> Why should any customer believe a Cluster bug will be fixed?
With these prices, the bug must be fixed!.
This is Vas's last chance as a leader in the chess computer market. If it fails then he might have to find another job or be as poor as most people
Of course the Hiarcs team can't fix bugs in their cluster software, since there is no :)
> Also you don't want to have many updates therefore and so on.
I don't think ANYONE, ANYWHERE would ever complain that R3 and R4 have had TOO many updates or fixes.. So you guys are safe in that regards.
> There's a distribution problem
Strange, can't imagine any other software-developing company having this problem (others push out updates weekly or even daily to suit their customers wishes), maybe it is just between Vas' ears? Please explain on that what... what distribution-problem makes it impossible to you to fix the Multi-PV-bug even after months? Alex
> Of course the Hiarcs team can't fix bugs in their cluster software, since there is no :)
Or you could put it like this Felix: The Hiarcs team has been too busy addressing the concerns of it's customers and providing said customers with good and timely service to actually spend time building a machine that hardly anyone can afford to rent time on.
one question about improving as vas said above made me wonder..... how much stronger would rybka get if vas or a programmer could use some of the most powerful computers (im talking the new ones you hear about that have 50x the power of vas power) and reved hundreds of moves daily to major plys against plys of 22 that takes ten miniutes to reach on some hardware- how much stronger elo wise would it be if those moves were now the moves instantly played in a set game
reason im interested in these subjects is because i guess that hardware is the only real way (for the most part) for improving chess because computers are so much stronger then us as it is. i think ply and increasing ply through brute power will be the only real way top programmers will be able to create better engines in the coming years- any other thoughts?
mass people are waiting for affordable Rybka 4/8 packs.
i also would not mind to pay for a classical Rybka updated versión, once it is obsolete for the Cluster, which i guess will happen from time to time
I'd like to hope a happy new year for you and all the rybka team.
Glad to see that the cluster project for public is a reality.
I must say that the offer is really very expensive, but one cannot imagine that the beginning prices could be much lower.
The solution could be two projects in parallel, a private cluster computer for future development of rybla, and a public newly built bunch of computers for public use.
Hope also that there could be a product may be named Cluster RYBKA 1.0 that could be sold apart from access to the cloud rybka product, and where we could have updates of the latest Rybka products.
> Would likely earn more profit that way
maybe it would, but i doubt it. i'm no expert, but chess engine users are addicts... and they'll be willing to wait a few hours (even if it's in the middle of the night) when demand is low and the price is right...
Sounds interesting! What I miss here is the "ICC". Maybe you should try to get their customers too. And ICC still has the biggest number of titled players (possible customers).
Just an idea :)
>Is the cluster running as smooth as on my computer?
Depends on your internet connection. From Germany to India the ping is around 250 ms. So bullet games won't be a good idea, but analyzing positions works perfectly well.
"The Rybka Cluster is a high-performance supercomputer located in Minden/Westfalen, Germany. Combining software and hardware engineering, the Rybka Cluster features exclusive Rybka versions not available to the public running on customized hardware built with the specific goal of providing top Rybka performance at an affordable price."
"A computer cluster is a collection of individual computers which act together as a single bigger computer. The challenge is to split the work in such a way that the relatively slow communication between the individual computers is tolerable. When this is accomplished, clustering provides a flexible and relatively inexpensive way to create and maintain a high-performance supercomputer."
May I ask (if it is not a secret) how was the problem of relatively slow communication between the individual computers resolved? I guess that parallel SMP search algorithm implemented in Crafty would not work? I wonder if it is possible to somehow efficiently share memory between processes running on different computers?
I was thinking to suggest to the organizers of the London Chess Classic Chess Tournament to rent say Rybka Cluster 100 (for say two days) and that the winner of the tournament plays (say) a (two game) mini match against the Rybka Cluster. If all the parties would be interested (and the funding secured) that would be a real treat for the spectators I would think. http://www.londonchessclassic.com/index.htm
But with a Rybka staying 30 or 40 ELO behind the strongest freeware, wallets will remain closed.
Vas, if the cluster-rybka is able to do specific astounding things, you should demonstrate it offensively.
And this would be a very nice promotion for UCI-Rybka-5 too.
And it would make the scene curious!
But if there is really nothing to tell us or to show us, I'm afraid that you have lost motivation in Rybka-development. :-(
Am I too anxious?
> When a customer is logged in, the cluster presents itself on his client computer as a
> standard (UCI) chess engine, which can be installed and used in any standard chess GUI.
Uh, i guess someone in this forum applied an EU-wide patent for exactly this, and other parts covering cluster-chess-software, just a few weeks ago...
No, seriously, it is really coincidence :)
anything less than 12 cores u can do from home cpus and as the core increase and the prices fall 24 and 32 way systems are just round the corner...
-1 for the senility, +2 for the drugs. this is ridiculous.
Powered by mwForum 2.25.1 © 1999-2011 Markus Wichitill