I’m seeing this warning message when the system wakes from sleep but not on a fresh boot.
nvidia-smi
±----------------------------------------------------------------------------+
| NVIDIA-SMI 410.48 Driver Version: 410.48 |
|-------------------------------±---------------------±---------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX TIT… Off | 00000000:65:00.0 On | N/A |
| 22% 59C P2 77W / 250W | 649MiB / 12209MiB | 0% Default |
±------------------------------±---------------------±---------------------+
WARNING: infoROM is corrupted at gpu 0000:65:00.0
I don’t recall seeing this before but I don’t use nvidia-smi a lot and may have missed it.
I’ve also seen intermittent issues with Tensorflow such as “failed initializing StreamExecutor for CUDA device ordinal 0: Internal: failed call to cuDevicePrimaryCtxRetain: CUDA_ERROR_UNKNOWN: unknown error”, and wonder if this could be related (although most of the time TF/cuda does seem to work properly).
Any ideas on what this error message means, given that it’s only on wake from sleep? This looks like a bug to me. Is there a separate place I should report this?