You should benchmark all implementations (using cpu time, not realtime) and choose the fastest and while benchmarking check whether the algorithm actually works.
+1 agreed. It’s not difficult or time-consuming for each user to do this at startup.
Re: Auto-detect for 128-bit 4-way SSE2
Figures: Jeff Garzik
Quote from: tcatm on September 09, 2010, 3:02:27 PM UTC