STS (v8.0) - "Advancement of f/g/h Pawns"

[swami]

Chapter 8 of the Strategic Test Suite: Advancement of f/g/h pawns is now available for download.

* Consists of carefully selected 100 questionairres on "Advancement of f/g/h Pawns"

* All of the questions in this test suite are selected by me and thoroughly verified by Dann Corbit through analysis with the help of top engines for hours. I initially selected about 200 problems and sent them to Dann, of which 100 has passed criteria. Dann then made the final selection.

* All the answers carry similar traits:

(1) Advancing the king side pawns in an attempt to gain more space/launch attack.
(2) Marching forward with either of the f/g/h pawns in an attempt to prevent opponent from advancing his pawns further. Helps in building the strong active defense.
(3) Tests your engine's understanding on King safety, Handling of King's side pawns, Gaining space on the edge of the board, Aggressiveness.

Download it! here:
http://sites.google.com/site/strategict ... f-g-h-pawn

Thanks to those who reported bugs and tested out the v1.0 to v7.0

Test Suite release time: 23rd of December, 2009
Swaminathan and Dann Corbit

[swami]

Alternate download links for Harun, Bob and those who had experienced problems downloading it from google site.

http://www.mediafire.com/?qyyywz3e4on

It's from my limited tests the very toughest one. I believe it will be the toughest ever.

Meanwhile the testimonials page is now updated with 2 new entries. Thanks guys!

http://sites.google.com/site/strategict ... stimonials

Feel free to report results from your engines/favorite engines. Suggestions are welcome.

Code: Select all

Recommended Test time: 

Short time control:                                      10 seconds each move. 
Intermediate Time Control:                               1 minute each move 
Long Time Control:                                       7 minutes each move. 

[Ferdy]

Alternate download links for Harun, Bob and those who had experienced problems downloading it from google site.

http://www.mediafire.com/?qyyywz3e4on

It's from my limited tests the very toughest one. I believe it will be the toughest ever.

Meanwhile the testimonials page is now updated with 2 new entries. Thanks guys!

http://sites.google.com/site/strategict ... stimonials

Feel free to report results from your engines/favorite engines. Suggestions are welcome.

Code: Select all

Recommended Test time: 

Short time control:                                      10 seconds each move. 
Intermediate Time Control:                               1 minute each move 
Long Time Control:                                       7 minutes each move. 

Deuterium v09.01.26.492 result:

Hardware: Pentium(R) Dual-Core CPU, E5300, 2.6Ghz, 4.00 GB RAM
OS: Windows Vista Home Basic SP 2, 32-bit

Engine setting: 1 cpu, 64 MB Main Hash, 4MB Pawn Hash, 16MB Eval Cache, egbb is OFF
Test Condition: 10 sec per position
Software: epd2wb

Code: Select all

<=Sec Solved Total  PosNr
----- ------ ------ -----------------------------------------------------------
    1     35     35   3  13  15  16  18  23  26  29  30  35  39  46  47  49  51
                     55  61  66  67  69  72  73  74  75  79  82  84  85  86  89
                     91  96  97  98 100
    2      1     36   4
    3      3     39   7  44  53
    4      1     40  37
    5      1     41  27
    6      1     42  81
    7      1     43  64
    9      1     44  58
   10      1     45  54


Failure               1   2   5   6   8   9  10  11  12  14  17  19  20  21  22
                     24  25  28  31  32  33  34  36  38  40  41  42  43  45  48
                     50  52  56  57  59  60  62  63  65  68  70  71  76  77  78
                     80  83  87  88  90  92  93  94  95  99

45 problems solved.
55 problems unsolved. 

This is indeed a difficult test for Deuterium, I heavily penalized moving king's pawn cover in this version. But this is one of the most important theme that an engine should know.

It would be great also if you can include typical openings that these positions will arise.

[beachknight]

Alternate download links for Harun, Bob and those who had experienced problems downloading it from google site.

http://www.mediafire.com/?qyyywz3e4on

It's from my limited tests the very toughest one. I believe it will be the toughest ever.

Meanwhile the testimonials page is now updated with 2 new entries. Thanks guys!

http://sites.google.com/site/strategict ... stimonials

Feel free to report results from your engines/favorite engines. Suggestions are welcome.

Code: Select all

Recommended Test time: 

Short time control:                                      10 seconds each move. 
Intermediate Time Control:                               1 minute each move 
Long Time Control:                                       7 minutes each move. 

