Dann Corbit wrote:If you analyze opening positions, Rybka is the only engine that is capable of sensible answers.
I disagree. In fact, I think that Rybka's opening is one of her biggest weaknesses, and for analyzing the opening and late opening other engines are better. Once Rybka is on the middle game she's so good that the opening loses relevance, and it's important to know when it's time to switch it on.
For example, in a simple opening position like this:
[d]rnbqkbnr/ppp1pppp/8/3p4/3P4/8/PPP1PPPP/RNBQKBNR w KQkq - 0 1
How much time does your Rybka take to avoid Nc3!? (That blocks c4!!) In positions in where we don't know what's the best move Rybka might be suggesting something bad, just because it was designed with an opening book in mind, that's why I think an engine like Pro Deo is better for opening analysis even though Rybka (and most other strong engines) will destroy it on the middle game.
For the above position, Rybka switches to g3 at 19 seconds (4 x 3GHz 64 bit cores on 64 bit OS). I never analyze for less than an *hour* on an important opening position. I might give one of casual interest a ten minute light dusting.
Try *your* favorite openings analysis engine on any gambit of your choice and see how it does. Then try Rybka. I have never seen any engine besides Rybka get them right.
Rybka is the only engine I can think of that can give you an accurate idea if white is better or worse. Ofcourse it is not perfect ... but it will be right more often than wrong more than any program ... or dare I say any GM. Many top GM's frequently quote "here Rybka thinks white (or black) is better". This list includes Anand, Magnus, Shirov etc... so for sure they value Rybka's opinion. If you run an engine like Fritz or Shredder ... you will get huge swings in evaluation in certain positions .... it is like the engine cannot really decide if this is good or bad ... but Rybka the evaluation is usually very steady and accurate.
Great strides seem to have been made with Rybka 3.0 ... and I would guess even more will come after the release of 3.0. When you have dedicated and highly skilled chess player and motivated individuals (Vas and Larry) putting 100% of their time and effort on a project ... the engine can only get stronger and stronger. I see Rybka 3.0 as a massive strength increase (for me even 50 ELO stronger than 2.3.2a is massive) ... but already I am more intrigued with what Rybka 4.0 will bring.
Dann Corbit wrote:I never analyze for less than an *hour* on an important opening position.
In my case I always prefer to analyze interactively when I have the time, and at least in my 32bit 1CPU 2.1Ghz computer, it is more effective to interactively analyze a position for one hour than to just let a program run infinite analysis for one hour. And when analyzing with two engines, I've been able to refute Rybka's moves in the opening with other engines successfully to the point in where not using Rybka in the opening has been faster than using it for finding the best moves.
However, your 4cores Rybka is much, much faster than mine, so my comments may not apply and perhaps Rybka's analysis quality on the opening is hardware dependent (if on fast hardware Rybka's opening is as strong as the middle game.)
If you are using Rybka to analyse 1. d4 d5 then you are misusing the engine. What are opening books for? I have never heard of anything so ridiculous. Even GMs when preparing for important matches are not so foolhardy, so I do not see why an amateur should do so, donate the CPU time to cancer research or something more productive. Get an Opening Encyclopaedia too if you want to learn the opening (your bookstore has several outstanding works by top GMs).
For a long time Fritz and ChessBase has dominated even though ChessMaster continues to outsell because it is a mass market product even found in your supermarket. However with new GUI Rybka can make a serious dent in the ChessBase GUI market. To many, buying a chess program is not a uci engine but the works (GUI plus engine). Then we will start seeing in the chess magazine, analysis from Rybka suggests .... rather than analysis from Fritz as is the case.
tjfroh wrote:With a 70% increase in search speed with the Vista 64bit version and even more speed with multiple core usage, do you think that Rybka 3.0 should be priced higher or should we keep it low and sell 10X as many programs?
TJ
Why do you think, you'd sell 10x as many for a difference like this?
tjfroh wrote:With a 70% increase in search speed with the Vista 64bit version and even more speed with multiple core usage, do you think that Rybka 3.0 should be priced higher or should we keep it low and sell 10X as many programs?
TJ
Why do you think, you'd sell 10x as many for a difference like this?
...Peter.
A New Way Comes Upon Earth.
God is an infinitely variable Constant.
Man marks his ground with ideologies.
Galaxies are the dreidels of God.
War is a punishment for implacability.
Peace flows from forgiveness of sins.
Dann Corbit wrote:$100 for the bare engine would be reasonable.
$100 for the aquarium gui would be reasonable.
Should I crosspost to the Rybka forum?
NO NO! PLEASE don't!
In fact, the possible good that could come from broadcasting this, is if Vas would feel it worthwhile to apply himself much more, being confident it would be appreciated and pay off.
A small resin dental filling costs about $110.00 here in the USA. Electricians charge over $100.00 per hour for their work. I don't think that Vas Rajlich is overpaid for his labors.
We once paid $150.00 for chess programs and upwards of $10,000.00 for stand-alone chess playing machines. Cars are becoming pricier.....why not chess programs that are so much better than yesterday's programs?
TJ
peter wrote:
tjfroh wrote:Dr. Martan,
OK....$80.00 it is.
Uncle Buck
Ok. Uncle Timothy, if I'd buy, I won't tell (;-)) but you have just lost 90% of your customers
P.S. You may call me Peter.
...Peter.
A New Way Comes Upon Earth.
God is an infinitely variable Constant.
Man marks his ground with ideologies.
Galaxies are the dreidels of God.
War is a punishment for implacability.
Peace flows from forgiveness of sins.
terminator wrote:If you are using Rybka to analyse 1. d4 d5 then you are misusing the engine.
I was giving an example of a position that Rybka mis-evaluates because it's an opening position, and I have seen many others "out of book" but 1. d4 d5 was one that came to mind. Anyway, in Dann's computer Rybka does so for only 19 seconds, when in my computer it takes 25 minutes for Rybka to avoid Nc3.
The point is in positions in where a novelty has been played and so doesn't appear on any opening encyclopedia. In there you want to know the best move, but if the position is like my example Rybka will be stuck with a low quality move like this for a long while, so finding the best move is done faster with other engines that are better in the opening (Unless you have fast hardware.)