Zappa Mexico Vs Rybka SMP

[Daniel Charles]

I will begin to make the big test between Rybka and Zappa on 8 cpu


A/ What is the time control you would like to see them play ?

1/The same time control like at Mexico ?
2/ A shorter time control
3/ A Longer time control

B/ what is the opening I could use for the test ?

1/ Draw book , (like perfect9 , HS6moves, ...)
2/ very strong book (Sheebar,TourBook, Rybka II,HS Rybka....)


Thanks


Daniel

[Werner]

Hi Charles,
I vote for

A) 40 moves in 40 minutes
B) draw book like perfect12.33.ctg

[Uri Blass]

I will begin to make the big test between Rybka and Zappa on 8 cpu


A/ What is the time control you would like to see them play ?

1/The same time control like at Mexico ?
2/ A shorter time control
3/ A Longer time control

B/ what is the opening I could use for the test ?

1/ Draw book , (like perfect9 , HS6moves, ...)
2/ very strong book (Sheebar,TourBook, Rybka II,HS Rybka....)


Thanks


Daniel

I vote for 3(longer time control) and 1.
I suggest 180/40 repeating(assuming that you use ponder off) and 120/40 in case of using ponder on.

Uri

[Sedat Canbaz]

Hello Daniel,

Seems to be a interesting test...

And its will be great match to see the performance between RybkaII.ctg (Rybka ) against my new latest Perfect X beta 3.ctg (Zappa) on your fast hardware ?

If you are interested i can send you my latest beta book ?!


Best,
Sedat

[Spock]

I would vote for neutral books for each engine, and I would go for a similar time control to Mexico, i.e. something that is actually used in chess engine tournaments, which is never a repeating time control

[geots]

I would vote for neutral books for each engine, and I would go for a similar time control to Mexico, i.e. something that is actually used in chess engine tournaments, which is never a repeating time control

I agree with you Ray. But in the end it wont matter much. Books, number of threads, time controls- it's all going to be pretty much irrelevant. When Zappa and Rybka neither one have someone overseeing each game, picking openings, i.e.- they just start and are left to their own devices, Rybka will add a complete new meaning to the phrase "kicking ass"

[Ovyron]

Hi George,

Remember when you told me to run a match of 50 games between Rybka 2.3.2a and Zappa Mexico? I was going to use TourBookII with learning on for Zappa and Perfect 12.33 for Rybka.

Well, since I can't use the computer while the games are being played (nor can I let the games to be played overnight), it has been very slow, but I finally have played 10 games, and the result is:

Code: Select all

 40'/40 repeating

                         1234567890
1   Rybka 2.3.2a 32-bit  1½½111½½½0    6.5/10
2   Zappa Mexico         0½½000½½½1    3.5/10

It was on my Athlon 2.1 Ghz with 64 MB of RAM for the engines.

The point was to prove that all Zappa needed to be on the level of Rybka on 1 CPU 32bit machines was a good book, but it seems I was wrong. And though 10 games are very inconclusive, I deem myself defeated and I won't be playing more games (They just take so much time to play and I don't think Zappa can recover).

So, congrats George.

Download games here or here.

[geots]

Hi George,

Remember when you told me to run a match of 50 games between Rybka 2.3.2a and Zappa Mexico? I was going to use TourBookII with learning on for Zappa and Perfect 12.33 for Rybka.

Well, since I can't use the computer while the games are being played (nor can I let the games to be played overnight), it has been very slow, but I finally have played 10 games, and the result is:

Code: Select all

 40'/40 repeating

                         1234567890
1   Rybka 2.3.2a 32-bit  1½½111½½½0    6.5/10
2   Zappa Mexico         0½½000½½½1    3.5/10

It was on my Athlon 2.1 Ghz with 64 MB of RAM for the engines.

The point was to prove that all Zappa needed to be on the level of Rybka on 1 CPU 32bit machines was a good book, but it seems I was wrong. And though 10 games are very inconclusive, I deem myself defeated and I won't be playing more games (They just take so much time to play and I don't think Zappa can recover).

So, congrats George.

Download games here or here.

