Author: Alain Lyrette
Date: 20:17:31 05/14/01
Go up one level in this thread
On May 14, 2001 at 13:16:33, Christophe Theron wrote: >On May 14, 2001 at 09:52:33, Steve Maughan wrote: > >>Christophe, >> >>>Chess Tiger 14.0 (AMD K6-2 450, 6Mb hash) avoids Rxb3 instantly (it plays g6). >>> >>>Chess Tiger 14.1 for Palm (PalmIIIx 26MHz, no WS, 48Kb hash) takes 137s to >>>avoid Rxb3, plays Qxh4, then takes 167s to play g6. >> >>Interesting. May I ask what speed differential you're getting on the Palm >>version? I guess it must be about 1/300 of the speed of the AMD K6-2 i.e. about >>800 nps. Am I close? >> >>Regards, >> >>Steve > > > >Yes you are close. > >I get approx 500 nps on the Palm (less in the opening, more in the endgame). > >On the K6-2 450 I get typically 150000 nps (more or less). So the PalmIIIx >@26MHz (no WS) is indeed 300 times slower... > >After the release of Tiger 12 I have rewritten a number of internal structures >and routines in order to use 16 bits words as often as possible because the >68000 is more comfortable with these. > >It worked, and I even got a speedup on the PC (I think it is because the L1 >cache is less saturated if I use a lot of 16 bits integers instead of 32 bits >ones). > >But I still believe I could optimise further by taking into account some other >architectural differences. > > > > Christophe Mmmm...very interesting.O.k then ...how much nps can we expect from your program on a pocketPC running on a strongarm 206 mhz?
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.