Author: Albert Silver
Date: 05:46:47 01/15/06
Go up one level in this thread
On January 15, 2006 at 07:37:55, Roger D Davis wrote:
>On January 14, 2006 at 20:09:05, Albert Silver wrote:
>
>>On January 14, 2006 at 18:00:22, Roger D Davis wrote:
>>
>>>On January 14, 2006 at 17:15:39, Joseph Ciarrochi wrote:
>>>
>>>>Hi folks,
>>>>
>>>>I am just wondering how much improvement vasik has managed from the original
>>>>beta and the newest beta (please post anything you have). My initial tests
>>>>against fritz 9 show (4 min 2 sec increment); noomen openings):
>>>>
>>>>rybka beta 10 37 - 22
>>>>rybka beta 35.5 - 23.5
>>>>
>>>>
>>>>(this tourni still going, will post final results soon.)
>>>>
>>>>My guess is vasik has maybe improved rybka by 30 points (please present any data
>>>>you have since this is truley guessing.)
>>>>
>>>>I am not sure that adding endgame knowledge will lead to big improvement. Maybe
>>>>a half point here or there, but i could be wrong. Maybe 20 point boost?
>>>>
>>>>Let's say the February release of rybka gets a total of 50 points improvement.
>>>>This will give it about a 90 point advantage over fruit, based on the CEGT. If
>>>>fabien keeps improving fruit as he has in the past (see
>>>>http://www.fruitchess.com/playing-strength.htm), then fruit could very well
>>>>catch rybka in the next release.
>>>>
>>>>
>>>>I really like the way Fruit 2.2. plays. It offers you very solid advice and
>>>>seems to understand opening play quite well (will post data on this soon). In
>>>>playing style, it seems very different from rybka and thus forms a nice
>>>>complement to this monster fish.
>>>>
>>>>
>>>>best
>>>>Joseph
>>>>
>>>>ps. did fabien ever offer an explanation about why on earth he called his
>>>>program fruit.
>>>
>>>
>>>Fruit versus Rybka could turn out to be a one of those classic sports rivalries
>>>(yes, chess programming is a sport), because it pits different kinds of
>>>expertise against each other. Rybka's author is obviously a great programmer and
>>>chess IM, but Fruit's author may be the best chess coder alive.
>>
>>I'm absolutely not putting down Fabien's skills by one iota, but it is strange
>>to see someone say that when a superior program, even if possibly temporarily,
>>exists. How do you explain this program's superiority and still claim Fabien is
>>better? What part of Rybka is so much stronger that it overcomes worse coding?
>>
>> Albert
>>
>>>
>
>Obviously, they're both great coders. I don't KNOW that it will develop into a
>great rivalry. Maybe and maybe not.
>
>If I remember correctly, Donninger reviewed one of the Rybka betas in a previous
>thread and stated that the code wasn't particularly well optimized.
I remember the post, but you have forgotten a very important element. He then
suggested that with money Vasik would be able to afford a better compiler. The
problem was with the compiling and not Vasik's coding.
Albert
In contrast,
>everyone has bragged about Fabien's code, including Dann Corbit.
>
>Of course, Vasil isn't done. He's only released betas. So we don't know how far
>he can go in making his code optimal and elegant, and may never know since it's
>not open source. We may never know. Apparently Donninger is knowledgeable enough
>to review the code at the assembly language level and render a judgment, but
>that has to be a rare talent. Whether Donninger or anyone else could "decompile"
>Rybka at the level of chess programming ideas and extract and adapt them for
>their own use, who knows.
>
>I hope it develops into a great rivalry...it's fun to watch. :)
>
>Roger
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.