Pharaon 3.0 beta : First impression.

[Johan Havegheer]

Geschrieben von:/Posted by: Johan Havegheer at 18 September 2004 16:58:14:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%
Best
Johan

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 18 September 2004 17:18:08:
Als Antwort auf:/In reply to: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Johan Havegheer at 18 September 2004 16:58:14:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%
Best
Johan

Where is 3.0 beta available?

[Johan Havegheer]

Geschrieben von:/Posted by: Johan Havegheer at 18 September 2004 17:21:43:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Norm Pollock at 18 September 2004 17:18:08:

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 September 2004 17:28:18:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Norm Pollock at 18 September 2004 17:18:08:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%
Best
Johan

Where is 3.0 beta available?

Hi Norm,
AEGT testers received Pharaon 3.0 beta from Franck Zibi for including it in Rook Class with AEGT 2.
I can confirm the first good impression Johan got, as I included Pharaon 3.00b in my Nunn Top 4 Blitz tournament. Results after 8 positions and 256 games are available in two hours at my web site. Rating improvement might be more than 50 points (guesses by Franck) at least in Blitz and so the new Pharaon will be top favourite in Rook Class.
Hope it will be released soon.
Best Regards
Heinz
http://www.husvankempen.de/nunn/

[Patrick Buchmann]

Geschrieben von:/Posted by: Patrick Buchmann at 18 September 2004 18:03:32:
Als Antwort auf:/In reply to: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Johan Havegheer at 18 September 2004 16:58:14:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%
Best
Johan

Hi,
I think Pharaon 3.00 beta is 50-70 points better as 2.62:
Gauntlet under Arena 1.09 and Fritz GUI (5')
Tournament in Division Senior (5'/40)
Regards,
Patrick


Le Fou numérique

[Gábor Szots]

Geschrieben von:/Posted by: Gábor Szots at 18 September 2004 18:47:32:
Als Antwort auf:/In reply to: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Johan Havegheer at 18 September 2004 16:58:14:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%

On a PIII 933 MHz machine with 48 MB hash (test under Arena, Pharaon installed as UCI) I got 2650 points on LCT2 which is in accordance with what you got.
However, LCT2 is a very unreliable test especially for very strong engines. Therefore I began the 100-position WM-test. Pharaon 3.00 solved 35, which is the same as Pharaon 2.62 did. Only, there are still some 30 positions left to test! I'm looking forward to the result of those.
Gábor

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 September 2004 18:57:22:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Heinz van Kempen at 18 September 2004 17:28:18:

Hi all ,
here are the interim standings from Nunn Top 4 after 8 positions. I gauntleted Pharaon 3.00b in on three computers Athlon 2600+ and 3000+. Games are 4 m. + 2 s. with Nunn positions.



1 Pro Deo 1.0               209.0 / 272 
2 SOS 4 for Arena           161.5 / 272 
3 Thinker 4.6c              150.5 / 272 
4 Pharaon 3.00b             150.0 / 272 
5 LG Revival 1.0.0.14       145.5 / 272 
6 AnMon 5.40                142.5 / 272 
7 Tao 5.7b4                 142.5 / 272 
8 Gothmog 1.0 beta 10       136.5 / 272 
9 Patriot 1.3.0             131.0 / 272 
10 Fruit 1.5                130.0 / 272 
11 SlowChess 2.94           126.5 / 272 
12 Crafty 19.15 BH          126.5 / 272 
13 Jonny 2.70               125.5 / 272 
14 Amyan 1.594              122.0 / 272 
15 The Baron 1.4.1b1        121.0 / 272 
16 Arasan 8.1               115.0 / 272 
17 Naum 1.2                 107.0 / 272 
18 The Baron 1.4.0          105.5 / 272 



Almost shocking to see Pharaon 3.00b on one level here with Thinker 4.6c. Of course 272 games are as good as nothing, we shall see if it will hold. First rating calculation after ten positions on Monday.
Nunn Top 4 will give 680 games to all new engines in it with all 20 Nunn positions till the end of the month. Some were already rated with a lot of games before, Naum also plays in Nunn Talents to have more connections between the groups.
All result tables for each positions and download of games available here:
http://www.husvankempen.de/nunn/
Best Regards
Heinz

[Pallav Nawani]

Geschrieben von:/Posted by: Pallav Nawani at 18 September 2004 19:42:30:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Gábor Szots at 18 September 2004 18:47:32:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%

On a PIII 933 MHz machine with 48 MB hash (test under Arena, Pharaon installed as UCI) I got 2650 points on LCT2 which is in accordance with what you got.
However, LCT2 is a very unreliable test especially for very strong engines. Therefore I began the 100-position WM-test. Pharaon 3.00 solved 35, which is the same as Pharaon 2.62 did. Only, there are still some 30 positions left to test! I'm looking forward to the result of those.
Gábor

Hello,
What are the LCT and WM tests?
Regards,
Pallav

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 18 September 2004 20:49:31:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Gábor Szots at 18 September 2004 18:47:32:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%

On a PIII 933 MHz machine with 48 MB hash (test under Arena, Pharaon installed as UCI) I got 2650 points on LCT2 which is in accordance with what you got.
However, LCT2 is a very unreliable test especially for very strong engines. Therefore I began the 100-position WM-test. Pharaon 3.00 solved 35, which is the same as Pharaon 2.62 did. Only, there are still some 30 positions left to test! I'm looking forward to the result of those.
Gábor

I do not think that the WM-test is more reliable.
I never use it in testing and I use only test suites like Arasan test suite or
ECM-GCP

