60 games Komodo 5 against Top4 at 120m+3s

[TimoK]

Dear Sedat,

Dear Timo,

Nice idea...

But however,if you are interested i can send you my latest Perfect 2012b book

It's much more improved version than Perfect 2012 book

Thanks for the offer, but I will wait until you are fully satisfied with optimizing the opening lines in the Perfect 2012b book. Otherwise somebody could come up with the idea that using a "beta" book isn't good for the performance of engine X or Y. You know what I mean... I think the former 2012 book is already very good balanced and fully sufficiant for this test.

Anyways I'm looking forward to download the new book when it's really finished of course. But take your time! Opening book work is a very hard and complicated task that needs its time...

One thing more,
Why you dont use Auto232 player (instead of Netchess adapter) for the current test ?
At least we will have no any suspicious about the engines playing strength

I used the NetChess adapter only for my former test when playing with 6cores for each engines (because I only have PCs with <=6 cores). The Komodo-Tests are all running with 1 core only for each engine so every match is played on the same PC. I don't need any adapters at all for it.

Best regards
Timo

[Sedat Canbaz]

Dear Sedat,

Dear Timo,

Nice idea...

But however,if you are interested i can send you my latest Perfect 2012b book

It's much more improved version than Perfect 2012 book

Thanks for the offer, but I will wait until you are fully satisfied with optimizing the opening lines in the Perfect 2012b book. Otherwise somebody could come up with the idea that using a "beta" book isn't good for the performance of engine X or Y. You know what I mean... I think the former 2012 book is already very good balanced and fully sufficiant for this test.

Anyways I'm looking forward to download the new book when it's really finished of course. But take your time! Opening book work is a very hard and complicated task that needs its time...

One thing more,
Why you dont use Auto232 player (instead of Netchess adapter) for the current test ?
At least we will have no any suspicious about the engines playing strength

I used the NetChess adapter only for my former test when playing with 6cores for each engines (because I only have PCs with <=6 cores). The Komodo-Tests are all running with 1 core only for each engine so every match is played on the same PC. I don't need any adapters at all for it.

Best regards
Timo

Hello dear Timo,

Actually Perfect 2012b does not mean its a beta version
For example,i am working over this book since several years
And Perfect 2012b book is already ready and compatible for 2 GUIs
Just i need to improve some lines for the rest Chess GUIs

But however,no problem...you can wait till the final release

Btw,the latest public release is Perfect 2012a:
http://www.sedatcanbaz.com/chess/perfect-2012a-book/

I wish you happy testings,
Sedat

[Sedat Canbaz]

We need more games...,but as i expected,Komodo is started to show its real power:
http://www.team-oh.de/livegames6/



Best,
Sedat

[Laskos]

We need more games...,but as i expected,Komodo is started to show its real power:
http://www.team-oh.de/livegames6/



Best,
Sedat

A bit funny, you have conclusions after 6 games, while Timo has now 110 games which show Komodo 5 as weaker than Komodo 4 at this long time control (and pretty significantly so, say 85% confidence).

Kai

[Uri Blass]

We need more games...,but as i expected,Komodo is started to show its real power:
http://www.team-oh.de/livegames6/



Best,
Sedat

A bit funny, you have conclusions after 6 games, while Timo has now 110 games which show Komodo 5 as weaker than Komodo 4 at this long time control (and pretty significantly so, say 85% confidence).

Kai

I think that the picture is now different
komodo has 14-14 against stockfish and Rybka and
is only losing 2 matches 16-12 against houdini and 16-14 against Critter.

[FWCC]

You are using an AMD machine,I thought it was understood that K5 is better on Intel architecture hence the mediocre results.











FWCC-Future World Chess Champion

[Laskos]

We need more games...,but as i expected,Komodo is started to show its real power:
http://www.team-oh.de/livegames6/



Best,
Sedat

A bit funny, you have conclusions after 6 games, while Timo has now 110 games which show Komodo 5 as weaker than Komodo 4 at this long time control (and pretty significantly so, say 85% confidence).

Kai

I think that the picture is now different
komodo has 14-14 against stockfish and Rybka and
is only losing 2 matches 16-12 against houdini and 16-14 against Critter.

After these wins of Komodo, there is still some 70% confidence that it's weaker than Komodo 4. And some >95% confidence it's weaker than Houdini 2.

Komodo 4 results
28.0 – 32.0
29.5 – 30.5
33.0 – 27.0
31.5 – 28.5
122.0 / 50.83%

Kai

[lkaufman]

You are using an AMD machine,I thought it was understood that K5 is better on Intel architecture hence the mediocre results.

That was our early hypothesis, but later data failed to substantiate it. More likely the problem is that the openings used are not representative of openings in general. Also it may be that our sudden death time allocation is not suitable for 2 hour per side games; we didn't expect people to use sudden death (which this is essentially) for such long games - hardly anyone does so for human events.










FWCC-Future World Chess Champion

[Houdini]

That was our early hypothesis, but later data failed to substantiate it. More likely the problem is that the openings used are not representative of openings in general. Also it may be that our sudden death time allocation is not suitable for 2 hour per side games; we didn't expect people to use sudden death (which this is essentially) for such long games - hardly anyone does so for human events.

Quite impressive series of excuses you've used so far for Komodo 5.
It's AMD.
It's Ponder.
It's repeating time controls.
It's sudded death time controls.
It's the openings.
It's not SSE4.

Learning every day .

[Sedat Canbaz]

We need more games...,but as i expected,Komodo is started to show its real power:
http://www.team-oh.de/livegames6/



Best,
Sedat

A bit funny, you have conclusions after 6 games, while Timo has now 110 games which show Komodo 5 as weaker than Komodo 4 at this long time control (and pretty significantly so, say 85% confidence).

Kai

I think you missed to read my previous posting:
We need more games...

So i will try just to make it more clear,
I dont care much about the previous testing,because Komodo 5 has been tested with different openings
And in the previous match,we noticed not very satisfied results by the performance of Komodo 5 (against Houdini 2.0c)

For example, right now Timo is repeating the same match,where this time Komodo 5 is using Perfect 2012 book

And so far... as we see,Komodo 5 is doing not bad under new conditions....

Btw, very soon the new standings should be:

Code: Select all

1. Komodo 5      4.5/8
2. Houdini 2.0c  3.5/8

Regards,
Sedat