Author: Walter Faxon
Date: 05:19:35 01/17/06
Go up one level in this thread
On January 17, 2006 at 06:50:44, Eelco de Groot wrote: >On January 17, 2006 at 01:15:40, Sandro Necchi wrote: > >>On January 16, 2006 at 22:34:24, Eelco de Groot wrote: >> >>>Hello Andreas, >>> >>>I had a little question, maybe Sandro knows, we could read that in Paderborn and >>>in Leiden Stefan played with his Mac version, and that Gravis actually sponsored >>>him with the machine: >>> >>>http://www.shredderchess.de/schach-news/shredder-news/ipccc2005b.html >>> >>>(Sorry folks, only in German but the text is easily translatable I think) >>> >>>But did Stefan Meyer-Kahlen not optimize Deep Shredder for 64-bit yet on the G5? >>>Stefan writes he got the machine from Gravis long in advance to optimize his >>>program, so it would seem plausible he already made a 64bit version? Or would >>>that not run on regular Macs, so Mac owners will then probably not get this >>>version? I have no idea about the latest Apple developments, I only know Apple >>>decided to switch to Intel chips instead of IBM, does that apply to the G5 >>>already or is that still IBM architecture? >>> >>>I thought that was the reason Sandro had to ask permission from Stefan, as this >>>probably is also the machine Shredder will be running in Turin in May WCCC, I >>>would think Shredder is already well optimized for this computer. >>> >>> Regards, Eelco >> >>Sorry, but what I sent has nothing to do with Shredder on quad. I have sent a >>book and not for Shredder. >> > >Ah, that makes more sense! I think I can guess for what program Albert wanted a >book from you! > > >>Stefan could not optimize the engine on the quad because he was very busy >>completing the work on the new commercial releases. >> >>The same is true for the engine as in the last 2 years and a half he only worked >>on the engine less than a month as we did not need to increase the strenght much >>because we were on the top. >> >>Now he is working on the engine full time and I think he will until March at >>least. > >Thanks for the info about Shredder Sandro. If Stefan Meyer-Kahlen can now work >fulltime on the engine for a while and last 2½ years could do that only for a >month, we can expect some progress! > >If Fruit makes a 100 elopoints gain by March, I think that would be an awful >lot, considering it is already a strong engine, it would be an enormous leap at >that level. I think 50 would be a more conservative bet, compare grandmasters >for instance, no human grandmaster can get 50 points stronger than their normal >level in a year. My guess is that Rybka cannot make the same kind of progress, >more than 50 points in a few months, simply because now Fabien has had more time >since the last version and Vasik also has to iron out some minor issues and work >on endings, analysis etc. This is only good for Rybka because the elonumbers are >not everything and also the more customers, the more diverse the wishes (and >vice versa I think)! So my own bet is Fruit could be strongest in March even >with a gain of say 50 points, but then again Ryan Benitez was not quite so >optimistic Fruit would come out on top. Or maybe Gambit Fruit, or Shredder of >course, or Glaurung makes another jump in strength.. > > >> >>This is why I am optimistic about the outcome. >> >>Sandro > > Good! > > Eelco There is some well-known result in decision theory/uncertainty analysis (but not so well-known that I can find it!) regarding making estimates that depend on uncertain parameters. If you have only a couple of unknowns your errors will have a large effect on the outcome. But if instead there are a large number of unknowns it turns out that the less you know, the more likely it is that your errors will "even out" to produce a reasonable ballpark estimate. But that's for a single estimate. I think you stretch this method too far by trying to compute the relative rankings for several such estimates. The new competitive landscape implies that someone will probably come out with a substantially improved chess engine within the next few months or a year. Don't take any bets on who will do it! -Walter
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.