Author: Vincent Diepeveen
Date: 13:39:28 10/02/02
Go up one level in this thread
On October 02, 2002 at 08:21:06, Joachim Rang wrote: >On October 02, 2002 at 07:41:19, Jorge Pichard wrote: > >>On October 02, 2002 at 07:12:38, George Sobala wrote: >> >>>On October 02, 2002 at 05:00:41, T. Opas wrote: >>> >>>>It is Deep Fritz 7, but the version is unclear, because there is an agreement, >>>>that Kramnik got the same version as used in the match three month before the >>>>match, means latest on the 4th of july. But the official release of DF7 for >>>>public was about the 15th of august! >>>>Hardware should be 8 x 900 MHz Compaq wizh 4 GB RAM. >>>> >>>>Best, >>>> Torsten >>> >>>So - 8x900MHz? Given the problems of non-linearity between number of processors >>>and "real" speedup, wouldn't a spanking new dual 2.4 or 2.8GHz P4 be as fast (or >>>even faster)? >> >> >>I agree that if they can get theit hands on a dual 2.8 Ghz it would be a little >>bit faster than 8x 900Mhz. >> >>Pichard. > >On this machine DF7 makes 2950Knodes/s. Name me a Dual-Board with this speed. >Don't forget the speed for chessprograms of the P4 is approximately 30% slower >than PIII or AMD. But you can only run it 4 cpu's at that 8 processor machine. So it's more like 4 x 450k nps = 1800k nps. minus some loss = 1600k nps. note that non-commercial deep fritz is faster than commercial deepfritz *perhaps*. Just guessing. It seems at dual K7 at least. Perhaps it is my imagination that Frans got more nps at world champs than deepfritz (6) got at my dual k7 :) I heard some rumours about cache line length and writing within the same cache line (64 bytes at DDR ram and 128 at dual P4 and 32 at old dual P3). Anyway. 8 processor Xeon 933Mhz ==> 1.6MLN nodes a second *at most*. 2 processor K7 ==> 2.2 MLN a second Best regards, Vincent
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.