Author: Ulrich Tuerke
Date: 07:40:08 02/19/99
Go up one level in this thread
On February 19, 1999 at 10:23:15, Harald Faber wrote: >On February 19, 1999 at 09:52:29, Ulrich Tuerke wrote: > >>>And I won't conclude ANYTHING out of this 11-7 because I also saw Hiarcs7-WChess >>>12-8 and with switched colours there was WChess-Hiarcs 12-8. >>>Anyway I admit I expected some better results for Shredder3 although these are >>>only 18 games. I thought 50% speed advantage would have a bigger effect. >> >>Some "fishy" estimate of the effect of a 50% speed advantage: >>1. we know that 100%advantage gives an ELO plus of about 40 - 50 points; > >This seems to be a bit less than I would estimate taking the average of all >increases from P90->P200MMX, or do you think it is more than 100%? >In general the increase is between 50 and 80 pts from P90->P200MMX I think, it is more than a factor of 2. It also depends on the program. As far as I know (not too far), larger engines profit even more from the MMX (probably a cache effect). > >>2. extrapolating to 50% we expect a plus of 20 - 25 ELO points > >Or 25-40... > >>3. a player with an advantage of 20-25 ELO may expect a win of 53% - 54% > >So in this case, do we assume Shredder3 being equal to Hiarcs7? If so Shredder3, >running on the faster hardware, should get the 53-54%, at the moment he even has >no 40%. Or does it count into the margin of error? > >>So, you may (almost) ignore Shredder's speed advantage. > >Oops. Ignoring means that Shredder3 should/would perform on a MMX166 like on >that K6-200?? >I don't think so. Don't miss the "almost" in my statement above. Sure if I had the choice; I'd take the K6-200. As I said, the estimate is fishy anyway and shouldn't be taken too seriously. However it should give the right order of magnitude. Uli
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.