Dirty ACCA 3 Released

[Andres Valverde]

Hi all,

The original version of Dirty (EveAnn+Buzz) that played in the 2009 Third Annual World Computer Rapid Chess Championships is available at :

http://www.dirtychess.com

It has a new search, modified, debugged and tested by Fonzy Bluemers. Thanks to him and to Christopher Conkie for the tourney book.

Also the eval DLL is slightly modified.

Cheers, Pradu & Andres

[swami]

Dirty ACCA 3
by Pradu Kannan and Andres Valverde

Strategic Test Suite Conditions:

Arena GUI
10 seconds per position
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 150 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
62/100, Grade: B

STS (v2.0) - Open Files and Diagonals:
59/100, Grade: C+

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
61/100, Grade: B

STS (v4.0) - Square Vacancy:
65/100, Grade: B+

STS (v5.0) - Bishop vs Knight:
69/100, Grade: B+

Overall Performance:

Total Score: 316/500

Overall Average: 63.2 %

Grade: B

[Graham Banks]

Thanks Andres and Pradu.

[Andres Valverde]

Thanks Swami, testing at 1' just now, shall send you the results

BTW : What do the letters mean ?

[Andres Valverde]

Thanks Andres and Pradu.

You are welcome Graham

BTW : Is the match vs Romi finished?
BTW2 : If not BTW... can you use this new version?

[Andres Valverde]

Dirty ACCA 3
by Pradu Kannan and Andres Valverde

Strategic Test Suite Conditions:

Arena GUI
10 seconds per position
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 150 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
62/100, Grade: B

STS (v2.0) - Open Files and Diagonals:
59/100, Grade: C+

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
61/100, Grade: B

STS (v4.0) - Square Vacancy:
65/100, Grade: B+

STS (v5.0) - Bishop vs Knight:
69/100, Grade: B+

Overall Performance:

Total Score: 316/500

Overall Average: 63.2 %

Grade: B

I tested at 2' in my latptop single core 1.73 Mhz (relative speed to my Quad 2.4Mhz is a bit less than 50%), which may be grossly equivalent to 1'. 128 Mb hash, Arena 2.0.1

STS (v1.0) - Undermining: 78/100
STS (v2.0) - Open Files and Diagonals: 77/100
STS (v3.0) - Knight Outposts/Centralization/Repositioning: 73/100
STS (v4.0) - Square Vacancy: 81/100
STS (v5.0) - Bishop vs Knight: 82/100

[Christopher Conkie]

Dirty ACCA 3
by Pradu Kannan and Andres Valverde

Strategic Test Suite Conditions:

Arena GUI
10 seconds per position
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 150 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
62/100, Grade: B

STS (v2.0) - Open Files and Diagonals:
59/100, Grade: C+

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
61/100, Grade: B

STS (v4.0) - Square Vacancy:
65/100, Grade: B+

STS (v5.0) - Bishop vs Knight:
69/100, Grade: B+

Overall Performance:

Total Score: 316/500

Overall Average: 63.2 %

Grade: B

I tested at 2' in my latptop single core 1.73 Mhz (relative speed to my Quad 2.4Mhz is a bit less than 50%), which may be grossly equivalent to 1'. 128 Mb hash, Arena 2.0.1

STS (v1.0) - Undermining: 78/100
STS (v2.0) - Open Files and Diagonals: 77/100
STS (v3.0) - Knight Outposts/Centralization/Repositioning: 73/100
STS (v4.0) - Square Vacancy: 81/100
STS (v5.0) - Bishop vs Knight: 82/100

Are you suggesting that longer time controls are better?

I would have thought so

Christopher

[swami]

Thanks Swami, testing at 1' just now, shall send you the results

BTW : What do the letters mean ?

You're welcome, Andres. Don't know what letters you may be referring to. If It's grading letters. Then I have these:

Code: Select all

80+ S 
75+ A+
70+ A
65+ B+
60+ B
55+ C+
50+ C
45+ D
40+ E
<40 F

[swami]

Dirty ACCA 3
by Pradu Kannan and Andres Valverde

Strategic Test Suite Conditions:

Arena GUI
10 seconds per position
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 150 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
62/100, Grade: B

STS (v2.0) - Open Files and Diagonals:
59/100, Grade: C+

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
61/100, Grade: B

STS (v4.0) - Square Vacancy:
65/100, Grade: B+

STS (v5.0) - Bishop vs Knight:
69/100, Grade: B+

Overall Performance:

Total Score: 316/500

Overall Average: 63.2 %

Grade: B

I tested at 2' in my latptop single core 1.73 Mhz (relative speed to my Quad 2.4Mhz is a bit less than 50%), which may be grossly equivalent to 1'. 128 Mb hash, Arena 2.0.1

STS (v1.0) - Undermining: 78/100
STS (v2.0) - Open Files and Diagonals: 77/100
STS (v3.0) - Knight Outposts/Centralization/Repositioning: 73/100
STS (v4.0) - Square Vacancy: 81/100
STS (v5.0) - Bishop vs Knight: 82/100

These are pretty good results, Ofcourse I have noticed that more the time control, many engines seem to score little higher.

With more time(1 min and beyond) in the conditions, I'd have changed the grading systems anyway S got to be 90+...A+ 85+...A 80+ and so on... since I usually test bulk of engines on 10 seconds per position, then I'd stick with the current grading system.

Btw, I'm currently working on "Re-Capturing" test suite.

[swami]

STS (v1.0) - Undermining: 78/100
STS (v2.0) - Open Files and Diagonals: 77/100
STS (v3.0) - Knight Outposts/Centralization/Repositioning: 73/100
STS (v4.0) - Square Vacancy: 81/100
STS (v5.0) - Bishop vs Knight: 82/100

We can infer from the above results that the engine's strength seems to be in Square Vacancy and Bishop-Knight trade-offs. Pretty good strength actually. Knight outposts and centralization is the one that really needed improvement...ofcourse the knowledge of square vacancy is particularly more important than Knight outposts/Centralization.