Author: Robert Hyatt
Date: 18:24:52 12/08/02
Go up one level in this thread
On December 08, 2002 at 16:42:37, Matt Taylor wrote: >On December 08, 2002 at 14:07:50, Christopher A. Morgan wrote: > >>Aaron, >> >>If I'm reading yours and Steffan's test results correctly, >>Athlon XP 2520 is 1.578 million nodes per second, and P4 3.03 >>with HT enabled is 1.161 million nodes per second. Your test >>on Crafty 18.15 and Steffan’s test on Crafty 18.11. >> >>Although the different versions of Crafty could account for some >>difference in speed (which way, I have no idea), the XP 2520 is some >>36% faster than the 3.03 with HT. Bob’s test indicated with two >>CPUs that with HT enabled the HT versions were some 33% faster >>than without HT enabled. He was comparing P4s at ~ 2850 GHz or so. >> >>I see your other results that show that the XP 2520 is substantially >>faster than the P4 at 3.25 GHz. >> >>This means that the XP 2520 is still the speed demon by a wide margin >>compared to the P4, with or without HT enabled. Is this correct? Thanks! >> >> >> >> >> >> >>On December 08, 2002 at 06:05:48, Aaron Gordon wrote: >> >>>I only have results from version 18.11 but here are a few... >>> >>>AthlonXP 2520MHz >>>Total nodes: 104161030 >>>Raw nodes per second: 1578197 >>>Total elapsed time: 66 >>>SMP time-to-ply measurement: 9.696970 >>> >>>Pentium 4 3255MHz without HT >>>Total nodes: 104161030 >>>Raw nodes per second: 1335397 >>>Total elapsed time: 78 >>>SMP time-to-ply measurement: 8.205128 >>> >>>AthlonXP 1860MHz >>>Total nodes: 104161030 >>>Raw nodes per second: 1197253 >>>Total elapsed time: 87 >>>SMP time-to-ply measurement: 7.356322 >>> >>>Pentium 4 2765MHz >>>Total nodes: 104161030 >>>Raw nodes per second: 1120011 >>>Total elapsed time: 93 >>>SMP time-to-ply measurement: 6.881720 >>> >>>http://speedycpu.dyndns.org/crafty/craftybench4.jpg - Crafty v18.11 benchmark >>>list. > >AthlonXP 2.52 GHz, not performance rating 2520. The AthlonXP 2800 is only >clocked at 2.25 GHz. The 1860 MHz AthlonXP is approximately AthlonXP 2200. Bus >speed is important to keep in mind, too, and none of these chips are running at >a standard (100/133/166 MHz) bus speed. > >It would be more valuable to compare the P4 3 GHz with and without HT to the >AthlonXP 2800 on their default bus speeds. I believe this is what Dr. Hyatt was >advocating, too. Nobody has done that, though. > >-Matt I also think that the numbers for the HT-enabled stuff might be wrong. The node counts should vary, because to use HT, it is necessary to run the SMP version of crafty and use mt=2 to use both logical processors. That will produce variability in the total nodes searched that I didn't see in any of the numbers displayed...
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.