Need a wrapper script for gpu-burn
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| checkbox-provider-gpgpu |
Expired
|
Medium
|
Unassigned | ||
Bug Description
Just realized that gpu-burn exits with 0 even if no CUDA devices are found:
terminate called after throwing an instance of 'std::_
No CUDA devices
GPU 0: A100-SXM4-40GB (UUID: GPU-c5185459-
GPU 1: A100-SXM4-40GB (UUID: GPU-02e538b2-
GPU 2: A100-SXM4-40GB (UUID: GPU-b0dc9492-
GPU 3: A100-SXM4-40GB (UUID: GPU-0fe6f3a7-
GPU 4: A100-SXM4-40GB (UUID: GPU-f9d7cc76-
GPU 5: A100-SXM4-40GB (UUID: GPU-a73545d2-
GPU 6: A100-SXM4-40GB (UUID: GPU-a1c775dc-
GPU 7: A100-SXM4-40GB (UUID: GPU-b6d1535c-
We need a wrapper to cause this to generate a non-zero exit code instead.
| tags: | added: hwcert-server |
| Changed in checkbox-provider-gpgpu: | |
| importance: | Undecided → Medium |
| Maksim Beliaev (beliaev-maksim) wrote | #1 |
| Changed in checkbox-provider-gpgpu: | |
| status: | New → Expired |
Bug was migrated to GitHub: https:/
Bug is no more monitored here.