm that referenced this pull request Apr 7, 2025
@DarkLight1337 @lulmer
[Bugfix] Fix size calculation of processing cache (vllm-project#15114)
d919aa3 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Louis Ulmer <ulmerlouis@gmail.com>
nishith-fujitsu pushed a commit to nishith-fujitsu/vllm that referenced this pull request Apr 9, 2025
@DarkLight1337 @nishith-fujitsu
[Bugfix] Fix size calculation of processing cache (vllm-project#15114)
ff536bf 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
@ckhordiasma ckhordiasma mentioned this pull request Apr 17, 2025
Closed
No Sign up for free to join this conversation on GitHub. Already have an account? No Sign in to comment
Reviewers

@ywang96 ywang96 ywang96 approved these changes

Assignees
No one assigned
Labels
multi-modality Related to multi-modality (#4194) ready ONLY add when PR is ready to merge/full CI is needed
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

[Bug]: Extremely slow inference + big waste of memory on 0.8.0 [Bug]: 0.8.0 ram leakage
3 participants
@DarkLight1337 @ywang96 @vllm-bot