reverse the amd ci test back to 1200s and split the 8-gpu deepseek job into two. by sunxxuns · Pull Request #11238 · sgl-project/sglang

sunxxuns · 2025-10-06T00:57:37Z

Motivation

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

Format your code according to the Format code with pre-commit.
Add unit tests according to the Run and add unit tests.
Update documentation according to Write documentations.
Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.

gemini-code-assist · 2025-10-06T00:57:46Z

Summary of Changes

Hello @sunxxuns, 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 standardizes the default timeout for continuous integration tests across all environments. It specifically reverts a previous adjustment that had increased the timeout for AMD (HIP) systems, ensuring a consistent 1200-second limit for the execution of each test file.

Highlights

CI Test Timeout Reversion: This pull request reverts the default timeout for CI tests, specifically for AMD (HIP) environments, back to 1200 seconds. Previously, the timeout was conditionally set to 1500 seconds for AMD and 1200 seconds for other environments.

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 by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

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 pull request 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 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.

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 reverts the timeout for AMD CI tests from 1500s back to 1200s, aligning it with the default timeout for other environments. The change simplifies the code by removing the conditional logic for the timeout value. My review includes a suggestion to add a comment explaining this change for future reference. It would also be helpful to fill out the pull request description to provide more context for the change.

gemini-code-assist · 2025-10-06T00:58:45Z

test/srt/run_suite.py

@@ -369,7 +369,7 @@ def auto_partition(files, rank, size):
    arg_parser.add_argument(
        "--timeout-per-file",
        type=int,
-        default=1500 if is_hip() else 1200,
+        default=1200,


While simplifying the code is good, the context behind why the AMD-specific timeout is being removed is lost. Adding a brief comment would help future maintainers understand the history of this value, especially if timeout issues arise again on AMD platforms.

Suggested change

default=1200,

# The default timeout is 1200s. A longer timeout was temporarily used for AMD CI.

default=1200,

merrymercy · 2025-10-06T05:36:42Z

test/srt/run_suite.py

+        TestFile("test_deepseek_v3_basic.py", 150),
+    ],
+    "per-commit-8-gpu-amd-mtp": [
+        TestFile("test_deepseek_v3_mtp.py", 150),


We do not need to split them into two suite. Just with two files should be good, because the timeout is per file.

merrymercy · 2025-10-06T05:37:28Z

.github/workflows/pr-test-amd.yml

@@ -290,7 +290,7 @@ jobs:
        run: |
          bash scripts/ci/amd_ci_exec.sh python3 run_suite.py --suite per-commit-2-gpu-amd

-  unit-test-backend-8-gpu-amd:
+  unit-test-backend-8-gpu-amd-basic:


do not need to split into two jobs

merrymercy · 2025-10-06T05:37:47Z

test/srt/test_deepseek_v3_basic.py

@@ -0,0 +1,77 @@
+import unittest


Splitting into two files is good!

…b into two. (sgl-project#11238) Co-authored-by: root <root@smci350-zts-gtu-e17-15.zts-gtu.dcgpu>

sglang-bot added the run-ci label Oct 6, 2025

gemini-code-assist bot reviewed Oct 6, 2025

View reviewed changes

sunxxuns requested review from merrymercy and zhyncs as code owners October 6, 2025 02:22

sunxxuns added run-ci and removed run-ci labels Oct 6, 2025

sunxxuns force-pushed the rever_amd_ci_timeout_change branch 2 times, most recently from ec4d793 to 077ee6e Compare October 6, 2025 04:10

sunxxuns changed the title ~~reverse the amd ci test back to 1200s~~ reverse the amd ci test back to 1200s and split the 8-gpu deepseek job into two. Oct 6, 2025

merrymercy reviewed Oct 6, 2025

View reviewed changes

test/srt/test_deepseek_v3_basic.py

@@ -0,0 +1,77 @@

import unittest

Copy link

Contributor

merrymercy Oct 6, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Splitting into two files is good!

sunxxuns force-pushed the rever_amd_ci_timeout_change branch 3 times, most recently from 2a34ed7 to a34d5f2 Compare October 6, 2025 06:08

Split unit-test-backend-8-gpu-amd into matrix partitions (basic and mtp)

c23ca4c

sunxxuns force-pushed the rever_amd_ci_timeout_change branch from a34d5f2 to c23ca4c Compare October 6, 2025 06:24

merrymercy approved these changes Oct 6, 2025

View reviewed changes

sunxxuns enabled auto-merge (squash) October 6, 2025 15:38

Merge branch 'main' into rever_amd_ci_timeout_change

acc5719

sunxxuns disabled auto-merge October 6, 2025 18:25

Merge branch 'main' into rever_amd_ci_timeout_change

b581a24

sunxxuns merged commit a57f0e3 into sgl-project:main Oct 6, 2025
82 of 88 checks passed

ch-tiger1 pushed a commit to ch-tiger1/sglang that referenced this pull request Oct 9, 2025

reverse the amd ci test back to 1200s and split the 8-gpu deepseek jo…

3133117

…b into two. (sgl-project#11238) Co-authored-by: root <root@smci350-zts-gtu-e17-15.zts-gtu.dcgpu>

lpc0220 pushed a commit to lpc0220/sglang that referenced this pull request Oct 29, 2025

reverse the amd ci test back to 1200s and split the 8-gpu deepseek jo…

6581e35

…b into two. (sgl-project#11238) Co-authored-by: root <root@smci350-zts-gtu-e17-15.zts-gtu.dcgpu>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

reverse the amd ci test back to 1200s and split the 8-gpu deepseek job into two.#11238

reverse the amd ci test back to 1200s and split the 8-gpu deepseek job into two.#11238
sunxxuns merged 3 commits intosgl-project:mainfrom
sunxxuns:rever_amd_ci_timeout_change

sunxxuns commented Oct 6, 2025

Uh oh!

gemini-code-assist bot commented Oct 6, 2025

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Oct 6, 2025

Uh oh!

merrymercy Oct 6, 2025

Uh oh!

merrymercy Oct 6, 2025

Uh oh!

merrymercy Oct 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Comments

	default=1200,
	# The default timeout is 1200s. A longer timeout was temporarily used for AMD CI.
	default=1200,

Conversation

sunxxuns commented Oct 6, 2025

Motivation

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

Uh oh!

gemini-code-assist bot commented Oct 6, 2025

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 Oct 6, 2025

Choose a reason for hiding this comment

Uh oh!

merrymercy Oct 6, 2025

Choose a reason for hiding this comment

Uh oh!

merrymercy Oct 6, 2025

Choose a reason for hiding this comment

Uh oh!

merrymercy Oct 6, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Comments