I do not trust positional test suites because they are not balanced and they may include positions when you need to sacrifice in order to win but not positions when the target is not to sacrifice because sacrificing is losing(Fritz failed in not sacrificing material in the last WCCC and lost against Diep and you will never see positions like the position that Fritz sacrificed when the target is to avoid a sacrifice).
ECM-GCP probably also include too many sacrifices and Arasan is probably better test suite but I use also ECM-GCP when I test only search changes.
Uri

[Gábor Szots]

Geschrieben von:/Posted by: Gábor Szots at 18 September 2004 22:17:08:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Pallav Nawani at 18 September 2004 19:42:30:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%

On a PIII 933 MHz machine with 48 MB hash (test under Arena, Pharaon installed as UCI) I got 2650 points on LCT2 which is in accordance with what you got.
However, LCT2 is a very unreliable test especially for very strong engines. Therefore I began the 100-position WM-test. Pharaon 3.00 solved 35, which is the same as Pharaon 2.62 did. Only, there are still some 30 positions left to test! I'm looking forward to the result of those.
Gábor

Hello,
What are the LCT and WM tests?
Regards,
Pallav

Hello Pallav,
LCT2 is a test suite of 35 positions (10 mins/pos), WM-Test is another one consisting of 100 positions (20 mins/pos).
Where to find them?
LCT2 is included in the collection of Jon Dart, author of Arasan. URL is www.arasanchess.org.
WM-Test can be found on the CSS site, I think its URL is at the bottom of the WinBoard Forum page.
Regards,
Gábor

[Gábor Szots]

Geschrieben von:/Posted by: Gábor Szots at 18 September 2004 22:24:25:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Uri Blass at 18 September 2004 20:49:31:

Hello,
I've just did a little test with Pharaon 3.0 Beta and it seems very promising !
LCT II tests : AMD XP2800+ 128Mb Hash.
Pharaon 2.62 : 2620 LCT Elo , POS = 56% , CMB = 81% , FIN = 72%
Pharaon 3.0 beta : 2695 LCT Elo, POS = 69% , CMB = 85% , FIN = 74%

On a PIII 933 MHz machine with 48 MB hash (test under Arena, Pharaon installed as UCI) I got 2650 points on LCT2 which is in accordance with what you got.
However, LCT2 is a very unreliable test especially for very strong engines. Therefore I began the 100-position WM-test. Pharaon 3.00 solved 35, which is the same as Pharaon 2.62 did. Only, there are still some 30 positions left to test! I'm looking forward to the result of those.
Gábor

I do not think that the WM-test is more reliable.
I never use it in testing and I use only test suites like Arasan test suite or
ECM-GCP

I do not trust positional test suites because they are not balanced and they may include positions when you need to sacrifice in order to win but not positions when the target is not to sacrifice because sacrificing is losing(Fritz failed in not sacrificing material in the last WCCC and lost against Diep and you will never see positions like the position that Fritz sacrificed when the target is to avoid a sacrifice).
ECM-GCP probably also include too many sacrifices and Arasan is probably better test suite but I use also ECM-GCP when I test only search changes.
Uri

In fact I don't think there are any reliable test suites out there. I prefer WM-Test because at least it was carefully selected and 100 positions is more than 35.
Test suites have the advantage of giving you an approximate first estimation within a couple of hours (or days), while you have to let your engine play for weeks to acquire the same estimation of strength, assuming that blitz games won't tell much.
Gábor

[Michael Scheidl]

Geschrieben von:/Posted by: Michael Scheidl at 19 September 2004 17:11:25:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Gábor Szots at 18 September 2004 22:17:08:

(...)

LCT2 is a test suite of 35 positions (10 mins/pos), WM-Test is another one consisting of 100 positions (20 mins/pos).
Where to find them?
LCT2 is included in the collection of Jon Dart, author of Arasan. URL is www.arasanchess.org.

The LCT2 is Loguet II-Test, not Lapuce 2 as it's called in the Arasan collection. I didn't check if it's only the name which is given wrong, or if these are different positions. But the "L" definetely means (F.) Louguet, of Chess Wizard, who has designed the LCT II-Test many years ago.
http://perso.wanadoo.fr/lefouduroi/test_lct_2.htm
WM-Test download (currently with 260 Engineresults):
http://www.computerschach.de/test/
Regards,
Mike Scheidl

[Michael Scheidl]

Geschrieben von:/Posted by: Michael Scheidl at 19 September 2004 19:57:09:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Michael Scheidl at 19 September 2004 17:11:25:

(...)

The LCT2 is Loguet II-Test, not Lapuce 2 as it's called in the Arasan collection. I didn't check if it's only the name which is given wrong, or if these are different positions.

Everything ok: It's an alternative name for the same test suite which was published by the magazine "La Puce Echiquenné."
Sorry, and thx Jon.
Regards,
M.Scheidl

[Pallav Nawani]

Geschrieben von:/Posted by: Pallav Nawani at 19 September 2004 21:10:40:
Als Antwort auf:/In reply to: Re: Pharaon 3.0 beta : First impression. geschrieben von:/posted by: Michael Scheidl at 19 September 2004 19:57:09:

(...)

The LCT2 is Loguet II-Test, not Lapuce 2 as it's called in the Arasan collection. I didn't check if it's only the name which is given wrong, or if these are different positions.

Everything ok: It's an alternative name for the same test suite which was published by the magazine "La Puce Echiquenné."
Sorry, and thx Jon.
Regards,
M.Scheidl

Thanks to both of you! (N/T)