It is hard to get access to GB200 instances, and when i do get access i want to be able to run my program quickly and not waste the time on setup.
So i want to simulate IMEX and DRA on 2 DGX spark systems (possibly more) and make sure my program works on a GB200 nvl72 system when i get access.
Are there any gotchas?
I’m not really familiar with these workloads but the main thing I would say is just to make sure that you have enough memory to run them on 2 Sparks.
so installing imex , simulating the nvl 72 domain is all good?
This is currently an untested workload so I cannot guarantee it will work with no issue on the Spark. But if it works on a GB200 and CUDA 13 and the 2 Spark setup has enough memory for your requirements, then it might work.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.