Skip to content

Upstream merge 2025 06 30 #586

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

Jump to bottom
Merged
merged 73 commits into from
Jul 1, 2025
Merged

Upstream merge 2025 06 30 #586

merged 73 commits into from
Jul 1, 2025

Conversation

gshtras
Copy link
Collaborator

@gshtras gshtras commented Jun 30, 2025

No description provided.

njhill and others added 30 commits June 25, 2025 13:38
@njhill
[Misc] Add parallel state node_count function (vllm-project#20045)
c40692b 
Signed-off-by: Nick Hill <[email protected]>
@QiliangCui
Fix the path to the testing script. (vllm-project#20082)
4e0db57 
Signed-off-by: Qiliang Cui <[email protected]>
@izhuhaoran
[Bugfix] default set cuda_graph_sizes to max_num_seqs for v1 engine (v…
9f0608f 
…llm-project#20062)

Signed-off-by: izhuhaoran <[email protected]>
@yaochengji
[TPU][Bugfix] fix kv cache padding (vllm-project#20048)
2cc2069 
Signed-off-by: Chengji Yao <[email protected]>
@njhill
[P/D] Avoid stranding blocks in P when aborted in D's waiting queue (v…
55c65ab 
…llm-project#19223)

Signed-off-by: Nick Hill <[email protected]>
@Chenyaaang
[TPU] Add TPU specific var VLLM_TPU_MOST_MODEL_LEN (vllm-project#19919)
2d7620c 
Signed-off-by: Chenyaaang <[email protected]>
@mgoin
[CI] Add SM120 to the Dockerfile (vllm-project#19794)
296ce95 
Signed-off-by: mgoin <[email protected]>
@mgoin
[Bugfix] Fix Mistral tool-parser regex for nested JSON (vllm-project#…
754b00e 
…20093)

Signed-off-by: mgoin <[email protected]>
@NickLucche
[PD] Skip tp_size exchange with rank0 (vllm-project#19413)
2582683 
Signed-off-by: NickLucche <[email protected]>
@ekagra-ranjan
[Benchmark][Bug] Fix multiple bugs in bench and add args to spec_deco…
9502c38 
…de offline (vllm-project#20083)
@eicherseiji
[Bugfix] Allow CUDA_VISIBLE_DEVICES='' in `Platform.device_id_to_ph…
65397e4 
…ysical_device_id` (vllm-project#18979)

Signed-off-by: Seiji Eicher <[email protected]>
@DarkLight1337
[Doc] Update docs for New Model Implementation (vllm-project#20115)
1d7c29f 
Signed-off-by: DarkLight1337 <[email protected]>
@yewentao256
[Refactor] Remove unused library (vllm-project#20099)
d188913 
Signed-off-by: yewentao256 <[email protected]>
@bigPYJ1151
[CPU] Fix torch version in x86 CPU backend (vllm-project#19258)
0567c82 
Signed-off-by: jiang1.li <[email protected]>
@reidliu41
[Misc] Use collapsible blocks for benchmark examples. (vllm-project#2…
167aca4 
…0017)

Signed-off-by: reidliu41 <[email protected]>
Co-authored-by: reidliu41 <[email protected]>
@windsonsea
[Docs] Improve frameworks/helm.md (vllm-project#20113)
84c260c 
Signed-off-by: windsonsea <[email protected]>
@tjtanaa
[Bugfix][V1][ROCm] Fix AITER Flash Attention Backend (Fix API Break a…
27c065d 
…nd Local Attention Logic: affecting Llama4) (vllm-project#19904)

Signed-off-by: tjtanaa <[email protected]>
@mgoin
Revert "[Bugfix] default set cuda_graph_sizes to max_num_seqs for v1 …
1f5d178 
…engine" (vllm-project#20128)
@yewentao256
[Bug Fix] Fix address/port already in use error for deep_ep test (vll…
c894c5d 
…m-project#20094)

Signed-off-by: yewentao256 <[email protected]>
@noooop
[Doc] Automatically signed-off by PyCharm (vllm-project#20120)
0907d50 
Signed-off-by: wang.yuqi <[email protected]>
@DarkLight1337
[Doc] Auto sign-off for VSCode (vllm-project#20132)
6393b03 
Signed-off-by: DarkLight1337 <[email protected]>
@DarkLight1337
[Doc] Rename page titles (vllm-project#20130)
34878a0 
Signed-off-by: DarkLight1337 <[email protected]>
@tlrmchlsmth
Spam folks if config.py changes (vllm-project#20131)
0bceac9 
Signed-off-by: Tyler Michael Smith <[email protected]>
@jikunshang
[Hardware][Intel GPU] Add v1 Intel GPU support with Flash attention b…
b69781f 
…ackend. (vllm-project#19560)

Signed-off-by: Kunshang Ji <[email protected]>
@yaochengji
[TPU] add kv cache update kernel (vllm-project#19928)
04e1642 
Signed-off-by: Chengji Yao <[email protected]>
@yewentao256
[Refactor] Rename commnication utils (vllm-project#20091)
5623088 
Signed-off-by: yewentao256 <[email protected]>
@kyolebu
[Doc] correct LoRA capitalization (vllm-project#20135)
07b8fae 
Signed-off-by: kyolebu <[email protected]>
@abmfy
[Feature] Expert Parallelism Load Balancer (EPLB) (vllm-project#18343)
e9fd658 
Signed-off-by: Bowen Wang <[email protected]>
@mgoin
[CI Failure] Fix OOM with test_oot_registration_embedding (vllm-proje…
71799fd 
…ct#20144)

