Author: Robert Hyatt
Date: 11:55:04 07/26/00
Go up one level in this thread
On July 26, 2000 at 08:29:51, Chris Carson wrote: >On July 26, 2000 at 01:26:38, Ed Schröder wrote: > >>On July 25, 2000 at 19:41:58, Dave Gomboc wrote: >> >>>On July 25, 2000 at 10:44:20, Chris Carson wrote: >>> >>>>On July 25, 2000 at 10:19:10, Ed Schröder wrote: >>>> >>>>>On July 25, 2000 at 08:44:57, Dave Gomboc wrote: >>>>> >>>>>>- the "1 million nodes/sec" figure is a peak figure, not an average >>>>>> - average is 200k nodes/sec >>>>> >>>>>From the IBM site (may 1997): >>>>> >>>>> "Deep Blue was now capable of examining and >>>>> evaluating an average of 100 >>>>> million chess positions per >>>>> second." >>>>> >>>>>Ed >>>> >>>>Thanks Ed! Accurate and factual as always. :) >>>> >>>>Best Regards, >>>>Chris Carson >>> >>>Boy, did you miss the point. >>> >>>We're talking about one chip, not the whole machine. >>> >>>Dave >> >>I don't thinks Chris or me missed the point. I have quoted from the >>IBM pages (see above). It says DB2 does 100M nodes AVERAGE and that's >>all what counts. >> >>Ed > >You have the right quote from the IBM page. When Bob's argument >goes south, he has to up the NPS to 1Billion (in a year it will be >more based on something only he knows about and was never before >disclosed to anyone else) or up the number of chips or talk about >DT vs 6502/386/486 machines or something. > My argument won't "go south". 480 processors is the right number. 2.2M nps per processor (average) is the right number. 480 * 2.2M is > 1B. Which is a right number. 200M is his 'effective number' which no other parallel program is reporting, they _all_ report the total nodes / total time, to get a raw NPS. Hsu's RAW nps is around 700M, based on his statement in several papers that he can drive the chess chips at about 70% duty cycle using the current SP2 configuration. Or factoring in his 30% efficiency, he claims 200M which is a very realistic (and conservative when compared with every other parallel program's NPS) number. Unlike yourself, I try to not manufacture data, nor twist it to suit my own purposes. I simply report _factual_ numbers for DB, numbers you could easily verify _if_ you really wanted to. But that would take the fun out of the constant arguing, wouldn't it??? > IBM tried to get the >most hype out of the match (nothing wrong with that, they were paying >for it) and put out the best, most flattering picture of DB they could, >they left out the 95 lose at the WCCC to Fritz3 and the fact that they >had not won a WCCC since 1989 (Chessmachine 1992, Fritz 1995, Shredder 1999). > >All the hype anyone needs is on the IBM pages. The easiest way to change >my opinion is to show me some legitimate publication (book, ACM journal, >Scientific America article, ...) with new facts. HSU and Murry have not >published anything that supports many out rageous statements made here. >I believe they would if there was more to tell. Silence says a lot. That _really_ shows what you are made of, academia-wise. Do you read IEEE Micro? Go find Hsu's article there. It details _everything_ about their hardware and speed. Do you read the JICCA? It has plenty of information published by them. AAAI? Same. There are plenty of articles out there written by members of the DB team. Just because _you_ haven't read them, doesn't mean they don't exist. To paraphrase our legal system, "ignorance of readily available technical details does not make a valid excuse for not knowing them." > >Deep Blue was a great achievement. I would still make it a favorite >by 25% today (3 years later). It was very impressive. It is dead. >Next month the Px-1.5GHZ will be out and by next year the 8x-700 will >be 50% or more slower than the fastest multi-processors. Next year, >I would give any ot the top commercials on the fastest multi-processors >equal to 25% chances over DB 97 and in 3 years, I would give any >of the top commercial programs on single processors equal to 25% over >DB 97. This may be to conservative considering diminishing returns for >the multi processors and the amount of improvement and continued >improvement in the comercial programs strength (Rebel, Chess Tiger, >Deep Junior, Fritz, Hiarcs, Shredder in no particular order). DB had it's >day and it is over (DT's performance has been surpassed, DB 95 HW surpassed, >DB 96 performance has been surpassed and DT 97 was only twice as fast as >DB 96). DB is dead, long live the new programs/hw/programmers. :) > >Hey, do not take my word for it, research it yourself, come >to you own conclusions, talk about it here if you like. I can accept >an opinion, I do not accept opinions as fact without documented proof, >but that is just me. :) > >Best Regards, >Chris Carson You don't bother to find documented anything it seems. Just find the IEEE Micro issue with the DB article. It gives numbers for everything you want to dispute. In black and white, even...
This page took 0.01 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.