A new LOW-MEDIUM tournament with Colchess an Raffaela

[Carlos E.A.Drake]

Geschrieben von:/Posted by: Carlos E.A.Drake at 17 February 2000 21:36:33:
How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2
TSCP 1.42/1.50 ......... 8 1/2
Raffaela 0.0.0.2 ....... 2

Pentium 133 with ~ 8 Mb HT and a level 5' + 2 "
See the PGN and the XLS files in Frank's page, I sended its./
Kind regards
Carlos

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 17 February 2000 21:40:15:
Als Antwort auf:/As an answer to: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Carlos E.A.Drake at 17 February 2000 21:36:33:

How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2
TSCP 1.42/1.50 ......... 8 1/2
Raffaela 0.0.0.2 ....... 2

Pentium 133 with ~ 8 Mb HT and a level 5' + 2 "
See the PGN and the XLS files in Frank's page, I sended its./
Kind regards
Carlos

Did you use 5.3? 5.4 is significantly stronger.
It is pretty amazing that TSCP did so well.


My FTP site

[Pete Galati]

Geschrieben von:/Posted by: Pete Galati at 17 February 2000 22:31:48:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Dann Corbit at 17 February 2000 21:40:15:

How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2
TSCP 1.42/1.50 ......... 8 1/2
Raffaela 0.0.0.2 ....... 2

Pentium 133 with ~ 8 Mb HT and a level 5' + 2 "
See the PGN and the XLS files in Frank's page, I sended its./
Kind regards
Carlos

Did you use 5.3? 5.4 is significantly stronger.
It is pretty amazing that TSCP did so well.

My guess (just a guess) would be that the most recent version of TSCP might do even better because it now understands time from what I hear. To be honest though, I still have not played any games against it to see if that's true, I've only run "bench" on a few different compiles of it (mine was the slowest compile)
This might be able to be improved on, see the "livello" (level) control in Raffaela's own interface, but as you'd see, setting it too high (much higher at all) would cause time problems.
Pete

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 17 February 2000 23:50:38:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Dann Corbit at 17 February 2000 21:40:15:

How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2

Did you use 5.3? 5.4 is significantly stronger.

Thanks a lot. Always interesting to find out how my program does against others.
Yes I think it is. You just wait and see how much I've just improved it by adding in one simple line and altering another. I _think_ I can prove that I can cut ouf up to half the tree with absolutely no drop in analysis accuracy. Wait for version 6.0 (which should be out before the end of the weekend)
Grr... Sjeng beat ColChess in the other tournament. Oh well.
Thanks a lot,
Colin

ColChess Homepage

[Frank Quisinsky]

Geschrieben von:/Posted by: Frank Quisinsky at 18 February 2000 00:02:56:
Als Antwort auf:/As an answer to: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Carlos E.A.Drake at 17 February 2000 21:36:33:
Hi Carlos,

See the PGN and the XLS files in Frank's page, I sended its./

And I added the tournament on my webpage !
Thanks Carlos
Frank

[Imran Hendley]

Geschrieben von:/Posted by: Imran Hendley at 18 February 2000 02:26:10:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Colin Frayn at 17 February 2000 23:50:38:

How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2

Did you use 5.3? 5.4 is significantly stronger.

Thanks a lot. Always interesting to find out how my program does against others.
Yes I think it is. You just wait and see how much I've just improved it by adding in one simple line and altering another. I _think_ I can prove that I can cut ouf up to half the tree with absolutely no drop in analysis accuracy. Wait for version 6.0 (which should be out before the end of the weekend)
Grr... Sjeng beat ColChess in the other tournament. Oh well.
Thanks a lot,
Colin

Great! How strong do you anticipate this to be? If you tell me some programs you think it can contend with I will try to run a tournament.

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 18 February 2000 17:43:33:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Imran Hendley at 18 February 2000 02:26:10:

Yes I think it is. You just wait and see how much I've just improved it by adding in one simple line and altering another. I _think_ I can prove that I can cut ouf up to half the tree with absolutely no drop in analysis accuracy. Wait for version 6.0 (which should be out before the end of the weekend)

Great! How strong do you anticipate this to be? If you tell me some programs you think it can contend with I will try to run a tournament.

Hmmm... well after significant testing it appears to perform slightly worse in WAC. Still not sure why.
If I can't get it's WAC score back up then I'll probably scrap this idea.
Cheers,
Col

ColChess Homepage

[Aaron]

Geschrieben von:/Posted by: Aaron at 18 February 2000 18:10:29:
Als Antwort auf:/As an answer to: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Carlos E.A.Drake at 17 February 2000 21:36:33:

How I talk with Colin, I released a new low-medium tournament, with some
new and not so programs for WB ( Colchess and Raffaela )
Techno - Knight 1.5 ... 21 1/2
Dragon 3.11 ........... 19 1/2
Averno 0.24 ........... 14
Colchess 5.3 .......... 10
Sjeng Vc ............... 8 1/2
TSCP 1.42/1.50 ......... 8 1/2
Raffaela 0.0.0.2 ....... 2

Nice tournament..Want to try the other new comers like crux,Larsen VB0.2,Noonian Chess,Freyer (is this compatiable with WB?) etc..Any else i left out?? Not Amy though..

[Dann Corbit]

Geschrieben von:/Posted by: Dann Corbit at 18 February 2000 18:53:01:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Colin Frayn at 18 February 2000 17:43:33:

Yes I think it is. You just wait and see how much I've just improved it by adding in one simple line and altering another. I _think_ I can prove that I can cut ouf up to half the tree with absolutely no drop in analysis accuracy. Wait for version 6.0 (which should be out before the end of the weekend)

Great! How strong do you anticipate this to be? If you tell me some programs you think it can contend with I will try to run a tournament.

Hmmm... well after significant testing it appears to perform slightly worse in WAC. Still not sure why.
If I can't get it's WAC score back up then I'll probably scrap this idea.

Could you explain your idea in detail and the math behind it? You have made me intensely curious. It might be the next giant leap in chess.
Cutting the tree in half will cause more than a linear speed up if you can achieve it.

My FTP site

[Colin Frayn]

Geschrieben von:/Posted by: Colin Frayn at 18 February 2000 20:07:03:
Als Antwort auf:/As an answer to: Re: A new LOW-MEDIUM tournament with Colchess an Raffaela geschrieben von:/posted by: Dann Corbit at 18 February 2000 18:53:01:

Hmmm... well after significant testing it appears to perform slightly worse in WAC. Still not sure why.
If I can't get it's WAC score back up then I'll probably scrap this idea.

Could you explain your idea in detail and the math behind it? You have made me intensely curious. It might be the next giant leap in chess.
Cutting the tree in half will cause more than a linear speed up if you can achieve it.

Hehehe yes it would be nice
The idea is just a simplification of futility pruning - nothing earth-shatteringly brilliant I'm afraid. It wasn't perfect either as I'd forgotten something rather irritating.
Anyway I've dumped that idea for now. ColChess v5.5 is now out and I did something rather incomprehensible in the quiescence search that I was hoping would improve it a little but in fact has got the WAC score up by 5, and that's taking into account the (slightly slow) corrections I had to make to fix some bugs in v5.4
I'll do an .exe soon.
Hopefully someone will do another tournament. ColChess 5.3 did OK in that low-mid tournament, but it should have done better. 5.5 should be quite a bit stronger than 5.3 (I hope).
Cheers,
Col

ColChess Homepage