philippwindischhofer
commented
1 year ago Moving to an optimized cubic interpolation kernel resulted in another huge improvement. At this point the runtime is dominated by the time it takes to load the weighting field, so I'm stopping for now and will wait until @cozzyd's storage library becomes available.
No SIMD intrinsics yet, just some minor optimizations in often-called functions.
Current speed improvement w.r.t.
main: 17%