Author: Robert Hyatt
Date: 10:21:53 07/06/03
Go up one level in this thread
On July 05, 2003 at 06:00:05, Aaron Gordon wrote: >On July 04, 2003 at 23:06:25, Robert Hyatt wrote: > >>On July 03, 2003 at 19:58:22, Tom Kerrigan wrote: >> >>>On July 03, 2003 at 19:32:37, Robert Hyatt wrote: >>> >>>>On July 03, 2003 at 18:55:19, Brian Richardson wrote: >>>> >>>>>On July 03, 2003 at 18:26:33, Slater Wold wrote: >>>>> >>>>>>What are your results with your current exe, on the Crafty 'bench' test? >>>>> >>>>> Elapsed SMP Raw >>>>> Version CPUs time time-to-ply NPS >>>>>Opteron 244 (1.7GHz) 19.03 2 28 22.857143 1976833 >>>>>Opteron 244 (1.7GHz) 19.03 1 43 14.883721 1234812 >>>> >>>>That leaves me pretty cold. Dual 2.8ghz xeon, SMT on: >>>> >>>>Total nodes: 122846715 >>>>Raw nodes per second: 2082147 >>>>Total elapsed time: 59 >>>> >>>>This is a different version (19.4) so the time is not really comparable, >>>>but the NPS is, and the xeon is faster than the opteron. That looks ugly, >>>>although it obviously means the opteron is running 32 bit mode only, and it >>>>isn't very happy doing it. Be nice to get real 64 bit boolean operations >>>>up and going. >>> >>>"Isn't very happy"? What's wrong with it? >> >>Simple. It is slower than a dual 2.8xeon box, yet it is AMD's latest. >> >>That's all I said. It is exactly what I meant. > >If you tried optimizing for the Opteron and used a decent compiler with decent >options it'd do a lot better. The P4 isn't worth a crap without specific P4 >optimizations and SSE2 support. That is a fact. You're comparing against a 32bit >non-PGO binary for the opteron with absolutely *NO* optimizations. No way you >can say, "My Xeons are faster". Certainly I can say it, exactly as I did. Based on the reported numbers, my NPS is higher. No question about it. And there is little doubt that if, as I said, a compiler would produce real 64 bit Opteron instructions, that the thing would run faster. But for 32 bit, it definitely did not. And SSE2 is irrelevant here as there is none used in my xeon executable. > >>> What kind of #s does a dual Athlon >>>1.8GHz get? If the Opteron improves over the Athlon with 32 bit code (and I >>>suspect it does) it seems like there should be very little cause for complaint. >>>Add to that the fact that Opteron will almost ceratinly run 64 bit Crafty much >>>faster and we're talking about a 1st generation Opteron anyway. Rumor is, 2GHz >>>Opterons will be available on the 6th, and those should be 5% faster than your >>>dual 2.8 Xeon. >> >>Maybe. My dual 2.8 is already out of date, of course. dual 3.06's are >>shipping with a 33% faster FSB. But I commented on the potential for a >>real 64 bit compiler. But the 32 bit numbers don't excite me. >> >> >> >> >>> >>>-Tom
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.