Hi, does this support GPU versions with more than 7.5 computing capacity, such as NVIDIA RTX 3090 or Tesla A100?

[XiaoqiWang]

Hi, does this support GPU versions with more than 7.5 computing capacity, such as NVIDIA RTX 3090 or Tesla A100?

[beckernick]

In case it's useful, KMeans and KNN are also implemented in cuML. Both algorithms are supported multi-GPU and multi-node. If you'd like to try it, I recommend using the RAPIDS release selector, which provides instructions for conda packages, docker containers, or source builds.