My code that I am deploying to Kubernetes can currently only use 1 GPU. My hardware has 2. I’d like to be able to deploy and manage CUDA_VISIBLE_DEVICES from a kubernetes deployment. Does the Nvidia K8s operator do this?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| NVIDIA_VISIBLE_DEVICES vs CUDA_VISIBLE_DEVICES | 2 | 5958 | April 23, 2021 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 508 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 336 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 301 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 306 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 337 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 586 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 330 | April 17, 2019 | |
| CUDA_VISIBLE_DEVICES doesn't work properly | 0 | 315 | April 17, 2019 | |
| CUDA Pro Tip: Control GPU Visibility with CUDA_VISIBLE_DEVICES | 3 | 1340 | April 26, 2021 |