![[ng]](/mwf/v12/flags/ng.png "Nigeria")
I wondered that no thread was seen celebrating 5 years of Rybka dominance in Computer Chess World. The question is are we going to celebrate 6 years next year with all these stockfishes and critters and ....
It seems unlikely that any other engine will be able to compete head-to-head with the cluster anytime soon, so Rybka will almost certainly be dominant one year from today. What may change is Rybka's position relative to other commercially available engines. This will depend mostly on how much time Vas puts into releasable engines, as opposed to working on the cluster, cloud Rybka, and consulting for elite GMs.
![[us]](/mwf/v12/flags/us.png "United States")
Well, Rybka cluster is not what we get. It is not commertial available. Nevertheless, Rybka the commercial available engine is loosing ground very fast.
Ricky
Ricky
![[de]](/mwf/v12/flags/de.png "Germany")
"Rybka cluster is not what we get"
we will see :)
we will see :)
I've already PM'd Vas offering to beta test remote Rybka.
I bet Lukas could make a fortune, charging people to remote in to his cluster!
I bet Lukas could make a fortune, charging people to remote in to his cluster!
A brilliant idea wouldbe something like Lukas holding a monthly raffle. $10 buy in, Winner gets 1 hr with the entire cluster. Loser gets $10 credit towards general remote access when it becomes available. Its a win/win. Vas would have a great and novel way to get users to join his cloud and users would get a chance to see what its like to run rybka on a 200 core machine.
After a month, $10 ticket x say 50 sells =$500 a month 500 raffle ticket sells = $5000/month :) :)
Now he's makin $$$ and what about Chessbase, would they advertise on their site such an event?(raffle) I'm sure there are at least 100 titled chess masters out there who would love a novel chance to get ahold of the cluster for an hour. Winner might even sell some of his time or divy out to others for his own special fee.
After a month, $10 ticket x say 50 sells =$500 a month 500 raffle ticket sells = $5000/month :) :)
Now he's makin $$$ and what about Chessbase, would they advertise on their site such an event?(raffle) I'm sure there are at least 100 titled chess masters out there who would love a novel chance to get ahold of the cluster for an hour. Winner might even sell some of his time or divy out to others for his own special fee.
> "Rybka cluster is not what we get"
>
> we will see :)
But that is cloud Rybka right?
I wondered that no thread was seen celebrating 5 years of Rybka dominance in Computer Chess World.
Can a chess engine dominate the computer chess world without a commercially available version? I think the answer is yes. In fact, it has already been done by earlier monsters like Deep Blue and Hydra.
Can a chess engine dominate the computer chess world without a commercially available version? I think the answer is yes. In fact, it has already been done by earlier monsters like Deep Blue and Hydra.
sure, but we are seeing the ascendancy of completely free for all engines making remarkable gains. i will use cloud rybka at a base plan if it's reasonable, but it does appear that rybka's position as top dog in the PC market is in jeopardy. critter 1.00, or stockfish 2.00 (or even later editions) could easily surpass it before the release of rybka 5. and more power to them.
It's very likely that a two tiered market will develop; Freely and commercially available UCI engines stronger than the current R4, and private engines that are much stronger. Vas will probably stay comfortably ahead of the other private engines for the foreseeable future. Vas can also stay ahead in the commercially available UCI engines if he wants to, but only at the expense of helping all the other UCI engines develop more quickly. It should be noted that the rapid recent gains in freely (and commercially) available engines is due to two sources; Tord's Stockfish effort, and the reverse engineering of Rybka.
i don't quarrel with yr latter statement, but i do think it's interesting that critter has posted consistent (massive) gains over a relatively consistent timeframe. i am positive both stockfish & reverse engineered rybka are part of this (as they should be, imo; who doesn't want to put forth the best they can), but it seems just as likely vida is just a top-notch programmer who hasn't come close to fulfilling his potential. i don't know if critter is open-source, but with the openness of the stockfish project, & vida's extremely rapid rise, he is either being incredibly canny for no financial reason, or a masterful computer chess programmer. i'd have to gamble on the latter. (he's also stated, i believe -- could be mixing messages with the stockfish team -- that he'll never sell critter).
I'm sure the developer of critter is both a top notch programmer and a skilled algorithm designer, but it's harder to be a pioneer than to take advantage of previous work. It will be interesting to see the pace of development when other engines reach the level of Rybka 4, assuming that Vas doesn't release another version anytime soon.
but the thing is that with critter, as with stockfish, we're seeing incremental gains that are closing the gap. this isn't a case of a programmer popping up with a rybka 3 level engine.
There are two ways to use other people's work:
1) Use other people's work as a framework to be modified. For instance, start with Stockfish and meld in ideas from reverse engineered Rybka. I believe there is a non-open source engine that I can't mention here that falls into this category.
2) Use your own framework and include algorithms from open source engines. This is a slower approach, but has more long term potential. Critter is probably in this category.
Either way, computer chess isn't an area where you start with just your own ideas and use them to build a competitive engine. It didn't work that way for Rybka 1 Beta, and it isn't that way for Critter either. Both made extensive use of algorithms developed by people going all the way back to Shannon.
1) Use other people's work as a framework to be modified. For instance, start with Stockfish and meld in ideas from reverse engineered Rybka. I believe there is a non-open source engine that I can't mention here that falls into this category.
2) Use your own framework and include algorithms from open source engines. This is a slower approach, but has more long term potential. Critter is probably in this category.
Either way, computer chess isn't an area where you start with just your own ideas and use them to build a competitive engine. It didn't work that way for Rybka 1 Beta, and it isn't that way for Critter either. Both made extensive use of algorithms developed by people going all the way back to Shannon.
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
Who or what is Shannon?
![[nz]](/mwf/v12/flags/nz.png "New Zealand")
Maybe there could be a Cluster version for those people with the same financial ability as Lukas to get top of the line hardware.It counts me out, but there must be a few thousand rich chess players in the whole world that could be tempted.Might be the only way to make money on it, apart from a mobile version which would make money overnight.
A cluster wouldn't have to cost a fortune though. 4 cheap quads = 16 cores for not too much money...
Given time, I'd say 16 core processors are a done deal (be a few years), for the enthusiast market. Its doubtful we'll see a chip with more than 16 cores though. Even now, the founderies have to get more chips out of a platter in order to make money on the multi billion dollar costs for a chip manufacturing plant.
Even with 8 core CPU's, your starting to see diminishing returns on core counts for the retail market. Even in chess engines, you have to throw more and more cores at the cluster in order to squeeze out a marginal gain.
I don't have a graph, but as I stated a few months ago: 64 cores isn't double the power of 32 cores.
Even with 8 core CPU's, your starting to see diminishing returns on core counts for the retail market. Even in chess engines, you have to throw more and more cores at the cluster in order to squeeze out a marginal gain.
I don't have a graph, but as I stated a few months ago: 64 cores isn't double the power of 32 cores.
my next (C) will be 8c/16t sandy bridge. already salivating. unless interlagos is something spectacular.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill