Comment by crest
9 days ago
I wouldn't be surprised for SSE4 to be the fastest cause it's easiest to map to NEON as both use 128 bit registers and offer a fairly simlar feature set.
9 days ago
I wouldn't be surprised for SSE4 to be the fastest cause it's easiest to map to NEON as both use 128 bit registers and offer a fairly simlar feature set.
Author here - agreed, we have been speculating that too.