We are deploying a commercial ML platform. The vendor of the platform suggest that we install a specific version (470.57.02) of driver on our T4 and V100 Nodes (CentOS 7.9). They came up with a link
We doubled checked the supported product tab in the link, there’s NO product in Tesla/Volta architecture in the list, only RTX series. We strongly doubt that this is not the correct driver for our case. However, in the release note of 470.57.02, Version 470.57.02(Linux)/471.41(Windows) :: NVIDIA Data Center GPU Driver Documentation
we’ve found T4/V100 are within the range of supported product list, but not RTX series.
We wonder if the link given by the vendor is the correct one or not. If not, where should I get the correct driver for our nodes, as we did NOT found driver version 470.57.02 in the official driver search page(Official Advanced Driver Search | NVIDIA), and there’s no 470.57.02 version of driver with the following query configuration:
Product: T4/V100
Operating System: Linux64 Bit
CUDA Toolkit: Any
Recommended/Beta: All