Why you're never going to catch Rybka

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

gladius
Posts: 568
Joined: Tue Dec 12, 2006 10:10 am
Full name: Gary Linscott

Re: Why you're never going to catch Rybka

Post by gladius »

The point is that Rybka 3 scales to multiple CPU's, and Rybka 1 doesn't. If you compare their strength on an 4 core machine, you will get the 300 ELO difference.

1 Rybka 3 x64 4CPU 3206 15 15 1857 80.0% 2965 28.6%
92 Rybka 1.0 Beta x64 2869 11 11 2558 57.4% 2817 39.5%

ELO difference = 337.

This doesn't mean that head to head Rybka 3 will be 300 ELO stronger, but that in the testing conditions that CEGT uses, it is.
User avatar
George Tsavdaris
Posts: 1627
Joined: Thu Mar 09, 2006 12:35 pm

Re: Why you're never going to catch Rybka

Post by George Tsavdaris »

Tony Thomas wrote:
George Tsavdaris wrote:
GenoM wrote:
Ovyron wrote:If Vas was smart enough to improve Rybka 1 by some 300 rating points I think he's smart enough to have created it from scratch.
Have any of users of Rybka 3 made any matches between Rybka 3 & 1 to prove such a bold statement (about 300+ rating points)?
I thought you knew something about computer Chess. I'm disappointed i was wrong.

Read these lists:
CEGT 40/40
[CCRL 40/40



CEGT and CCRL have 1 CPU Rybka 3 to be around 240 to 245 ELO points stronger than Rybka 1.0.
So it's stronger by about 300 ELO points.
240-245 shouldnt be rounded up to 300..unless you are counting oodles.
I didn't round up anything.
I've just said(with indirect way) that the word "about" can be used with a free way.....
Also the 300 to 245 is approximately 22% more, so in common language it's about the same.
After his son's birth they've asked him:
"Is it a boy or girl?"
YES! He replied.....
Tony Thomas

Re: Why you're never going to catch Rybka

Post by Tony Thomas »

George Tsavdaris wrote:
Tony Thomas wrote:
George Tsavdaris wrote:
GenoM wrote:
Ovyron wrote:If Vas was smart enough to improve Rybka 1 by some 300 rating points I think he's smart enough to have created it from scratch.
Have any of users of Rybka 3 made any matches between Rybka 3 & 1 to prove such a bold statement (about 300+ rating points)?
I thought you knew something about computer Chess. I'm disappointed i was wrong.

Read these lists:
CEGT 40/40
[CCRL 40/40



CEGT and CCRL have 1 CPU Rybka 3 to be around 240 to 245 ELO points stronger than Rybka 1.0.
So it's stronger by about 300 ELO points.
240-245 shouldnt be rounded up to 300..unless you are counting oodles.
I didn't round up anything.
I've just said(with indirect way) that the word "about" can be used with a free way.....
Also the 300 to 245 is approximately 22% more, so in common language it's about the same.

OK, so if I give you 245 dollars you will give me 300 dollars few minutes later?? After all, they are practically the same. :lol: