vllm - 💡(How to fix) Fix [CI Failure]: mi355_2: Qwen3-30B-A3B-FP8-block Accuracy (B200-MI355) [1 comments, 2 participants]

TL;DR

Investigate the mi355_2 hardware status and the recent changes in the nightly build to identify the cause of the failing test group.

Guidance

• Review the build logs for amd-ci #7854 to check for any errors or warnings that might indicate the cause of the failure.
• Investigate the hardware status of mi355_2 to determine if there are any issues with the hardware configuration or environment.
• Check the test group Qwen3-30B-A3B-FP8-block Accuracy (B200-MI355) to see if there have been any recent changes to the test cases or data that could be causing the failure.
• Compare the latest nightly build with the last successful nightly build from 2026-04-20 to identify any changes that might have introduced the failure.

Notes

The issue seems to be related to a specific hardware and test group, so the investigation should focus on these areas. The fact that the test group was passing previously and started failing recently suggests that there might be a change in the environment, hardware, or test cases that is causing the failure.

Recommendation

Apply workaround: Investigate and fix the issue with the mi355_2 hardware or the test group, as the problem seems to be specific to this configuration.