17:19:14 <sech1> tevador you might be interested: https://github.com/xmrig/xmrig/pull/1986
19:09:43 <tevador> sech1: is it basically due to faster mov? dataset init only happens twice per week anyways
19:10:31 <sech1> It's faster because it initializes 5 items at a time (1 item using integer registers, 4 items using AVX 256-bit registers)
19:10:42 <sech1> so it also has more time to prefetch data from cache
19:10:50 <sech1> 5 parallel fetches per thread
19:11:24 <sech1> but it's not faster on all CPUs...
19:15:12 <sech1> It would've been much much faster if there were 64-bit multiplication instructions for AVX...
19:15:35 <sech1> Basically main loop is ~4 times slower than the regular loop, but it initializes 5 elements
20:26:05 <hyc> older CPUs with only 128bit pathways for AVX
22:37:16 <tevador> I guess I could try to test AVX2 with HashX, it could be a bit faster there since it doesn't use mulh