[Hardware][CPU] Support chunked-prefill and prefix-caching on CPU #10355
Conversation
|
👋 Hi! Thank you for contributing to the vLLM project. 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 do one of these:
🚀 |
mergify
bot
added
documentation
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
e9ddc0e to
d33a175
Compare
| # reset distributed env properly. Use a value > 1 just when you test. | ||
| @pytest.mark.parametrize("tensor_parallel_size", [1]) | ||
| @pytest.mark.parametrize("attention_backend", ["TORCH_SDPA"]) | ||
| @pytest.mark.cpu_only |
There was a problem hiding this comment.
Can you define this mark in pyproject.toml?
There was a problem hiding this comment.
Let's use the cpu_model defined before.
|
I'll leave the detailed review to @Isotr0py and @WoosukKwon . |
|
This pull request has merge conflicts that must be resolved before it can be |
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
96cafe5 to
a7534ec
Compare
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
a7534ec to
f43c276
Compare
|
Seems that the kernel tests are failing, perhaps we need merging from main branch? |
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
f43c276 to
3fb5054
Compare
|
Hmm, seems the kernel test are broken before, they were also failed in some previous commits. |
|
cc @youkaichao Seems that VLLM config is not initialized in kernel tests. |
should be fixed by #10431 . please merge main to fix the test error. |
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
4 times, most recently
from
e219bab to
eb1c775
Compare
|
Should be fixed now. |
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
eb1c775 to
883c6f5
Compare
|
Hmm, CPU tests failed. |
|
@DarkLight1337 ops, a conflict with #10441... |
Head branch was pushed to by a user without write access
|
@DarkLight1337 The failed tests are not related to this PR. Can we ignore these cases and merge this PR manually? |
|
I don't have permission to force merge. Can you just merge from main? |
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Signed-off-by: jiang1.li <[email protected]>
Head branch was pushed to by a user without write access
bigPYJ1151
force-pushed
the
upstream_chunked_prefill
branch
from
504c82b to
fda58f7
Compare
|
@DarkLight1337 No problem, let's try again and pray for all cases green😂 |
…lm-project#10355) Signed-off-by: jiang1.li <[email protected]> Signed-off-by: Tyler Michael Smith <[email protected]>
…lm-project#10355) Signed-off-by: jiang1.li <[email protected]>
This PR provides CPU chunked-prefill and prefix-caching support. For now only FP32 and BF16 are supported, ipex-2.6 will provide the FP16 support.
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:
format.shto format your code.docs/source/if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.Adding or changing kernels
Each custom kernel needs a schema and one or more implementations to be registered with PyTorch.
Tensorsrequire meta-functions. Meta-functions should be implemented and registered in python so that dynamic dims can be handled automatically. See above documents for a description of meta-functions.torch.libary.opcheck()to test the function registration and meta-function for any registered ops. Seetests/kernelsfor examples.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-requiredand 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:
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.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!