SYS-420GP-TNR ; GPU -- PCIE LINK SPEED ISSUE ; PCIe link speeds are reporting as downgraded via lspci: Expected value 16GT/s, but detected 2.5GT/s
I've performed the following troubleshooting steps:
*hard power cycle -- issue persists
*reseat gpus + cables -- issue persists
*check mellanox link speed -- full speed / only gpus affected
*reseat pcie cables -- issue persists
Checked nvidia-smi and found those GPUs are in idle state. Also BIOS/BMC FWs are out of date.
Requested to update BIOS/BMC and to stress GPUs and related concerns resolved