Thanks for the info. Im certainly no expert, but in my gauntlet im about to post the final results in- i used the same generic book for both. And same results after 20 games- Rybka more or less doubled up on Zappa. The only thing that will keep Zappa close on 8 cores is Rybka doesnt scale that well on 8 cores- but i imagine it scales enough to handle Zappa. If you see my results, Zappa on 1 CPU is between Naum and Junior 10, just bunched in with the rest that cant deal with Rybka. That engine is just too strong- in a class by itself. The only engine head up on 1 cpu that can even make it close with Rybka at times is Fritz 10. Thankd you again for your effort- and you are wise to go ahead and stop it now. CPU time is precious.

Best to You,

[Tony Thomas]

Hi George,

Remember when you told me to run a match of 50 games between Rybka 2.3.2a and Zappa Mexico? I was going to use TourBookII with learning on for Zappa and Perfect 12.33 for Rybka.

Well, since I can't use the computer while the games are being played (nor can I let the games to be played overnight), it has been very slow, but I finally have played 10 games, and the result is:

Code: Select all

 40'/40 repeating

                         1234567890
1   Rybka 2.3.2a 32-bit  1½½111½½½0    6.5/10
2   Zappa Mexico         0½½000½½½1    3.5/10

It was on my Athlon 2.1 Ghz with 64 MB of RAM for the engines.

The point was to prove that all Zappa needed to be on the level of Rybka on 1 CPU 32bit machines was a good book, but it seems I was wrong. And though 10 games are very inconclusive, I deem myself defeated and I won't be playing more games (They just take so much time to play and I don't think Zappa can recover).

So, congrats George.

Download games here or here.

Thanks for the info. Im certainly no expert, but in my gauntlet im about to post the final results in- i used the same generic book for both. And same results after 20 games- Rybka more or less doubled up on Zappa. The only thing that will keep Zappa close on 8 cores is Rybka doesnt scale that well on 8 cores- but i imagine it scales enough to handle Zappa. If you see my results, Zappa on 1 CPU is between Naum and Junior 10, just bunched in with the rest that cant deal with Rybka. That engine is just too strong- in a class by itself. The only engine head up on 1 cpu that can even make it close with Rybka at times is Fritz 10. Thankd you again for your effort- and you are wise to go ahead and stop it now. CPU time is precious.

Best to You,

Fritz 10? WTF, you forgot about Shredder on a single CPU 32bit system.

[Graham Banks]

Hi George,

Remember when you told me to run a match of 50 games between Rybka 2.3.2a and Zappa Mexico? I was going to use TourBookII with learning on for Zappa and Perfect 12.33 for Rybka.

Well, since I can't use the computer while the games are being played (nor can I let the games to be played overnight), it has been very slow, but I finally have played 10 games, and the result is:

Code: Select all

 40'/40 repeating

                         1234567890
1   Rybka 2.3.2a 32-bit  1½½111½½½0    6.5/10
2   Zappa Mexico         0½½000½½½1    3.5/10

It was on my Athlon 2.1 Ghz with 64 MB of RAM for the engines.

The point was to prove that all Zappa needed to be on the level of Rybka on 1 CPU 32bit machines was a good book, but it seems I was wrong. And though 10 games are very inconclusive, I deem myself defeated and I won't be playing more games (They just take so much time to play and I don't think Zappa can recover).

So, congrats George.

Download games here or here.

Thanks for the info. Im certainly no expert, but in my gauntlet im about to post the final results in- i used the same generic book for both. And same results after 20 games- Rybka more or less doubled up on Zappa. The only thing that will keep Zappa close on 8 cores is Rybka doesnt scale that well on 8 cores- but i imagine it scales enough to handle Zappa. If you see my results, Zappa on 1 CPU is between Naum and Junior 10, just bunched in with the rest that cant deal with Rybka. That engine is just too strong- in a class by itself. The only engine head up on 1 cpu that can even make it close with Rybka at times is Fritz 10. Thankd you again for your effort- and you are wise to go ahead and stop it now. CPU time is precious.

Best to You,

Fritz 10? WTF, you forgot about Shredder on a single CPU 32bit system.

Shredder 11 gets wasted in a head to head battle with Rybka.