Signed-off-by: mgoin <[email protected]>
@dsikka @kylesayrs
[Quantization] Bump to use latest compressed-tensors (vllm-project#…
a57d57f 
…20033)

Signed-off-by: Dipika <[email protected]>
Co-authored-by: Kyle Sayers <[email protected]>
s3woz and others added 19 commits June 28, 2025 08:16
@s3woz
[Bugfix] Correct behavior of GraniteMoeHybrid for TensorParallel exec…
daec9de 
…ution (vllm-project#20137)

Signed-off-by: Stanislaw Wozniak <[email protected]>
@yewentao256
[Refactor] Create a function util and cache the results for `has_deep…
4d36693 
…gemm`, `has_deepep`, `has_pplx` (vllm-project#20187)

Signed-off-by: yewentao256 <[email protected]>
@mgoin
[CI Fix] Try fixing eagle e2e test OOM by reducing block allocation (v…
7b1895e 
…llm-project#20213)

Signed-off-by: mgoin <[email protected]>
@dsikka
[Quantization] Add compressed-tensors NVFP4 MoE Support (vllm-project…
6f2f53a 
…#19990)

Signed-off-by: Dipika Sikka <[email protected]>
Signed-off-by: Dipika <[email protected]>
@huydhn
Fix cuda_archs_loose_intersection when handling sm_*a (vllm-project#2…
6c9837a 
…0207)

Signed-off-by: Huy Do <[email protected]>
@redmoe-moutain
[Model] support dots1 (vllm-project#18254)
65b1cbb 
Signed-off-by: redmoe-moutain <[email protected]>
@xuechendi
[BUGFIX][DEEPSEEK][MODEL_LOAD] fix w13, w2 weight not initialized ass…
5a52f38 
…ert (vllm-project#20202)

Signed-off-by: Chendi Xue <[email protected]>
@WoosukKwon
[Misc] Fix import (vllm-project#20233)
19108ef 
Signed-off-by: Woosuk Kwon <[email protected]>
@reidliu41
[doc] Add Slack and Forum to the top navigation (vllm-project#20208)
022c58b 
Signed-off-by: reidliu41 <[email protected]>
@noiji
[Bugfix] Skip loading extra parameters for modelopt Qwen3 MoE model (v…
f5dfa07 
…llm-project#19598)

Signed-off-by: noiji <>
@Isotr0py
[Bugfix] Fix processor initialization in transformers 4.53.0 (vllm-pr…
e936e40 
…oject#20244)

Signed-off-by: Isotr0py <[email protected]>
@jeejeelee
[Quantization] Improve BitsAndBytesModelLoader (vllm-project#20242)
8fe7fc8 
Signed-off-by: Jee Jee Li <[email protected]>
@windsonsea
[Docs] Fix 1-2-3 list in v1/prefix_caching.md (vllm-project#20243)
3ee56e2 
Signed-off-by: windsonsea <[email protected]>
@lihaoyang-amd @lixixicommute
[Bugfix] fix quark ptpc (vllm-project#20251)
1c50e10 
Signed-off-by: Haoyang Li <[email protected]>
Co-authored-by: Haoyang Li <[email protected]>
@WoosukKwon
[Spec Decode] Refactor spec decoding into a separate function (vllm-p…
2062c07 
…roject#20238)

Signed-off-by: Woosuk Kwon <[email protected]>
@WoosukKwon
[Spec Decode] Clean up spec decode example (vllm-project#20240)
2965c99 
Signed-off-by: Woosuk Kwon <[email protected]>
@WoosukKwon
[Optimization] Use Shared CachedRequestData Instance Across All Req…
2863bef 
…uests (vllm-project#20232)

Signed-off-by: Woosuk Kwon <[email protected]>
@yewentao256 @gemini-code-assist
[Unit Test] Add unit test for deep gemm (vllm-project#20090)
551ef16 
Signed-off-by: yewentao256 <[email protected]>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
@gshtras
Merge remote-tracking branch 'upstream/main'
d171777
@gshtras gshtras merged commit 0f7ec48 into main Jul 1, 2025
5 checks passed
@gshtras gshtras deleted the upstream_merge_2025_06_30 branch July 1, 2025 19:10
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@shajrawi shajrawi shajrawi approved these changes

@Alexei-V-Ivanov-AMD Alexei-V-Ivanov-AMD Awaiting requested review from Alexei-V-Ivanov-AMD

@maleksan85 maleksan85 Awaiting requested review from maleksan85

@sunway513 sunway513 Awaiting requested review from sunway513

@hongxiayang hongxiayang Awaiting requested review from hongxiayang

@divakar-amd divakar-amd Awaiting requested review from divakar-amd

@charlifu charlifu Awaiting requested review from charlifu

@mawong-amd mawong-amd Awaiting requested review from mawong-amd

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.