H100 support?

Open sjtuzq opened this issue 1 year ago • 1 comments

Hi, thanks for your great work! This is super useful, however, one minor issue is, it seems that this framework can only support A100 node, and get stuck on H100 node, I wonder whether the H100 support feature is undergoing or not?

Mar 12 '24 16:03 sjtuzq

Hi @sjtuzq We are running well on a cluster NV H GPUs. Could you please provide more details about your issue? This helps us investigate and reproduce. Thanks.

Mar 14 '24 10:03 binmakeswell

We run our experiments on NV H GPUs. Our code supports H100 training.

Mar 18 '24 06:03 zhengzangw

Does it the cuda version need to be 12.4? Cuda 12.1 cannot works with a problem of core dump.

Jun 03 '24 03:06 zoelearner