Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / rybka very slow compared to fritz in Chessbase analysis
- - By nzmose Date 2007-05-13 00:24
I have the latest rybka, and chessbase.

When I go to use the rybka and fritz10 engines together when playing over a game using chessbase the nodes per second for fritz 10 reads around
326 kNodes/second but rybka is always around 9 kNodes /second.

This is a huge difference.

what is going on?  I have a reasonably fast computer. Why is rybka so much slower in doing the same analysis as fritz 10  ?
Parent - - By kblaine (**) Date 2007-05-13 04:11 Edited 2007-05-13 04:13
Yours is a very common question. Don't worry. Rybka counts nodes per second very differently than other engines. It doesn't mean anything is wrong.
Parent - By Kapaun (****) Date 2007-05-20 22:17
Well, 9 is a very low number, though. Even for Rybka...
Parent - - By FICGS (**) Date 2007-05-22 02:06
Did Vasik comment this issue ?  Actually I would not be very surprised to see a slow engine with more chess knowledge beating Fritz and others... How is it possible to count nodes so differently ?
Parent - By kblaine (**) Date 2007-05-22 04:10 Edited 2007-05-22 04:24
Search the Rybka Support section of this forum for discussions on nodes per second (Kn/s). At least 2 new users ask about it every week. And that is understandable since we are accustomed to seeing much higher rates. It's true 9 is exceptionally low but most likely caused by an unusual setting in the GUI. Try running Rybka alone on infinite analysis and see what you get. It should be a considerably higher rate than 9 Kn/s but significantly less than most other engines.
Parent - - By insipid (**) Date 2007-05-25 03:28
Well, Garry Kasparov sees around 1-5 positions per second, yet this was enough to defeat Deep Blue's 200 million. (at least in one game..)  There's more to chess than positions per second, Rybka's great strength is knowing where to look, not how fast.

If you look at depths Rybka is rarely behind the other engines.

Yes 9kn/s is quite low, your computer is a dog, 300k is also very low for Fritz..
Parent - - By diskamyl (**) Date 2007-05-25 10:38
I don't think kasparov sees 1-5 positions per second.
Parent - - By insipid (**) Date 2007-05-26 06:21
The extremely oft-quoted figure, around the time of the '96 match, was that Kasparov calculates and evaluates three positions per second.  I think this is probably reasonable and correct, and it was never disputed (even agreed with by Garry, IIRC, though a quick google search doesn't turn up any such quote), and I increased the bounds to 1-5 to give some room for opinions.

It certainly is not much higher.
Parent - By diskamyl (**) Date 2007-05-26 11:10 Edited 2007-05-26 11:12
maybe for the first second, let's say he can analyze variations to 5 plies, that makes 5 positions. but after that, for example in the 3rd second, I think it's not possible for a human being to get near 5. so I'm almost sure that 5 is out of question.

what I meant was that kasparov cannot analyze 300 positions when he thinks for 5 minutes on a move. so that doesn't make even 1 position per second. but maybe I'm underestimating grandmasters, I'm just speculating here.
Up Topic The Rybka Lounge / Computer Chess / rybka very slow compared to fritz in Chessbase analysis

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill