Author: Ed Schröder
Date: 04:22:35 10/19/97
Go up one level in this thread
>Posted by Milton Zucker on October 19, 1997 at 03:03:47: >Can someone say what the expected difference in playing strength would >be with Rebel 9 on a Pentium 200MMX as compared to a Pentium 200 >(non-MMX) at blitz and tournament speeds? Also, is there a rule about >the effect of MMX technology on the strength of chess programs in >general? I thought the main advantage of MMX technology was supposed to >be the increased speed of programs that were graphics intensive. The difference between a P200 and a P200MMX is 25% for Rebel9 see: http://www.xs4all.nl/~rebchess/bench.htm The 25% is *not* related to MMX but to the doubling of the level-1 cache of the P200MMX. However it's quite well possible to speed up chess programs using the new MMX instructions. Somebody from Intel told me 20-25%. Since I have to support all kind of Pc's (386/486/Pentium whatever) it's simply not worth to invest time in the MMX instruction set. Same story with the Pentium PRO, this chip has some nice ARM based new instructions to avoid expensive branches. Based on my ARM-2 (chessmachine) experiences this would allow at least another 10-15% speed gain. Also not worth to invest time in it because of compatibility reasons :(( - Ed Schroder -
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.