microsoft / knossos-ksc

Compiler with automatic differentiation
Other
45 stars 10 forks source link

Support tensor rank either 1 or 2 in CUDA vrelu3 #815

Closed dcrc2 closed 3 years ago

dcrc2 commented 3 years ago

I've done this in a fairly simple way which results in some duplication. Unless there's an easy fix for this that I've not spotted yet, I'm not sure it's worth trying to improve this while the code is still experimental.