Interesting.
How fast can Futhark be compared to a standard CUDA loop with a few arithmetic, load and save operations? Basically, suppose you're doing simple gathering and scattering?
How fast can Futhark be compared to a standard CUDA loop with a few arithmetic, load and save operations? Basically, suppose you're doing simple gathering and scattering?
@arthas, the author, at some point made comparisons against implementations and it was a most twice as slow, but often faster.