Hyper Bullet Match: Houdini 1.5 Vs Rybka 4 (1852 Games)

[mwyoung]

Fritz 12 Gui
Tablebase 5 man
64Mb Hash
TC: Game in 10 seconds.


1 Houdini 1.5 w32 2500 +182 +1195/=355/-302 74.11% 1372.5/1852
2 Deep Rybka 4 w32 2500 -182 +302/=355/-1195 25.89% 479.5/1852

[lucasart]

Fritz 12 Gui
Tablebase 5 man
64Mb Hash
TC: Game in 10 seconds.


1 Houdini 1.5 w32 2500 +182 +1195/=355/-302 74.11% 1372.5/1852
2 Deep Rybka 4 w32 2500 -182 +302/=355/-1195 25.89% 479.5/1852

10 sec and no inc ? this is just silly...

[Damir]

but the result speak for itself, don't you think ?...
even with time increment, I doubt Rybka would have played any better than it already has..

[Graham Banks]

but the result speak for itself, don't you think ?...
even with time increment, I doubt Rybka would have played any better than it already has..

There is no doubt that Houdini 1.5 is stronger than Rybka 4 by at least 50 ELO. Anybody who can't see that is denying the obvious.

[Milos]

Fritz 12 Gui
Tablebase 5 man
64Mb Hash
TC: Game in 10 seconds.


1 Houdini 1.5 w32 2500 +182 +1195/=355/-302 74.11% 1372.5/1852
2 Deep Rybka 4 w32 2500 -182 +302/=355/-1195 25.89% 479.5/1852

Rybka TM is so crappy that even with the best settings that CCRL (pardon me, I mean Rybka) testers have found it's still like 10% worse in hyper bullet than in regular TC against any serious engine.
With default settings is 20% worse. That speaks volumes about Rybka author.
Obvious motive is just another obfuscation. Since most engine authors do their engine tuning in hyper bullet, Vas found a "genious" way to screw them.
Moreover that should give some (false of course) impression that Rybka is better as the TCs increase, because it has some "enigmatic" knowledge, when in reality it has mediocre evaluation but greatly tuned.

[Damir]

but the result speak for itself, don't you think ?...
even with time increment, I doubt Rybka would have played any better than it already has..

There is no doubt that Houdini 1.5 is stronger than Rybka 4 by at least 50 ELO. Anybody who can't see that is denying the obvious.

Judging from the result above, I would say it is much much more than +50 elo increase.