Some time ago I saw this blog post, https://sarah-ek.veganb.tw/blog/nano-gemm/, which shows a new implementation of matrix multiplication that's quite fast for our use case ('square', size = 50) and in particular faster than ndarray's by maybe 20%. We should investigate how to use the crate and see if that translates into faster evolution methods. The repository of nano-gemm is at https://github.com/sarah-ek/nano-gemm.
Some time ago I saw this blog post, https://sarah-ek.veganb.tw/blog/nano-gemm/, which shows a new implementation of matrix multiplication that's quite fast for our use case ('square', size = 50) and in particular faster than
ndarray's by maybe 20%. We should investigate how to use the crate and see if that translates into faster evolution methods. The repository ofnano-gemmis at https://github.com/sarah-ek/nano-gemm.