vllm-project
[Misc]: Implement CPU/GPU swapping in BlockManagerV2
A draft PR for #3666, supporting CPU-GPU swapping for Block Manager V2.
PR Checklist (Click to Expand)
Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.
PR Title and Classification
Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:
[Bugfix]for bug fixes.[CI/Build]for build or continuous integration improvements.[Doc]for documentation fixes and improvements.[Model]for adding a new model or improving an existing model. Model name should appear in the title.[Frontend]For changes on the vLLM frontend (e.g., OpenAI API server,LLMclass, etc.)[Kernel]for changes affecting CUDA kernels or other compute kernels.[Core]for changes in the core vLLM logic (e.g.,LLMEngine,AsyncLLMEngine,Scheduler, etc.)[Hardware][Vendor]for hardware-specific changes. Vendor name should appear in the prefix (e.g.,[Hardware][AMD]).[Misc]for PRs that do not fit the above categories. Please use this sparingly.
Note: If the PR spans more than one category, please include all relevant prefixes.
Code Quality
The PR need to meet the following code quality standards:
- We adhere to Google Python style guide and Google C++ style guide.
- Pass all linter checks. Please use
format.shto format your code. - The code need to be well-documented to ensure future contributors can easily understand the code.
- Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.
- Please add documentation to
docs/source/if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.
Notes for Large Changes
Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with rfc-required and might not go through the PR.
What to Expect for the Reviews
The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:
- After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.
- After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.
- After the review, the reviewer will put an
action-requiredlabel on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR. - Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion.
Thank You
Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!
I passed through the code. I think this PR is mixing the abstraction boundaries too much. I wrote out some pseudocode for a better way of using the existing APIs -- take a look.
https://docs.google.com/document/d/157VCEOs5EIB1aztD51ZrT2RvdjxMx1spazyBM4LS4aQ/edit
Is it ready for review again?
Nope, still working on the test, will push the new version later today or tmr and notify you. Thanks for the attention!
@cadedaniel Hi cade, i think it's ready for another round of review. One small point is currently the preemption_mode parameter in _preempt() are default to RECOMPUTE and never be passed in. Add some tiny changes around that to support testing preemption by swapping, should we add those in an separate MR?
@cadedaniel Hi cade, i think it's ready for another round of review. One small point is currently the preemption_mode parameter in _preempt() are default to RECOMPUTE and never be passed in. Add some tiny changes around that to support testing preemption by swapping, should we add those in an separate MR?
It's preferable to put it in a separate pull request, but for this one let's try to get it in quickly. I will review !
One missing item from this PR is better testing; the E2E tests and unit test you have are great but they don't allow testing edge cases that could be done with a unit test at the
cpu gpu block allocatororblock allocatorlevel.
@cadedaniel For the testing at cpu_gpu_block_allocator level, is there anything else other then testing the get_num_blocks_touched? It seems that is the only thing that affect can_swap(), and the the correctness of swap() should be verified by the e2e tests. WDYT?
The e2e test has good coverage but it's bad at verifying lower level properties in edge cases. Eg that swapped block ids are correct. We want to have coverage of the happy path (done) and the important edge cases the happy path doesn't see. I can look later and provide more specific test outline for you.
We want to have coverage of the happy path (done) and the important edge cases the happy path doesn't see. I can look later and provide more specific test outline for you.
Maybe just give a few hints about what to test, I am happy to add those comprehensive tests tmr.
@cadedaniel Hi cade, just rebased the master into my branch and left a TODO for the test. Should we merge this one first and get a follow up PR about the tests?
SG, I will take a look by Monday
@cadedaniel Sure! Btw is there any way we can rerun the CI tests? It seems there's always random failures not because of the bugs but environment.
I retried the AMD ones for you. Best way is to push an empty commit to restart things. If it keeps happening with AMD let's see if we should auto retry or we should fix them.
Merge conflicts but tests passing - @Kaiyang-Chen can you rebase?
@cadedaniel @richardliaw rebased, but still some randomly failed tests.
retrying
Hi Cade, can you help to retry the last failed test one more time? Thanks!
@cadedaniel Hi cade, solved the new conflicts. Can you help to retry and get it merged? TY!
thanks @Kaiyang-Chen . i was out last week. I will get this merged this week, thanks for all your work!
Enabled auto merge
@cadedaniel It seems this cannot be auto merged because I don't have write access to the repo?