> The Fritz engine is a completely new version
Just like Rybka 4.1 was a completely new version of Rybka 4
Anyway, quite surprised someone still cared about rybka . You must be old as the last major release was almost 10 years ago, young chess players today will probably have no idea what rybka is.
>So why, the forum is still named "rybka"?
No one changed the name. A lot of us are old Rybka fans that supported Vas through all the drama.
Some corrections to the above, Rybka hasn't exactly been 'rebranded' as Fritz, it's just that Fritz now uses an updated Rybka engine. It isn't all that strong by today's standards, but was +100 Elo I believe on the previous Fritz which is what chessbase was looking for. So there could still be a Rybka 5, it's just really unlikely because it would have to best today's programs and this is now extremely difficult.
Also, there wasn't a '1-core' rental scheme planned, but instead a zillion core rental scheme, the idea being that you would rent time on an enormous cluster of top hardware for your analysis and then access it remotely. It ran a special version of Rybka that was designed to take advantage of multiple machines in a cluster setup.
Vas abandoned the Rybka project entirely shortly after its inception however, and when other engines surpassed Rybka it didn't make much sense as the persons involved weren't going to run other engines, and they wouldn't run on a cluster setup out of the box anyways.
>Also, there wasn't a '1-core' rental scheme planned,
Vas did plan to launch single core rental scheme as a 'starter' package - http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=259914#pid259914
And should cost 750 euros per year - http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=13820
There was also an FPGA version planned, it would have been cool if it had been launched but I think it would had been just a high tech paper weight in a few years with the arrival of stockfish framework had it been launched.
>Vas did plan to launch single core rental scheme as a 'starter' package
I don't think it was ever planned, was under consideration more like. Even on 1 core though, if the Elo difference to the client's box is high enough it can make sense financially to rent like this, it's just that such an Elo is really high. The nice thing about renting for the developer is it completely prevents any software piracy or reverse engineering.
I gave Vas the chess accelerator idea and he was looking for ways to implement it, but he didn't like the software tools that came with the FPGA stuff he looked at, and I think he considered FPGA a reasonable prerequisite to an ASIC. All computer hardware becomes a high tech paperweight eventually. If it added enough Elo, or nodes/sec it would still make financial sense today though. If I could say pay $ 200 for an accelerator board that would give me 500 mN/sec vs. say my 20 mN/sec on a fairly modern PC, that would be quite interesting because otherwise you'd need a ~$50k machine to achieve that.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill