Skip to content

[CI] run_llama.sh: Add returncode propagation of llama-bench#2232

Merged
mhalk merged 1 commit into
ROCm:aomp-devfrom
mhalk:mhalk/amd/dev/mhalkenh/feat/ci-llama-returncode-propagation
May 28, 2026
Merged

[CI] run_llama.sh: Add returncode propagation of llama-bench#2232
mhalk merged 1 commit into
ROCm:aomp-devfrom
mhalk:mhalk/amd/dev/mhalkenh/feat/ci-llama-returncode-propagation

Conversation

@mhalk
Copy link
Copy Markdown
Contributor

@mhalk mhalk commented May 27, 2026

Motivation

Even when errors occurred, nothing was reported to the user or calling script.
tee would mask any returncode, e.g. 139 / SEGFAULT.

Technical Details

Added wrapper for the benchmark call which has the required error handling.
That is: Print error message and capture & propagate returncode of llama-bench.

Test Plan

Small amount of manual runs, e.g. with AOMP_GPU set to a wrong GPU arch.

Test Result

Script reports errors and may return with the corresponding returncode of llama-bench.

Submission Checklist

Copy link
Copy Markdown
Contributor

@Kewen12 Kewen12 left a comment

Choose a reason for hiding this comment

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

lgtm

@mhalk mhalk merged commit 8a3058d into ROCm:aomp-dev May 28, 2026
1 check passed
@mhalk mhalk deleted the mhalk/amd/dev/mhalkenh/feat/ci-llama-returncode-propagation branch May 28, 2026 12:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants