TimBlechmann at gmx.net
Fri Apr 22 10:53:15 CEST 2005
> neither is sickle version of matrix~, nor any other sickle clone.
> When I was experimenting back then with gcc-2.95, the consistent
> pattern was that -funroll-loops performed slightly better than
> unrolling by hand.
well, that's interesting ...
there are three different factors for speeding up things:
- using SIMD instructions (single instruction multiple data)
- using aligned memory operations movaps / movups
- loop unrolling ... loop unrolling is necessary for simd instructions,
since the parallel instructions can be used (although i've never
seen a compiler producing the specific code)
cheers ... tim
mailto:TimBlechmann at gmx.de ICQ: 96771783
latest mp3: kMW.mp3
latest cd: Goh Lee Kwang & Tim Blechmann: Drone
After one look at this planet any visitor from outer space
would say "I want to see the manager."
William S. Burroughs
More information about the Pd-list