Zappa Mexico II v Naum 3- Is Naum Running Out of Time?!

[geots]

Zappa Mexico II 32-bit v Naum 3 32-bit



Fritz 11 GUI
Intel Dual Core
Single Processor- 3.2ghz
128mb hash
3 & 4 man egtb
Perfect 12.32.ctg book (w/ 12 move CCRL limit)
40 moves in 58 minutes repeating (Adapted to CCRL 40/40)
Both Engines = 32-bit
Both Engines = 1 CPU
Match = 30 Games Total
Run by George Speight


After Rybka Match- Possible Naum Retribution?

Latest/Most Recent Version of Zappa Mexico Is Being Used Here

Updates Posted Every Few Days

Games 19 thru 22: 2 draws and 1 win each (Same result as prior 4 games)

To Date: Zappa Mexico II- 12.0 , Naum 3- 10.0


Beginning to look like Naum is going to have a tough time making up a 2 game deficit with 8 to play.


PGN package available at match conclusion.

[Uri Blass]

The rating list that I see suggest that Naum3 is not only stronger than Zappa Mexico 32 bits but also stronger than Zappa Mexico 64 bits.

I wonder if Naum is relatively twice slower in your computer or if there is a different reason for your results.

The main problem when testers use different computers is that
I suspect that it may be possible that we may have something like the following:

Computer A:
Zappa 200 knodes per second
Naum 400 knodes per second

Computer B
Zappa 400 knodes per second
Naum 400 knodes per second

I think that it may be good to have some table of speed for the programs that are tested in different computers based on time to solve some test positions without learning(in case that the program is not deterministic for example because of parallel search we need to repeat the experiment more than one time) .

Here are results from more than one rating list when Naum3 seems to be clearly stronger than Zappa Mexico

CEGT 40/20

39 Naum 3.0 w32 1CPU 2892 25 25 457 57.4% 2840 39.6%
60 Zappa Mexico II w32 1CPU 2850 34 34 234 50.6% 2846 43.2%


CCRL 40/40
Naum 3 32-bit 2966 +33 −32 53.8% −25.6 50.0% 278
Zappa Mexico II 64-bit 2945 +35 −35 48.2% +10.5 47.8% 247
Zappa Mexico II 32-bit 2941 +85 −84 53.8% −25.5 51.3% 39

CEGT 40/4

66 Naum 3.0 w32 1CPU 2853 13 13 2129 62.8% 2762 31.0%
71 Zappa Mexico x64 1CPU 2848 11 11 2429 50.3% 2846 35.8%
78 Zappa Mexico II w32 2CPU 2833 17 17 1058 45.4% 2865 37.1%

[geots]

The rating list that I see suggest that Naum3 is not only stronger than Zappa Mexico 32 bits but also stronger than Zappa Mexico 64 bits.

I wonder if Naum is relatively twice slower in your computer or if there is a different reason for your results.

The main problem when testers use different computers is that
I suspect that it may be possible that we may have something like the following:

Computer A:
Zappa 200 knodes per second
Naum 400 knodes per second

Computer B
Zappa 400 knodes per second
Naum 400 knodes per second

I think that it may be good to have some table of speed for the programs that are tested in different computers based on time to solve some test positions without learning(in case that the program is not deterministic for example because of parallel search we need to repeat the experiment more than one time) .

Here are results from more than one rating list when Naum3 seems to be clearly stronger than Zappa Mexico

CEGT 40/20

39 Naum 3.0 w32 1CPU 2892 25 25 457 57.4% 2840 39.6%
60 Zappa Mexico II w32 1CPU 2850 34 34 234 50.6% 2846 43.2%


CCRL 40/40
Naum 3 32-bit 2966 +33 −32 53.8% −25.6 50.0% 278
Zappa Mexico II 64-bit 2945 +35 −35 48.2% +10.5 47.8% 247
Zappa Mexico II 32-bit 2941 +85 −84 53.8% −25.5 51.3% 39

CEGT 40/4

66 Naum 3.0 w32 1CPU 2853 13 13 2129 62.8% 2762 31.0%
71 Zappa Mexico x64 1CPU 2848 11 11 2429 50.3% 2846 35.8%
78 Zappa Mexico II w32 2CPU 2833 17 17 1058 45.4% 2865 37.1%

I will do the best i can here. I am running two computers, Uri. An XP and a Vista. Where you came up with those stats im not real sure. But i can promise you that you will never, ever see either of these engines where kns are concerned- double from one computer of mine to the other. And in the end that is all i have control over- my two computers. Tho i do have them naturally adapted to CCRL 40/40 time controls. As to ratings & percentages you have quoted above- i dont think they apply here. At least not in 32-bit. Tho i could be wrong, i know of no one running a head-up engine-engine 32 bit 1CPU match other than me for CCRL. With either Naum v Rybka or Naum v Zappa. AFAIK im the only one. And you cannot compare head up results to games against common opponents. As for CEGT, i havent a clue as to their machines, specs, controls or the like. And im not about to get caught up in that trap. Add to that the fact that, first, only 22 games have been played. Thats not exactly a landmark figure. Secondly, in 8 more games you will have access to the complete 30 game package between Zappa-Naum, being run on my Vista box. You already have access to the 30 game package that was run prior to this between Naum- Rybka, which was run on my XP box. That should be plenty of ammo for you to blow me out of the water. Just be patient.

Best,


PS: Personally- no stats to back it up, just from watching the play- my gut feeling tells me that head-up against each other Zappa Mexico II 32-bit is slightly stronger than Naum 3 32-bit on 1 CPU. If Naum ends up with a higher rating, it will be due to the fact that it did better against a number of common opponents than did Zappa. Nothing scientific here- just gut hunch.

PS2: Any other head-up results in engine-engine matches between the above mentioned done by CCRL were, Im guessing, Blitz results. Or they were not 32 bit, single CPU.

[geots]

One note i failed to mention, Uri. I believe it was Game 14- you can check to be sure, in the Zappa - Naum match to which we refer now. It was a dead draw, and i adjudicated it as such. I did not feel like watching them wade around for another 100 moves to decide it themselves. I had bigger fish to fry at the moment.

Best,

[Spock]

PS: Personally- no stats to back it up, just from watching the play- my gut feeling tells me that head-up against each other Zappa Mexico II 32-bit is slightly stronger than Naum 3 32-bit on 1 CPU.

Yes. You can see on our list that on 1 CPU, Naum 3 x64 has played Zappa Mexico II x64. Zappa WON that match 20.5 - 15.5

So yes, on 1 CPU Zappa beats Naum in a direct fight.