Deleted Lambda Stack as it wasn’t working with Pytorch or Tensorflow.
Reinstalled fresh Ubuntu 22.04, new nvidia drivers etc. Pytorch and Tensorflow worked for ~3 weeks.
Last Thursday training started suddenly going very slowly
Checked nvidia-smi :
NVIDIA-SMI has failed because it couldn’t communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.
Reinstalled 515 drivers through Software & Updates , restarted and got nothing but a flashing cursor on restart.
Have now:
- Uninstalled / reinstalled Ubuntu 22.04, drivers, etc, same issue
- Tried Ubuntu 20.04, same issue
Interestingly the nvidia gpu output of lspci | grep VGA is VGA compatible controller: NVIDIA Corporation Device 2460 (rev a1) when this should be something like NVIDIA Geforce Ti 3080 or similar
If I change Chipset > GPU Mode to Discrete GPU Only I can get to the login screen, but the trackpad and mouse don’t work.
I can’t get the recovery iso image off the website as the link doesn’t work
So what’s the next move here? Calling Lambda tomorrow but wondering if anyone else has experienced similar?