Sungorus 1.2

[Tord Romstad]

Sungorus 1.2 is out.

And it's open source! I didn't know until now. It compiles right out of the box under Mac OS X. At first sight, the source code looks very readable and pleasantly terse. I miss the days when my own source code looked like that.

Thanks a lot for your program!

[Pablo Vazquez]

Sungorus 1.2 is out.

And it's open source! I didn't know until now. It compiles right out of the box under Mac OS X. At first sight, the source code looks very readable and pleasantly terse. I miss the days when my own source code looked like that.

Thanks a lot for your program!

Thanks for the compliment, Tord, and glad you liked it

[swami]

Sungorus 1.2

Strategic Test Suite Conditions:

Arena GUI
10 seconds each move
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 130 - 145 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
54/100, Grade: C

STS (v2.0) - Open Files and Diagonals:
43/100, Grade E

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
46/100, Grade: D

STS (v4.0) - Square Vacancy:
48/100, Grade: D

STS (v5.0) - Bishop vs Knight:
56/100, Grade: C+

Overall Performance:

Total Score: 247/500

Overall Average: 49.4%

Grade: D

[Pablo Vazquez]

Sungorus 1.2

Strategic Test Suite Conditions:

Arena GUI
10 seconds each move
500 positions
Core2Quad 32 bits, Q6600, 2 GB RAM, 2.4GHZ
Engine uses 130 - 145 Mb Hash.

Subject-wise Scores:

STS (v1.0) - Undermining:
54/100, Grade: C

STS (v2.0) - Open Files and Diagonals:
43/100, Grade E

STS (v3.0) - Knight Outposts/Centralization/Repositioning:
46/100, Grade: D

STS (v4.0) - Square Vacancy:
48/100, Grade: D

STS (v5.0) - Bishop vs Knight:
56/100, Grade: C+

Overall Performance:

Total Score: 247/500

Overall Average: 49.4%

Grade: D

Thanks Swami. Currently, the evaluation in sungorus sucks quite a bit, so I will concentrate exclusively on the evaluation for the next version.

[JVMerlino]

Hi Swami,

I use these tests a lot in determining improvements in my own engine, but your use of 10 seconds per position surprises me a bit, considering that the website where the suites are hosted says, for example, that for the undermining suite "It's advisable that you test your engine on 7 minutes per position".

For the other two tests, no specific time is recommended. So what I've been doing is three minutes per position for the undermining suite, and one minute per position on the outposts and files/diagonals suites.

jm

[swami]

Hi Swami,

I use these tests a lot in determining improvements in my own engine, but your use of 10 seconds per position surprises me a bit, considering that the website where the suites are hosted says, for example, that for the undermining suite "It's advisable that you test your engine on 7 minutes per position".

For the other two tests, no specific time is recommended. So what I've been doing is three minutes per position for the undermining suite, and one minute per position on the outposts and files/diagonals suites.

jm

Underming test is several months old, I should have modified that part about the time control. Anyway when you happen to download the latest test suite "Bishop vs Knight, You will see that the recommended time written there:

http://sites.google.com/site/strategict ... -vs-knight

Recommended Test time:

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


Therefore all the tests suites should be tested in either of these time controls. (just a ideal suggestion though, not must)

Also, Miguel did test his engine on 10 sec/position, 1 minute/position and 7 minutes position and posted his results.

Thanks for pointing that out, I'll now modify the time control part in older test suites to avoid confusion.

[JVMerlino]

Underming test is several months old, I should have modified that part about the time control. Anyway when you happen to download the latest test suite "Bishop vs Knight, You will see that the recommended time written there:

http://sites.google.com/site/strategict ... -vs-knight

Recommended Test time:

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


Therefore all the tests suites should be tested in either of these time controls. (just a ideal suggestion though, not must)

Also, Miguel did test his engine on 10 sec/position, 1 minute/position and 7 minutes position and posted his results.

Thanks for pointing that out, I'll now modify the time control part in older test suites to avoid confusion.

Got it. Thanks very much!

When I have my next official release, I'll post the results as requested above (although I may pass on the 7 minutes per position option as I use my work computer to run these tests overnight ).

jm