rightnow-ai/gpuci - GPU Kernel Testing
Test CUDA kernels across multiple GPUs via SSH
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Maintainedlast commit Jan 25, 2026
- License
- None
Pinned Snippet
uses: rightnow-ai/gpuci@d6a659f4633e2d526a411eb8ac582ca1642aa01b # v0.1.1tags can be moved; commit SHAs can't. why a SHA?
Inputs
| name | description | required | default |
|---|---|---|---|
| kernel | Path to CUDA kernel file(s). Supports glob patterns. | yes | — |
| config | Path to gpuci.yml config file | no | gpuci.yml |
| target | Filter to specific target(s) by name or GPU type | no | — |
| runs | Number of benchmark iterations | no | 10 |
| warmup | Number of warmup iterations | no | 3 |
| ssh-key | SSH private key for connecting to GPU machines | no | — |
| brev-token | Brev.dev API token (if using Brev provider) | no | — |
| post-comment | Post results as PR comment (true/false) | no | true |
| fail-on-error | Fail the workflow if any GPU test fails | no | true |
Outputs
| name | description |
|---|---|
| results | JSON results from gpuci |
| passed | Whether all tests passed (true/false) |