Matt Fraser
Matt Fraser
Looking into it.
There were issues with Fabric Manager 550.54.15, so we had to revert FM and the GPU driver. As per NVIDIA, this version of CUDA should be compatible with the GPU...
yes, this continued when we moved to AlmaLinux from CentOS. It bothers me too, so I'll look into it.
When I asked, I was told this was intentional. I'm not sure on the reason.
Will look into for the next release
As a FYI, development using Spack was halted, but I will be working on porting simplified tests back to master
We're currently evaluating which distros to expand support to
> You could use any image with a supported Infiniband instance, and you can build Mellanox from scratch. I would suggest go with the stable 5.8-4.1.5.0-LTS. https://network.nvidia.com/products/infiniband-drivers/linux/mlnx_ofed/ > > If...
This is still an issue for me. The state seems to always be unavailable, but if I look at the history there are periods where speed is recorded, and then...
I also had my 3 WAN sensors disappear, but it looks like @bonroyage has nailed this.