i suspect the k/native direct to blas will be much...

# datascience

[kyonifer]

i suspect the k/native direct to blas will be much faster, looking at the generated assembly, even with the native team saying they havent optimized for performance yet