Cytnx
Cytnx copied to clipboard
Add the function for randomized SVD.
Add the function for randomized SVD based on the paper: https://arxiv.org/pdf/2403.00562
FYI I believe the svd we used on CUDA side is via rsvd
FYI I believe the svd we used on CUDA side is via rsvd
How could one make the implementation of #619 faster on the GPU side by using what CUDA provides directly?
Implemented in #619 .