AH_Endgames-250 and AH_Tactics-250 Results

[Aser Huerga]

Have You checked 6 piece endings from Nalimov database? At least this has 2 equal solutions

[D]2k5/8/rn6/6R1/7R/1K6/8/8 w - - 0 1

Rc5+ and Rh8+ are both mate in 30.

Jouni

Results are updated, thanks again Jouni.

[Aser Huerga]

------------ AH_Endgames-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Hiarcs 13.2                242/250                2777         93.83

------------ AH_Tactics-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Hiarcs 13.2                239/250                3358         92.54

Endgames: http://ah-chesstestsuites.blogspot.com/ ... s-250.html
Tactics: http://ah-chesstestsuites.blogspot.com/ ... s-250.html

(Last entry in bold)

[Aser Huerga]

------------ AH_Endgames-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Shredder 12 32-bit         228/250                5172         88.51

------------ AH_Tactics-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Shredder 12 32-bit         217/250                8470         81.18

Endgames: http://ah-chesstestsuites.blogspot.com/ ... s-250.html
Tactics: http://ah-chesstestsuites.blogspot.com/ ... s-250.html

(Last entry in bold)

[MM]

I'm going to post here test results for AH_Endgames-250 (http://www.mediafire.com/?m6u7w3g4r9m1cu4) and AH_Tactics-250 (http://www.mediafire.com/?0xfu5bh4zcwvoou) EPD collections.
(The links above contains corrected EPDs)

Hardware: i5-2410M 2.30Ghz Laptop
Operating system: Microsoft Windows 7 64 Bit
All engines 1 CPU (SSE versions when available)
Arena 2.0.1 GUI
128MB Hash Tables
3-4-5 EGTBs
180 seconds for each position

I will post individual entries and links to the updated lists with all engines tested.

Each entry shows engine name, seconds to solve the EPD, and the corresponding percentage (simply seconds referred to 100).

Hi,

thank you very much.

Very interesting tests.

Best regards

[Jouni]

Previously Shredder was endgame expert

Jouni

[rodolfoleoni]

I tried the suites with The Baron 3.28a, at 1 minute per position, without TBs/BBs. Hardware was an i7 2630 QM, hyperthreaded 8 cores. Baron got the following results:

Tactics: 238/250 in 1638 seconds (95,2%)
Endgame: 219/250 in 2416 seconds (87,6%)

I don't know how hardware difference does impact, or if giving only 1 min/position compensates. I also notice hyperthreading works fine with The Baron as 8 cores is always 30% faster than 4 cores.

[MM]

Previously Shredder was endgame expert

Jouni

Hi,

if im not wrong the latest release of shredder (12) is dated summer 2009, an eternity...and the next is not exactly coming soon...

Regards

[Aser Huerga]

I would like a new strong Shredder version so much ... when I brought the 12 version I have 32-bit OS yet.
I have use bitbases for the test, but as Maurizio says, 2009 is so before ....

I tried the suites with The Baron 3.28a, at 1 minute per position, without TBs/BBs. Hardware was an i7 2630 QM, hyperthreaded 8 cores. Baron got the following results:

Tactics: 238/250 in 1638 seconds (95,2%)
Endgame: 219/250 in 2416 seconds (87,6%)

I don't know how hardware difference does impact, or if giving only 1 min/position compensates. I also notice hyperthreading works fine with The Baron as 8 cores is always 30% faster than 4 cores.

BTW, Rodolfo, my % is related with seconds (from 0 [all positions solved in short than 0 seconds] to 45,000 [no position solved]) and not with number of positions solved.


I will be glad to add results from my hardware and conditions (for homogeneity) of any engine in actual development, if someone ask for.

Cheers.

[rodolfoleoni]

I found your test so interesting that I couldn't resist...

As a Baron tester, I found useful to measure how much pruning does impact on finding the only one move per position. I think I got some interesting answers.

Thanks for the links!

[Aser Huerga]

------------ AH_Endgames-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Hannibal 1.1 64-bit        228/250                6312         85.97

------------ AH_Tactics-250:

Code: Select all

ENGINE                   POS. SOLVED             SECONDS         %    
Hannibal 1.1 64-bit        237/250                4825         89.28

Endgames: http://ah-chesstestsuites.blogspot.com/ ... s-250.html
Tactics: http://ah-chesstestsuites.blogspot.com/ ... s-250.html

(Last entry in bold)