Skip to content

ci: run CodeQL on an 8-core runner - #1377

Draft
roivanov wants to merge 3 commits into
mainfrom
ci/codeql-cpu8
Draft

roivanov wants to merge 3 commits into
mainfrom
ci/codeql-cpu8

Conversation

@roivanov

@roivanov roivanov commented Oct 2, 2026 •

Copy link
Copy Markdown
Collaborator

Summary

Rust CodeQL on ubuntu-latest took 42m 51s on a0cc6cc (4 vCPU / 16 GB). This moves the job to linux-amd64-cpu8 (m7i.2xlarge, 8 vCPU / 32 GiB), the smallest nv-cpu-general size above that.

Changes

  • Set runs-on in .github/workflows/codeql.yml to linux-amd64-cpu8.
  • Also push on ci/codeql-cpu8 so this draft can run. Self-hosted runners reject pull_request. Drop that branch from the trigger before merge.

Runner cost

NVIDIA/cuda-rust is public. Standard ubuntu-latest minutes are free. GitHub larger runners are billed on public repositories too. nv-cpu-general is self-hosted EC2: GitHub does not meter it (the proposed $0.002/min self-hosted platform charge was postponed and is not in the current price list). The bill is the instance.

Rates are list price. GitHub rounds each job up to the next minute (runner pricing). EC2 is us-east-1 Linux on-demand; the runner group's region and any enterprise discount are not in the runner docs. The 150 GiB volume on these labels is extra and small next to the instance (gp3 is about $0.08/GB-month).

The Rust job below is priced at the measured 43 billed minutes. A shorter run on a larger machine costs less than the figure in its row.

Runner Machine $/min 43 min
ubuntu-latest (this repo, public) 4 vCPU / 16 GB $0 $0
GitHub larger Linux 4-core 4 vCPU / 16 GB $0.012 $0.52
GitHub larger Linux 8-core 8 vCPU / 32 GB $0.022 $0.95
GitHub larger Linux 16-core 16 vCPU / 64 GB $0.042 $1.81
GitHub larger Linux 32-core 32 vCPU / 128 GB $0.082 $3.53
linux-amd64-cpu4 m7i.xlarge, 4 vCPU / 16 GiB $0.00336 $0.14
linux-amd64-cpu8 (this PR) m7i.2xlarge, 8 vCPU / 32 GiB $0.00672 $0.29
linux-amd64-cpu16 m7i.4xlarge, 16 vCPU / 64 GiB $0.01344 $0.58
linux-amd64-cpu32 m7i.8xlarge, 32 vCPU / 128 GiB $0.02688 $1.16

At the same 43 minutes, m7i.2xlarge is $0.29 of EC2 against $0.95 for GitHub's 8-core larger runner. Against today's public ubuntu-latest job the GitHub invoice stays $0, and this runner adds about $0.29 of EC2 per Rust analysis if the duration does not drop.

Testing

Workflow-only change, so the crate checks do not apply. The pull_request run fails in runner setup; the push run is the measurement.

Checklist

  • All commits signed off (git commit -s)
  • SPDX headers on new source files

Questions about the review? Ping us in #contributors on Discord.

Rust analysis took about 43 minutes on ubuntu-latest. linux-amd64-cpu8 is the smallest nv-cpu-general size above that, and this branch is included in the push trigger so that runner can start the job.

Signed-off-by: Roman Ivanov <rivanov@nvidia.com>
@NVIDIA NVIDIA deleted a comment from copy-pr-bot Bot Oct 2, 2026
Self-hosted runners reject pull_request events, same as cutile-rs.

Signed-off-by: Roman Ivanov <rivanov@nvidia.com>
@copy-pr-bot

copy-pr-bot Bot commented Oct 2, 2026

Copy link
Copy Markdown

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

The nv-cpu-general image has no rustup, so the extractor exits 0 with an empty database.

Signed-off-by: Roman Ivanov <rivanov@nvidia.com>

This branch has not been deployed

No deployments
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.

1 participant