Author: Mike S.
Date: 09:41:43 11/01/01
Go up one level in this thread
On November 01, 2001 at 02:09:59, Jouni Uski wrote: >I quess no MMX is simply "no Man Machine eXtension" :-) :o) But serious: Different speed of different CPU's MMX code is an issue here. The better results of "No MMX" vs. the standard version came from an AMD K6-350 CPU. I thought there would be a few hardware experts here to answer, if MMX code in AMD CPU's is equally fast, or (much) slower than in Intel's. This *could* explain, at least partially, why in these tests the standard engine was much less successful than No MMX. Ernst Walet quoted Frans Morsch in a posting below, having said that he uses MMX registers for bitboard operations. If my assumption is true, we should see results from Intel CPUs which are more narrow than from AMD's CPUs... I don't know though, if there is another difference between K6 and Athlon regarding MMX implementation. Computer chess is difficult! :o) (That's why sometimes, even I prefer to ask the experts, instead of studying and answering myself...) Regards, M.Scheidl
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.