Zappa Mexico (8 cpu) vs Top free UCI engine

[Spock]

I'm pretty sure Deep Fritz 10 supports 8 CPUs as I think that is what they played on vs Kramnik ? I'm pretty sure Deep Junior 10 will as well. Hiarcs 11.2 definitely does.

[Tony Thomas]

If not a 100 point difference, then pretty massive, 50 elo minimum. 1.3.1 gets better results, bar none. We might as well put Rybka 1.2 in a match vs Zappa Mexico to see which engine is better.

According to CCRL the difference between both versions is 42 points in 40/40. In 40/4 the rating difference is reduced to a mere 26 point advantage for the current version. Since Daniel's time control is even faster, I doubt that the gains the new version would make is minimal. I still dont understand why he used the old version, so far for me a personality of the old version performs better than the current version, I guess it depends on what kind of computer a person is using.

[Daniel Charles]

You are right Ray ,

Deep Fritz, Deep Junior , Hiars should supports fully 8 cpu ...
maybe Hiars testers could tell us more about this .

Deep Fritz was on 2 Xeon 5160 when it played against Kramnik (4 cpu X 2.66 GHZ )

About the choice of Toga, this is just because This is not really Toga 1.2.1 but a new beta version that I just wanted to test.


Thanks


Daniel Charles

[jdart]

I'd be interested in seeing the games.

[George Tsavdaris]

I'd be interested in seeing the games.

Me too, but with evaluations, depths, etc by the engines in the PGN.....

[Eelco de Groot]

I'd be interested in seeing the games.

Me too, but with evaluations, depths, etc by the engines in the PGN.....

Well personally I would be very interested in any news about the new Toga beta that Daniel apparently has for testing. It's disguised as a Toga 1.2.1a but he don't fool us! We're on to him! Daniel, is it, perhaps, related to some rumours I read elsewhere about a SMP Toga made by Teemu?

Regards, Eelco

[Vempele]

Well personally I would be very interested in any news about the new Toga beta that Daniel apparently has for testing. It's disguised as a Toga 1.2.1a but he don't fool us! We're on to him! Daniel, is it, perhaps, related to some rumours I read elsewhere about a SMP Toga made by Teemu?

Unrelated AFAIK and not by me (Zach did all the hard work, I just added rootsplit and fixed any bugs I could find).

[Eelco de Groot]

Well personally I would be very interested in any news about the new Toga beta that Daniel apparently has for testing. It's disguised as a Toga 1.2.1a but he don't fool us! We're on to him! Daniel, is it, perhaps, related to some rumours I read elsewhere about a SMP Toga made by Teemu?

Unrelated AFAIK and not by me (Zach did all the hard work, I just added rootsplit and fixed any bugs I could find).

That sounds really interesting Teemu, also because Zach wrote a while back in Winboard Forum that he had not gotten his SMP version to work. So my congratulations to the both of you then!

Daniel writes about a Toga 1.2.1a par, and he tests all kinds of multithreading engines in his tournament, so maybe this means there are now two new parallel Togas and we hear about them on exactly the same date! Is this what they call synchronicity? Maybe they have time to do another match in Mexico City..

Is there any strength increase noticeable with your version Teemu?

Regards, Eelco

[Vinvin]

Could you estimate the speed up nodes/sec on 1 CPU vs nodes/sec 8 CPUs on your machine, please ?

Why no response to my question

[Daniel Charles]

Dear Vincent

very good question

I Just Upload a picture from the KN/s from Zappa 8 CPU (2891 kN/s for the 32 bits version)

http://www.superchessengine.com/september07.htm

1 CPU around 400 Kn/s

Rybka SMP from Vasik
2 Processors: 1.7
4 Processors: 2.8
8 Processors: 4.4


Zappa Mexico scale very well on 8 CPU !!! 28XX/ 4XX = > 7 or more


maybe Anthony Cozzie could tell us more about this

Daniel