[Doc]Add the user_guide doc file regarding fine-grained TP.#5084
Merged
jianzs merged 8 commits intovllm-project:mainfrom Dec 19, 2025
Merged
[Doc]Add the user_guide doc file regarding fine-grained TP.#5084jianzs merged 8 commits intovllm-project:mainfrom
jianzs merged 8 commits intovllm-project:mainfrom
Conversation
Signed-off-by: zzhx1 <zzh_201018@outlook.com>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request adds user documentation for the Fine-Grained Tensor Parallelism feature. The documentation is comprehensive, covering the overview, benefits, usage, and deployment recommendations. However, I've identified a few critical issues. The documentation for mlp_tensor_parallel_size is misleading due to an underlying bug where its value is incorrectly tied to embedding_tensor_parallel_size. Additionally, the documentation for o_proj is missing a key restriction regarding its use in prefill-decode disaggregation scenarios. Finally, there is a syntax error in one of the configuration examples. These issues should be addressed to ensure the documentation is accurate and prevents user errors.
|
|
||
| ### Component & Execution Mode Support | ||
| - **`embedding`, `lm_head`, and `mlp`**: Can be configured with fine-grained TP in any execution context—prefill, decode, or mixed deployment. | ||
| - **`o_proj`**: Currently, fine-grained TP for the attention output projection is **only supported in graph-capture mode** (e.g., CUDA Graphs). It cannot be enabled in eager execution. |
Contributor
There was a problem hiding this comment.
The documentation for o_proj fine-grained TP is incomplete. The implementation in vllm_ascend/ascend_config.py (lines 195-198) reveals an additional critical restriction: oproj_tensor_parallel_size is only supported in prefill-decode (PD) disaggregation scenarios and can only be used on the decode (consumer) nodes. This should be explicitly mentioned to prevent users from encountering runtime errors in unsupported configurations.
I suggest updating the line to be more explicit about all restrictions:
- **`o_proj`**: Currently, fine-grained TP for the attention output projection has two key restrictions:
- It is **only supported in graph-capture mode** (e.g., CUDA Graphs) and cannot be enabled in eager execution.
- It is **only supported in prefill-decode (PD) disaggregation scenarios and must be used on a decode (consumer) node**.
Contributor
|
👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:
If CI fails, you can run linting and testing checks locally according Contributing and Testing. |
github-actions
Bot
added
the
documentation
Signed-off-by: zzhx1 <zzh_201018@outlook.com>
zzhx1
force-pushed
the
finegrained-TP
branch
from
ff932a5 to
d700184
Compare
Signed-off-by: 秋刀鱼 <jaychou1620@Gmail.com>
zzhx1
changed the title
jianzs
approved these changes
Dec 19, 2025
Signed-off-by: zzhx1 <zzh_201018@outlook.com>
845473182
pushed a commit
to 845473182/vllm-ascend
that referenced
this pull request
Dec 19, 2025
…to eplb_refactor * 'main' of https://github.com/vllm-project/vllm-ascend: (52 commits) [Doc]Add the user_guide doc file regarding fine-grained TP. (vllm-project#5084) [pref] qwen3_next add triton ops : fused_sigmoid_gating_delta_rule_update (vllm-project#4818) [Feature] Add token mask for DispatchGmmCombineDecode operator (vllm-project#5171) [CI] Improve CI (vllm-project#5078) [Refactor] remove some metadata variables in attention_v1. (vllm-project#5160) Add Qwen3-VL-235B-A22B-Instruct tutorials (vllm-project#5167) [Doc] Add a perf tune section (vllm-project#5127) [Image] Refactor image build (vllm-project#5175) [refactor] refactor weight trans nz and