Tony Thomas wrote:I actually wouldnt mind a 10 game match of Hydra vs Rybka 3..I would like to know about the progress they have made with Hydra that only one person in the world can afford.
Well this is what we do know about Hydra as it is posted on their website:
"It runs on a 64CL and evaluates about 200,000,000 chess positions per second, roughly the same as the much older Deep Blue, but with several times more overall computing power. Whilst FPGAs generally have a lower performance level than ASIC chips, Moore’s Law allows modern-day FPGAs to run about as fast as the older ASICs used for Deep Blue. The engine is on average able to evaluate up to a depth of about 18 ply (9 moves by each player), deeper than Deep Blue, which only evaluated to about 12 ply on average. HYDRA’s search uses alpha-beta pruning as well as null-move heuristics. The extra search depth over Deep Blue is due to its use of more modern type B forward pruning techniques that are slightly less perfect but generally play better due to the greater search depth these techniques permit."
Based on this I would say Rybka III would win if Hydra were not improved. Rybka's built in knowledge would destroy Hydra on quad.
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
Tony Thomas wrote:I actually wouldnt mind a 10 game match of Hydra vs Rybka 3..I would like to know about the progress they have made with Hydra that only one person in the world can afford.
Well this is what we do know about Hydra as it is posted on their website:
"It runs on a 64CL and evaluates about 200,000,000 chess positions per second, roughly the same as the much older Deep Blue, but with several times more overall computing power. Whilst FPGAs generally have a lower performance level than ASIC chips, Moore’s Law allows modern-day FPGAs to run about as fast as the older ASICs used for Deep Blue. The engine is on average able to evaluate up to a depth of about 18 ply (9 moves by each player), deeper than Deep Blue, which only evaluated to about 12 ply on average. HYDRA’s search uses alpha-beta pruning as well as null-move heuristics. The extra search depth over Deep Blue is due to its use of more modern type B forward pruning techniques that are slightly less perfect but generally play better due to the greater search depth these techniques permit."
Romichess on my measly celeron searched 18 ply (I am kidding, it was depth 18 ) in a game vs ZCT in the last WCRCC. I am pretty sure most of the moves Romi found were wrong, otherwise Romi would have won the game. At one time I heard that Chrilly was going to use much faster FPGAs, but I never heard anything else about it after that.. I did read about them coming out with a robotic man who plays chess using the Hydra, but that doesnt necessarily tell us weather or not the chess playing part of it was improved. Since Chrilly was the first to reverse engineer Rybka 1.0, I dont see why he cant do the same with Rybka 3.0, and take some of the ideas and apply it in his venture..
I voted no, not because I would have no interest in such a match, but because I think it would be far more interesting to see a match between a 2800 level super-GM and Rybka. I am pretty confident Rybka would win such a match, but a look through the games might also help resolve the "Compared to GMs, computers are positional patzers" versus "Computers are positionally just as strong or stronger than GMs" debate.
Zappa was able to compete with Rybka because it had amazing scalability ... much better than Rybka. So as hardware increased Zappa was able to close the gap. I noticed that in my test matches at home and this progression was quite linear with hardware increase. It seems that Vas has "fixed" this problem and that Rybka 3 scales as good if not better than Zappa ... so Zappa would have to be improved on to be able to compete. As for Hydra ... well it has been static for a few years, and if you would look at how Hydra played against other engines I would guess that it would have had tremendous problems with Rybka 2.3.2a ... Rybka 3 would be not a good matchup unless again a new version was made or new hardware was used. As for humans ... let's not go there ... I think the best GM would be squirming to get a draw ... and that is not what I want to see.
swami wrote:I think Anthony Cozzie had said he retired from computer chess...
He said something similar before making Zappa Mexico, and he supposedly made it in 10 days! He said that he was just 96% retired (or something like that), this time he could be 99% retired so there's still hope.
And he could be working on a new engine from scratch, so I'll still keep on dreaming (Just like I dream on Achilles, a new version of Fruit, Chess64, etc. )
) in a game vs ZCT in the last WCRCC. I am pretty sure most of the moves Romi found were wrong, otherwise Romi would have won the game. At one time I heard that Chrilly was going to use much faster FPGAs, but I never heard anything else about it after that.. I did read about them coming out with a robotic man who plays chess using the Hydra, but that doesnt necessarily tell us weather or not the chess playing part of it was improved. Since Chrilly was the first to reverse engineer Rybka 1.0, I dont see why he cant do the same with Rybka 3.0, and take some of the ideas and apply it in his venture..
He said that he was just 96% retired (or something like that), this time he could be 99% retired so there's still hope.