15 Commits

Author SHA1 Message Date
Zetong Li
52abd47f8c [Bugfix][SHM] Use writer lock by default and remove redundant env (#4117)
### What this PR does / why we need it?
This PR aims to remove env introduced by #3988 and use lock by default.
As described in https://github.com/vllm-project/vllm/issues/27858, we
have tested the writer lock method in various scenarios and the
performance is almost unaffected. Therefore, we believe that it would be
safe to enable the lock by default and remove the redundant env
`SHM_BARRIER` now.

After discussion, we decide to preserve env and set it as true by
default.

### Does this PR introduce _any_ user-facing change?
`SHM_BARRIER` is set as true by default.

### How was this patch tested?
by ci

---------

Signed-off-by: Zetong Li <slippersss@126.com>
2025-12-01 22:27:01 +08:00
wangxiyuan
b321e3846a [cherry-pick]【main】patch sched_yield (#3648) (#3687)
### What this PR does / why we need it?
On Arm systems, os.sched_yield() does not take effect, causing the GIL
(Global Interpreter Lock) to remain unrelinquished and resulting in CPU
bound issues. This PR applies a patch to sched_yield in vLLM, making the
process execute time.sleep(0) instead to release the GIL. ### Does this
PR introduce _any_ user-facing change?

Signed-off-by: fems14 <1804143737@qq.com>
Co-authored-by: fems14 <74094523+fems14@users.noreply.github.com>
2025-10-24 00:24:58 +08:00
ZYang6263
6975d46627 [v0.11.0][Perf] Eliminating the zerolike operator through patch (#3632)
### What this PR does / why we need it?
There is a zero-like operator before the attention operation in each
decoding stage. After analysis, this operator can be eliminated. The
purpose of this PR is to remove this operator and improve performance.

---------

Signed-off-by: ZYang6263 <zy626375@gmail.com>
2025-10-23 14:49:28 +08:00
whx
6464c97ff9 [BugFix][v0.11.0] Fix quantization related mtp bug with patch (#3619)
vLLM 0.11.0 didn't bring PR
(https://github.com/vllm-project/vllm/pull/25805) thus missing the
prefix of mtp's SharedHead. This PR fixes this bug with a patch to
vllm's deepseek_mtp.

---------

Signed-off-by: whx-sjtu <2952154980@qq.com>
2025-10-22 23:06:09 +08:00
wangxiyuan
13e8e75143 [Refactor] refactor patch module (#3555)
### What this PR does / why we need it?
we notice that `patch_main` is never used. Usually the patch is for all
version. And if it's for specified version, we can use `vllm_version_is`
instead. So let's remove the useless sub folder in patch module to make
it clear.


- vLLM version: v0.11.0rc3
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-10-21 20:19:46 +08:00
Mengqing Cao
61866b8ac6 [Quickfix] update CachedRequestState as NewRequestData changed (#2367)
### What this PR does / why we need it?
1. update `CachedRequestState` as `NewRequestData` changed in
https://github.com/vllm-project/vllm/pull/22570
2. drop maintenance of vllm v0.10.0 in the branch main

### Does this PR introduce _any_ user-facing change?
N/A

### How was this patch tested?
CI passed with existing test.


- vLLM version: v0.10.0
- vLLM main:
92ff41abea

---------

Signed-off-by: MengqingCao <cmq0113@163.com>
2025-08-15 07:35:27 +08:00
Yikun Jiang
17a430f7b8 Upgrade vLLM to v0.10.0 (#1927)
### What this PR does / why we need it?
- Upgrade to v0.10.0
- Drop v0.9.2 version compatibility
- Add patch for
`vllm_ascend/patch/worker/patch_common/patch_sampler_gather_logprobs.py`
as workaround of
f3a683b7c9
for v0.10.0 and also add e2e test `test_models_prompt_logprobs`
- Pin transformers<4.54.0 as workaround of
https://github.com/vllm-project/vllm-ascend/issues/2034

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
- Test locally:
`VLLM_USE_MODELSCOPE=true pytest -sv
tests/e2e/singlecard/test_offline_inference.py::test_models_prompt_logprobs`
- CI passed

- vLLM version: v0.9.2
- vLLM main:
7728dd77bb

---------

Signed-off-by: Yikun Jiang <yikunkero@gmail.com>
2025-07-26 15:43:29 +08:00
wangxiyuan
830332ebfc Clean up v0.9.1 code (#1672)
vllm has released 0.9.2. This PR drop 0.9.1 support.

- vLLM version: v0.9.1
- vLLM main:
b942c094e3

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-07-09 08:52:24 +08:00
wangxiyuan
4f5964420e [CI] Upgrade vllm to 0.9.1 (#1165)
1. upgrade vllm to 0.9.1. 0.9.0 is not supported for main branch now.
keep doc to 0.9.0 until we release the first 0.9.1 release.
2. disable V0 test for PR
3. move actionlint check to lint job

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-06-11 16:33:11 +08:00
wangxiyuan
f6e5decc10 [CI] upgrade to vllm 0.9.0 (#959)
Upgrade to vllm 0.9.0.
0.8.5 will not be supported any more.

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-05-28 21:18:41 +08:00
Yikun Jiang
afe1767c17 [Core] Cleanup triton patch which has been fixed in vllm (#764)
### What this PR does / why we need it?
- Revert "Re-patch TritonPlaceholder on main to make CI happy (#753)"
because upstream main CI already merged:
https://github.com/vllm-project/vllm/pull/17446
- Keep 0.8.5.post1 compatible

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
CI passed

---------

Signed-off-by: Yikun Jiang <yikunkero@gmail.com>
2025-05-06 18:52:15 +08:00
Yikun Jiang
d7e1110c8e Re-patch TritonPlaceholder on main to make CI happy (#753)
### What this PR does / why we need it?
Re-patch TritonPlaceholder on main to make CI happy
- Add triton patch back until
https://github.com/vllm-project/vllm/pull/17446 resolved
- Move patch_main before patch_common to resolve minicpm triton import
issue
- Add `0.8.5` and `0.8.5.post1` to make patch work on 0.8.5 all versions

Related:
- https://github.com/vllm-project/vllm-ascend/pull/704
- https://github.com/vllm-project/vllm-ascend/pull/690

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
All CI passed include main

Signed-off-by: Yikun Jiang <yikunkero@gmail.com>
2025-05-05 23:22:24 +08:00
wangxiyuan
f8350569e6 [CI] upgrade vllm to 0.8.5 (#715)
1. Upgrade vllm to 0.8.5
2. Drop 0.8.4 support
3. Keep doc to 0.8.4rc2 until we release 0.8.5

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-04-30 09:15:50 +08:00
wangxiyuan
42c7fbb10e [Misc] Fix import error and address nits to make CI happy (#563)
1. Add `vllm_version_is` function to check vllm version.
2. `ensure_kv_transfer_initialized` and `get_kv_transfer_group ` have
been moved to other place in vllm main branch via
3408e47159
, this patch fix the import error.

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
2025-04-18 12:23:32 +08:00
wangxiyuan
bbe7ccd366 [MISC] Add patch module (#526)
This PR added patch module for vllm
1. platform patch: the patch will be registered when load the platform
2. worker patch: the patch will be registered when worker is started.

The detail is:
1. patch_common: patch for main and 0.8.4 version
4. patch_main: patch for main verison
5. patch_0_8_4: patch for 0.8.4 version
2025-04-16 09:28:58 +08:00