[xpu]feat: support multi-lora on xpu #20616

yma11 · 2025-07-08T07:54:19Z

This PR enables multi-lora support on xpu platform, passing UT like pytest -s -v tests/lora/test_quant_model.py

Signed-off-by: yan <yan.ma@intel.com>

gemini-code-assist

Summary of Changes

Hello @yma11, 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 introduces comprehensive support for multi-LoRA functionality on the XPU platform. It involves adapting core LoRA operations for platform-specific dispatch, refining tensor handling for multiple LoRA weights, and enabling XPU-aware model deserialization. The changes also include XPU-specific configurations and a temporary workaround to ensure stability with torch.compile.

Highlights

XPU LoRA Operations Dispatching: Modified lora_expand_op.py and lora_shrink_op.py to use current_platform.dispatch_key when registering custom operations. This enables platform-specific implementations for LoRA expand and shrink operations, crucial for XPU compatibility.
Multi-LoRA Pointer Handling Improvement: Updated utils.py to explicitly cast tensor pointers and slice offsets to torch.uint64 when creating tensors for multiple LoRA weights. This change enhances robustness and precision, particularly important for memory addressing on XPU.
XPU Model Deserialization Support: Adjusted tensorizer.py to dynamically select the appropriate device (xpu or cuda) during model deserialization. This ensures that models are loaded onto the correct hardware platform based on current_platform.is_xpu().
XPU Platform Configuration and Workarounds: Added a get_punica_wrapper method to xpu.py to specify the Punica wrapper path for XPU. Additionally, a temporary measure was introduced to disable torch.compile (forcing eager mode) for XPU under specific conditions, addressing current stability considerations with compilation on the platform.

Using Gemini Code Assist

The 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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

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 .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist is currently in preview and 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 to provide feedback.

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.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This pull request introduces multi-LoRA support for the XPU platform. My feedback focuses on improving code style and readability for better maintainability.

gemini-code-assist · 2025-07-08T07:56:09Z

vllm/model_executor/model_loader/tensorizer.py

+                device=f'xpu:{torch.xpu.current_device()}'
+                if current_platform.is_xpu() else
+                f'cuda:{torch.cuda.current_device()}',


This ternary expression can be written on a single line for better readability.

device=f'xpu:{torch.xpu.current_device()}' if current_platform.is_xpu() else f'cuda:{torch.cuda.current_device()}',

gemini-code-assist · 2025-07-08T07:56:09Z

vllm/platforms/xpu.py

+        if envs.VLLM_USE_V1 and vllm_config.model_config is not None and \
+            not vllm_config.model_config.enforce_eager:


Use parentheses for grouping expressions over multiple lines to improve readability and avoid potential issues with trailing whitespace.

Suggested change

if envs.VLLM_USE_V1 and vllm_config.model_config is not None and \

not vllm_config.model_config.enforce_eager:

if (envs.VLLM_USE_V1 and vllm_config.model_config is not None and

not vllm_config.model_config.enforce_eager):

gemini-code-assist · 2025-07-08T07:56:09Z

vllm/platforms/xpu.py

+            vllm_config.compilation_config.level = \
+                CompilationLevel.NO_COMPILATION


This assignment can be written on a single line for better readability.

vllm_config.compilation_config.level = CompilationLevel.NO_COMPILATION

Signed-off-by: yan <yan.ma@intel.com>

vllm/lora/ops/triton_ops/utils.py

github-actions · 2025-07-08T08:35:15Z

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

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 either: Add ready label to the PR or enable auto-merge.

🚀

jeejeelee

Considering that these changes are mostly only related to XPU, we can merge after the CI is all green

Signed-off-by: yan <yan.ma@intel.com>

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: Patrick von Platen <patrick.v.platen@gmail.com>

Signed-off-by: yan <yan.ma@intel.com>

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: avigny <47987522+avigny@users.noreply.github.com>

Signed-off-by: yan <yan.ma@intel.com>

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>

Signed-off-by: yan <yan.ma@intel.com>

support multi-lora on xpu

29dd078

Signed-off-by: yan <yan.ma@intel.com>

yma11 requested a review from jeejeelee as a code owner July 8, 2025 07:54

gemini-code-assist bot reviewed Jul 8, 2025

View reviewed changes

address comments

4ab93dc

Signed-off-by: yan <yan.ma@intel.com>

jeejeelee reviewed Jul 8, 2025

View reviewed changes

vllm/lora/ops/triton_ops/utils.py Show resolved Hide resolved

jeejeelee approved these changes Jul 8, 2025

View reviewed changes

jeejeelee added the ready ONLY add when PR is ready to merge/full CI is needed label Jul 8, 2025

DarkLight1337 merged commit a4c2331 into vllm-project:main Jul 8, 2025
79 checks passed

huydhn pushed a commit to huydhn/vllm that referenced this pull request Jul 8, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

f97c6c3

Signed-off-by: yan <yan.ma@intel.com>

Chen-zexi pushed a commit to Chen-zexi/vllm that referenced this pull request Jul 13, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

5ea7eee

Signed-off-by: yan <yan.ma@intel.com>

patrickvonplaten pushed a commit to patrickvonplaten/vllm that referenced this pull request Jul 15, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

26810e0

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: Patrick von Platen <patrick.v.platen@gmail.com>

LyrisZhong pushed a commit to LyrisZhong/vllm that referenced this pull request Jul 23, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

8058ffb

Signed-off-by: yan <yan.ma@intel.com>

avigny pushed a commit to avigny/vllm that referenced this pull request Jul 31, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

7acc5f0

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: avigny <47987522+avigny@users.noreply.github.com>

Pradyun92 pushed a commit to Pradyun92/vllm that referenced this pull request Aug 6, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

4243dcc

Signed-off-by: yan <yan.ma@intel.com>

npanpaliya pushed a commit to odh-on-pz/vllm-upstream that referenced this pull request Aug 6, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

7df3a2c

Signed-off-by: yan <yan.ma@intel.com>

jinzhen-lin pushed a commit to jinzhen-lin/vllm that referenced this pull request Aug 9, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

cc946d2

Signed-off-by: yan <yan.ma@intel.com> Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>

epwalsh pushed a commit to epwalsh/vllm that referenced this pull request Aug 27, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

f1eb646

Signed-off-by: yan <yan.ma@intel.com>

googlercolin pushed a commit to googlercolin/vllm that referenced this pull request Aug 29, 2025

[xpu]feat: support multi-lora on xpu (vllm-project#20616)

69c5e4f

Signed-off-by: yan <yan.ma@intel.com>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

[xpu]feat: support multi-lora on xpu #20616

[xpu]feat: support multi-lora on xpu #20616

Uh oh!

yma11 commented Jul 8, 2025

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Jul 8, 2025

Uh oh!

gemini-code-assist bot Jul 8, 2025

Uh oh!

gemini-code-assist bot Jul 8, 2025

Uh oh!

Uh oh!

github-actions bot commented Jul 8, 2025

Uh oh!

jeejeelee left a comment

Uh oh!

Uh oh!

Uh oh!

		if envs.VLLM_USE_V1 and vllm_config.model_config is not None and \
		not vllm_config.model_config.enforce_eager:

		vllm_config.compilation_config.level = \
		CompilationLevel.NO_COMPILATION

Uh oh!

[xpu]feat: support multi-lora on xpu #20616

[xpu]feat: support multi-lora on xpu #20616

Uh oh!

Conversation

yma11 commented Jul 8, 2025

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Jul 8, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jul 8, 2025

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jul 8, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot commented Jul 8, 2025

Uh oh!

jeejeelee left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!