reanimator/bee
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

Stabilize DCGM GPU discovery

Browse Source
This commit is contained in:
Mikhail Chusavitin
2026-04-03 09:50:33 +03:00
parent 7f6386dccc
commit 7a843be6b0
3 changed files with 76 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
audit/internal/platform/sat.go
View File

@@ -286,7 +286,25 @@ func (s *System) RunNvidiaAcceptancePack(baseDir string, logFunc func(string)) (
// gpuIndices: specific GPU indices to test (empty = all GPUs).
// ctx cancellation kills the running job.
func (s *System) RunNvidiaAcceptancePackWithOptions(ctx context.Context, baseDir string, diagLevel int, gpuIndices []int, logFunc func(string)) (string, error) {
return runAcceptancePackCtx(ctx, baseDir, "gpu-nvidia", nvidiaDCGMJobs(diagLevel, gpuIndices), logFunc)
resolvedGPUIndices, err := resolveDCGMGPUIndices(gpuIndices)
if err != nil {
return "", err
}
return runAcceptancePackCtx(ctx, baseDir, "gpu-nvidia", nvidiaDCGMJobs(diagLevel, resolvedGPUIndices), logFunc)
}
func resolveDCGMGPUIndices(gpuIndices []int) ([]int, error) {
if len(gpuIndices) > 0 {
return dedupeSortedIndices(gpuIndices), nil
}
all, err := listNvidiaGPUIndices()
if err != nil {
return nil, err
}
if len(all) == 0 {
return nil, fmt.Errorf("nvidia-smi found no NVIDIA GPUs")
}
return all, nil
}
func (s *System) RunMemoryAcceptancePack(ctx context.Context, baseDir string, logFunc func(string)) (string, error) {