Just slap that green flag to see my fastest and simplest perlin noise implementation so far! If anyone would like to compete with me in a codegolf / perf test, I would be willing ;)
Improvements from V2: - no weird broadcast system - optimized the gradient index calculations - only one custom block used to sample - only 2 smootherstep calculations per sample (as opposed to three) I further optimized in my V3.1 project in competition with @xXName77Xx You could make this faster by calculating the indexes and sampling the whole chunk instead of recalculating it with each sample. I actually made this a month ago, but didn't share it for some reason.
