![[cz]](/mwf/v12/flags/cz.png "Czech Republic")
Greetings to everyone,
I have a problem with settings of RYBKA and Fritz when I want to start an Engine match between them. I have installed RYBKA 2.3.2a yesterday and I use the interface of FRITZ 8. I have some other free UCI engines installed, so occasionally I let the programs play some engine matches. The problem is, that RYBKA and FRITZ do not want to play with each other probably :-), for when I start the match, one of them - mostly black - just stops playing and lose on time. I have discovered that this happens by the time they both finish their lines in Opening library book. When RYBKA or FRITZ playing with other engines, no problems are visible. The funny thing is that when I used just DEMO of RYBKA everything was OK. Could someone possibly give me a hint how to solve this prob?
Thank you in advance.
K.
I have a problem with settings of RYBKA and Fritz when I want to start an Engine match between them. I have installed RYBKA 2.3.2a yesterday and I use the interface of FRITZ 8. I have some other free UCI engines installed, so occasionally I let the programs play some engine matches. The problem is, that RYBKA and FRITZ do not want to play with each other probably :-), for when I start the match, one of them - mostly black - just stops playing and lose on time. I have discovered that this happens by the time they both finish their lines in Opening library book. When RYBKA or FRITZ playing with other engines, no problems are visible. The funny thing is that when I used just DEMO of RYBKA everything was OK. Could someone possibly give me a hint how to solve this prob?
Thank you in advance.
K.
![[nl]](/mwf/v12/flags/nl.png "Netherlands")
Are you using ponder=on during these matches?. If so, try to disable pondering.
![[hu]](/mwf/v12/flags/hu.png "Hungary")
I would suggest checking in the task manager to see what is happening when the games start. It might be pondering, it might be something else swamping your computer.
Vas
Vas
If nothing else helps, delete the file "engines.recent" from ChessProgram8's application data.
I never got it what this file is for anyway, but at least twice I solved "mysterious" problems by simply deleting it. I think it is possible that it starts to cause problems if it grows too big.
I never got it what this file is for anyway, but at least twice I solved "mysterious" problems by simply deleting it. I think it is possible that it starts to cause problems if it grows too big.
Unfortunately, I can't find the file you have mentioned in your message. Could you be so kind as to give me a closer guide to find it? Thanks.
K.
K.
This depends on the Windows version. I have it under
C:\Users\User\AppData\Roaming\ChessBase\ChessProgram8
C:\Users\User\AppData\Roaming\ChessBase\ChessProgram8
I have checked the tips about pondering and task manager and it is now clear they were useful. As I can see, the pondering was the main problem, when unchecked during engines settings, the games took their part. But still, after 2 or 3 games one of the engines freezes. Maybe there could be a problem with my RAM (512 MB), especially RYBKA takes a great amount of the memory - over 250 MB and FRITZ's usage of the RAM is rising along with the game's progress. It seems to me that nothing else swamps the computer significantly. So I assume that extension of my RAM could solve the problem. Any other tips are welcome.
K.
K.
What are the hash sizes you use for the engines?. On a 512MB system these should be really small to prevent disk trashing.
I use 256 MB for my HASH tables. Till today, there was not a single problem with it, but I try to reduce the size and see what happens. Thanks for tip.
K.
K.
Please note that in engine-engine matches the system will use 2*256MB memory, so that is a problem on a 512MB machine. Try reducing them to 128 MB or better 64MB. Or even better, buy more memory :)
Did you recently add end game tablebases?. Every table base will take some additional memory.
>Till today, there was not a single problem with it
Did you recently add end game tablebases?. Every table base will take some additional memory.
The problem is definitely solved. You was obviously right with the size of RAM, I reduced it to 128 MB for both programs and the games run with no single problem - except RYBKA is not winning every time :-). I even put back in life the pondering function and still there is no problem.
Thank you for opening my eyes.
K.
Thank you for opening my eyes.
K.
Be carefull with pondering on, on a single machine. The most greedy engine will get the most CPU time this way. If you want to make sure both engines get equal amount of CPU time you have to test with ponder is off.
Ponder off is more effective on a single system, in terms of depth per time. The expected reply will be played in (roughly) 65% of the cases. So, to ~35% one engine will waste half the cpu performance while the other one gets only half of the cpu, too. But with ponder off, each engine has 100% (or 95%+) while having the move.
![[no]](/mwf/v12/flags/no.png "Norway")
Be carefull with pondering on, on a single machine. The most greedy engine will get the most CPU time this way. If you want to make sure both engines get equal amount of CPU time you have to test with ponder is off.
An engine cannot be greedy itself. The OS' scheduler decides how much cpu time each process will get. The scheduler is, however, not always fair, but on average most of them try to be. If the two engines are given different process priorities, it will of course be highly unfair.
I of course agree that tournaments should be run with ponder off! :)
An engine cannot be greedy itself. The OS' scheduler decides how much cpu time each process will get. The scheduler is, however, not always fair, but on average most of them try to be. If the two engines are given different process priorities, it will of course be highly unfair.
I of course agree that tournaments should be run with ponder off! :)
Correct, but on a multiprocessor/core system Rybka gives all it's own threads (expect one) 'below normal' priority. If the other engine uses 'normal' you have a problem.
It was a typical observation that Chessmaster would constantly use (nearly) 100% CPU if run in it's own interface, in older CM versions. It did that even if ponder - which was called "deep thinking" in CM - was switched off. I have often seen postings on rgcc and elsewhere, where people had played engine matches manually, Fritz vs. Chessmaster and asked "Why does Fritz lose all games...?!" That was for that reason. Appearantly it set it's own task to a high priority or something like that.
The King engine alone does not have that behaviour though, although if an interface can do it, I assume an engine could do it as well. That is why I think an engine actually CAN be greedy, it's just unusual or any good engine should of course not do it. But if I don't forget it, I always check CPU consumption with a task tool if I test a new engine in matches.
(In addition, there are other chess programs (not engines) where disabling ponder was not possible, like Socrates.)
The King engine alone does not have that behaviour though, although if an interface can do it, I assume an engine could do it as well. That is why I think an engine actually CAN be greedy, it's just unusual or any good engine should of course not do it. But if I don't forget it, I always check CPU consumption with a task tool if I test a new engine in matches.
(In addition, there are other chess programs (not engines) where disabling ponder was not possible, like Socrates.)
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill