Author: Vincent Diepeveen
Date: 04:39:43 10/01/03
Go up one level in this thread
On September 29, 2003 at 23:44:07, Jay Urbanski wrote: >On September 28, 2003 at 19:58:49, Gian-Carlo Pascutto wrote: > >>On September 28, 2003 at 18:21:39, enrico carrisco wrote: >> >>>I am curious to know if these scores have been accepted by SPEC. >> >>I doubt so, at the very least they are not on the SPEC site itself. >> >>-- >>GCP > >I think the ~1200 specint score they are using for the P4 3 Ghz has only 512K L2 >cache and no L3 cache. The P4 "Extreme" they overclocked to 3.8Ghz has 512K L2 >cache and 2MB L3 cache, so it's possible. the P4 EE doesn't exist currently. it's modded Xeon MPs right now. So the scores of P4 EE are simply not true. If you buy P4 EE in the shops in a while you won't see the same score simply. Note the big L3 cache didn't matter for diep. the SMT speedups reported were too high simply. As if diep would get 25% in SMT. It's like this for years already. The SMT / HT speedups of 'testmachines' from intel are higher than when you buy such a thing in the shop. I remember NALIMOV reporting 30% for crafty. others tried similar cpu's and got to 10-15%. I remember several reports from well known testers indicating 20% at P4 3.06Ghz for diep. When a few months later some production machines diep was tested it was 10%. Other programs have the same experiences (Fritz). Intel test machines are different from what you can buy in the shop. I would really be amazed when you can get a cheap Xeon MP for this price. All you need is a small socket adapter. 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.