[Doc] fix DSV3.1 PD configs#7187
Conversation
Signed-off-by: chenmenglong <chenmenglong1@huawei.com>
dragondream-chen
requested review from
LCAIZJ,
Yikun and
wangxiyuan
as code owners
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request rectifies an inaccuracy in the DeepSeek-V3.1 documentation by updating specific configuration parameters. The changes ensure that the provided example for the 2P1D scenario accurately reflects the correct Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Changelog
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
github-actions
bot
added
the
documentation
|
👋 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. |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request fixes an inconsistency in the Prefill-Decode Disaggregation configuration for DeepSeek-V3.1 in the documentation. The engine_id and kv_port for the second decoder node are updated to match the first, which is correct for a single logical decoder engine spanning multiple nodes. However, my review found that all kv_port values in the document conflict with the recommended port range specified in other documentation for 16-NPU nodes. This could lead to port conflicts and runtime errors. I've added a comment with a suggestion to use ports outside the reserved range to ensure stability.
| "kv_port": "30200", | ||
| "engine_id": "2", |
There was a problem hiding this comment.
This change correctly makes the engine_id consistent for the logical decoder engine. However, the kv_port values used throughout this document appear to conflict with the guidance in pd_disaggregation_mooncake_multi_node.md.
For 16-NPU nodes like the Atlas 800 A3, that guide recommends kv_port >= 36000 to avoid port conflicts with AscendDirectTransport, which can cause zmq.error.ZMQError: Address already in use. The ports 30000, 30100, and 30200 used in this document are within the reserved range.
I recommend updating all kv_port values in this file to follow that guidance. For example, you could use 36000 and 36100 for the prefill nodes, and 36200 for the decode nodes.
Here is a suggested change for this block. Please apply similar changes to the other nodes' configurations in this file.
| "kv_port": "30200", | |
| "engine_id": "2", | |
| "kv_port": "36200", | |
| "engine_id": "2", |
…to qwen3next_graph * 'main' of https://github.com/vllm-project/vllm-ascend: (88 commits) [main][bugfix] Fixed the problem of speculative decoding in FULL mode (vllm-project#7148) fixed fia pad logic in graph mode. (vllm-project#7144) [Doc] fix DSV3.1 PD configs (vllm-project#7187) refactor: add a check before layer_sharding logging (vllm-project#7186) [Build] Add support for Ascend950 chip (vllm-project#7151) Revert "[CI] fix skiped e2e test when upgrade vllm version (vllm-project#6654)" (vllm-project#7166) [MODELRUNNERV2]fix penality ops (vllm-project#7013) [Bugfix][LoRA] Fix the issue when enable LoRA + tp + fully_sharded_loras (vllm-project#6650) [KV Pool]get_num_new_matched_tokens return 0 if token length < block_size (vllm-project#7146) [CI] Build Image for v0.16.0rc1 (vllm-project#7155) [CI] Skip `test_mooncake_layerwise_connector.py` in `ut` (vllm-project#7147) [BugFix]Fix recomputed scheduler bug (vllm-project#7137) [Model] Support Minimax-m2.5 on NPU (vllm-project#7105) [P/D]Mooncake Layerwise Connector supports hybrid attention manager with multiple kvcache groups (vllm-project#7022) Add patch_qwen3_5 for triton ops fused_recurrent_gated_delta_rule (vllm-project#7109) [Doc][ReleaseNote] Add release notes for v0.16.0rc1 (vllm-project#7067) [Misc] Download on both hk and guiyang region (vllm-project#7129) [bugdix] The problem that the w4a8 weight fails to be loaded when the EP is not enabled is resolved. (vllm-project#7090) [eagle][cp] fix eagle_cp enable bug2 (vllm-project#7079) [CI]Upgrade niglty multi-node-tests max-parallel to 2 (vllm-project#7035) ...
### What this PR does / why we need it? Modify the `kv_port` and `engine_id` config of DeepSeek-V3.1/R1 in the 2P1D scenario - vLLM version: v0.16.0 - vLLM main: vllm-project/vllm@4034c3d Signed-off-by: chenmenglong <chenmenglong1@huawei.com>
2 participants
What this PR does / why we need it?
Modify the
kv_portandengine_idconfig of DeepSeek-V3.1/R1 in the 2P1D scenarioDoes this PR introduce any user-facing change?
How was this patch tested?