transpose (vllm-project#4878) [BugFix]Fix precision issue for LoRA feature (vllm-project#4141) 【Doc】Deepseekv3.1/R1 doc enhancement (vllm-project#4827) support basic long_seq feature st (vllm-project#5140) [Bugfix] install trition for test_custom_op (vllm-project#5112) [2/N][Pangu][MoE] Remove Pangu Related Code (vllm-project#5130) [bugfix] Use FUSED_MC2 MoE comm path for the op `dispatch_ffn_combine` (vllm-project#5156) [BugFix] Fix top_p,top_k issue with EAGLE and add top_p,top_k in EAGLE e2e (vllm-project#5131) [Doc][P/D] Fix MooncakeConnector's name (vllm-project#5172) [Bugfix] Fix in_profile_run in mtp_proposer dummy_run (vllm-project#5165) [Doc] Refact benchmark doc (vllm-project#5173) [Nightly] Avoid max_model_len being smaller than the decoder prompt to prevent single-node-accuray-tests from failing (vllm-project#5174) ... Signed-off-by: 白永斌 <baiyongbin3@h-partners.com>
chenaoxuan
pushed a commit
to chenaoxuan/vllm-ascend
that referenced
this pull request
Dec 20, 2025
…ject#5084) ### What this PR does / why we need it? Add user guide for **Fine-Grained Tensor Parallelism** feature. Documents usage, supported components (`embedding`, `lm_head`, `o_proj`, `mlp`/`dense_ffn`), model compatibility, and deployment guidelines. - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e --------- Signed-off-by: zzhx1 <zzh_201018@outlook.com> Signed-off-by: chenxiao <Jaychou1620@Gmail.com> Signed-off-by: 秋刀鱼 <jaychou1620@Gmail.com> Co-authored-by: chenxiao <Jaychou1620@Gmail.com> Co-authored-by: Jade Zheng <zheng.shoujian@outlook.com>
ZRJ026
pushed a commit
to ZRJ026/vllm-ascend
that referenced
this pull request
Feb 28, 2026
…ject#5084) ### What this PR does / why we need it? Add user guide for **Fine-Grained Tensor Parallelism** feature. Documents usage, supported components (`embedding`, `lm_head`, `o_proj`, `mlp`/`dense_ffn`), model compatibility, and deployment guidelines. - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e --------- Signed-off-by: zzhx1 <zzh_201018@outlook.com> Signed-off-by: chenxiao <Jaychou1620@Gmail.com> Signed-off-by: 秋刀鱼 <jaychou1620@Gmail.com> Co-authored-by: chenxiao <Jaychou1620@Gmail.com> Co-authored-by: Jade Zheng <zheng.shoujian@outlook.com> Signed-off-by: zrj026 <zhangrunjiang026@gmail.com>
ZRJ026
pushed a commit
to ZRJ026/vllm-ascend
that referenced
this pull request
Mar 4, 2026
…ject#5084) ### What this PR does / why we need it? Add user guide for **Fine-Grained Tensor Parallelism** feature. Documents usage, supported components (`embedding`, `lm_head`, `o_proj`, `mlp`/`dense_ffn`), model compatibility, and deployment guidelines. - vLLM version: v0.12.0 - vLLM main: vllm-project/vllm@ad32e3e --------- Signed-off-by: zzhx1 <zzh_201018@outlook.com> Signed-off-by: chenxiao <Jaychou1620@Gmail.com> Signed-off-by: 秋刀鱼 <jaychou1620@Gmail.com> Co-authored-by: chenxiao <Jaychou1620@Gmail.com> Co-authored-by: Jade Zheng <zheng.shoujian@outlook.com> Signed-off-by: zrj026 <zhangrunjiang026@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this PR does / why we need it?
Add user guide for Fine-Grained Tensor Parallelism feature.
Documents usage, supported components (
embedding,lm_head,o_proj,mlp/dense_ffn), model compatibility, and deployment guidelines.Functionality implemented in:
add mlp tp optimze #2120 (MLP TP)
[feat]: oproj tensor parallelism in pure DP and graph-mode scenarios. #2167 (OProj TP)
[Feat]: Add custom lmhead tensor model parallel #2309 (LM Head TP)
[Feat] Add custom Embedding tensor model parallel #2616 (Embedding TP)
[Feat] Support MLP_TP feature, exclude MOE layer #4999 (Dense FFN TP)
vLLM version: v0.12.0
vLLM main: vllm-project/vllm@ad32e3e