Author: Joshua Lee
Date: 10:12:37 10/06/01
Go up one level in this thread
> >I do not know about the HP chip but Wchess used pentium90 when it drew Deep Blue >Prototype. > >Uri This is the same Wchess as on the SSDF list version 1.06 53 WChess2000 64MB P200 MMX 2355 31 -32 522 36% 2457 83 WChess 1.06 Pentium 90 MHz 2231 20 -20 1222 39% 2308 90 WChess 1.03 486/50-66 MHz 2198 27 -27 677 49% 2204 Ok so if anything Deep Thought/Blue/Prototype Only managed a draw against most likely WChess 1.06 which is 2231 on a P90 so for that one game Deep is within 396 elo points Which would make the prototype Around 2627 At best according to statistics I.E if you play 100 games and draw 1 lose the rest. I don't think the list ratings are Accurate, Deep Fritz is rated 2726 but on a 4x500 1550 it is Around 151 points weaker, On a 500Mhz machine Fritz is Between 2355 and 2500 with 2423 being a safe average of two tournaments and On the list 2623 is obviously too high for Fritz 6 on a 450Mhz machine. 14-16cpu Deep Blue Prototype played in Hong Kong in 95 wasn't the actual Deep Blue If all this is 80-100% true Deep Prototype searching 4-5Million positions a second would be (Fritz 2623 - 200 =2423 Wchess 2231 - 200 = 2031 ?? Not sure about that ) This would make Deep Prototype 2427 Which is not logical, it is safe to say some programs are incorrectly rated while others are not From SSDF's own Human Vs Computer Calibration you can Add or subtract 100-200 Points , Then if you include Chris Carson's list you can With some of the results Fritz Deep Fritz, show that without a doubt This list is so bogus to Claim on a 1200Mhz machine a chess program is playing in the same league with Karpov, Kasparov, Botvinnik, Spassky etc Alekhine at his peak Was 2690 , this is 36 points lower than DF If you think the ratings are accurate there has to be something wrong with you're brain in light of the facts. As far as Deep Blue is concerned they added Knowledge and power to this Tripling Their numbers 3 times +270 points this would put Deep Blue at 2897 which makes the First Number Too High by 255 points as 2642 was the performance from the 96 match with Deep Looking really like a (2535 positionally) The numbers don't match up here but suffice to say that It proves SSDF is far off the mark despite what seirawan thinks.
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.