Use AVX2 #11
Description
AVX2 is available for ~7ish years by now in processors; we should take it for granted by now and use it.
Use it to speed up mem copies, finds, bitsets operations, etc.
Requires
-march=flag in cflags-mavx2and related flags in cflags- intrinsics header and avx2 implementation
- thinking about runtime dispatch vs ifdef'ing it out at compile-time vs always requiring avx2
Requires benchmarks first #7
Ideas
- linear scan / binary search / exponential search / interpolating search: start here
- all pair shortest path (matrix multiplication): start here