Thanks, Swami. Just downloaded from the mf link.
I'll take a deeper look at this test suit, but later

Best,

[Graham Banks]

Chapter 8 of the Strategic Test Suite: Advancement of f/g/h pawns is now available for download.

If these all involve the advancement of pawns upon the enemy king, perhaps that would be a more appropriate name for it.

Cheers,
Graham.

[swami]

Chapter 8 of the Strategic Test Suite: Advancement of f/g/h pawns is now available for download.

If these all involve the advancement of pawns upon the enemy king, perhaps that would be a more appropriate name for it.

Cheers,
Graham.

Yes, It was suggested by Dann and I accepted his suggestion. I had initially called it "Advancement of King Side Pawn Cover" but Dann showed the position number 99 where the king is castled long and that pawn "cover" wouldn't be more appropriate in this case. So, we settled it on f/g/h pawns instead.

[swami]

Deuterium v09.01.26.492 result:

Hardware: Pentium(R) Dual-Core CPU, E5300, 2.6Ghz, 4.00 GB RAM
OS: Windows Vista Home Basic SP 2, 32-bit

Engine setting: 1 cpu, 64 MB Main Hash, 4MB Pawn Hash, 16MB Eval Cache, egbb is OFF
Test Condition: 10 sec per position
Software: epd2wb

Code: Select all

<=Sec Solved Total  PosNr
----- ------ ------ -----------------------------------------------------------
    1     35     35   3  13  15  16  18  23  26  29  30  35  39  46  47  49  51
                     55  61  66  67  69  72  73  74  75  79  82  84  85  86  89
                     91  96  97  98 100
    2      1     36   4
    3      3     39   7  44  53
    4      1     40  37
    5      1     41  27
    6      1     42  81
    7      1     43  64
    9      1     44  58
   10      1     45  54


Failure               1   2   5   6   8   9  10  11  12  14  17  19  20  21  22
                     24  25  28  31  32  33  34  36  38  40  41  42  43  45  48
                     50  52  56  57  59  60  62  63  65  68  70  71  76  77  78
                     80  83  87  88  90  92  93  94  95  99

45 problems solved.
55 problems unsolved. 

This is indeed a difficult test for Deuterium, I heavily penalized moving king's pawn cover in this version. But this is one of the most important theme that an engine should know.

It would be great also if you can include typical openings that these positions will arise.

Thanks for running the test. With more time, I believe Deuterium could score >55's.

Doch 1.2 could only manage 64/100 in 10 seconds per move. With more time I believe it could manage 75's.

This is certainly toughest, and I don't believe anything in future will be more tougher than this.

[swami]

I heavily penalized moving king's pawn cover in this version. But this is one of the most important theme that an engine should know.

Advancing the king side pawns is done only in cases

* where the opponent's position is bit passive on the other end directly opposite.

* And number of pawns are fixed in the center.

* where pawn advance acts as an active defense weapon against pawn storms on the opposing end.

It would be great also if you can include typical openings that these positions will arise.

Usually in King's Indian Defense. Sicilian. Slav Accepted, Catalan. It can happen in most semi closed positions openings.

[Graham Banks]

Chapter 8 of the Strategic Test Suite: Advancement of f/g/h pawns is now available for download.

If these all involve the advancement of pawns upon the enemy king, perhaps that would be a more appropriate name for it.

Cheers,
Graham.

Yes, It was suggested by Dann and I accepted his suggestion. I had initially called it "Advancement of King Side Pawn Cover" but Dann showed the position number 99 where the king is castled long and that pawn "cover" wouldn't be more appropriate in this case. So, we settled it on f/g/h pawns instead.

Thanks for the clarification. Keep up the good work.

[noctiferus]

Only 32bit STS short time control:

Swaminathan Sts :100 test/ set http://sites.google.com/site/strategictestsuite/
1cpu 32 bit 10 second/move 2 Ghz – Shredder GUI 12 –hash 64 Mb

Performance over 100 x 8 tests :
Sts
g3 e4 YY Ryb3
1 95 96 95 92
2 88 88 89 84
3 86 86 85 84
4 80 80 81 87
5 84 84 87 87
6 85 85 86 84
7 85 85 84 82
8 74 73 73 71

Total out of 800
677 677 680 671
percent
.846 .846 .85 .839

PS: YY is another engine that "could" be suspect... No further wars, please.