Nvidia-smi only see 3 of 4 A100, the OS see them all 4

Hi,
I have 4 x A100 on a Dell server linked with a Liqid box. I configure the 4 cards on vSphere (v6.7) with a passthru to 1 VM (Ubuntu 20.04). The OS see them all (lspci) but the driver show only 3 of them (nvidia-smi). Any idea why only 3 are showed? I have the nvidia-bug-report.log but it doesn’t seem that I can attach a file to this post (first time here).
Regards

Please make sure you properly set up the vm (EFI boot, enough address space for all Teslas) this also applies to the host bios. See:
https://blogs.vmware.com/apps/2018/10/how-to-enable-nvidia-v100-gpu-in-passthrough-mode-on-vsphere-for-machine-learning-and-other-hpc-workloads.html

1 Like

After several tests, I have a defect card! No configurstion or driver problem. Thanks generix for your answer.