Engine Score Me
01: Mediocre 200,5/400 ····················
02: NanoSzachy 18,0/20 11111==1111011111111
03: Hamsters 17,5/20 11111111=11011111110
04: Diablo 17,0/20 11101101111011111111
05: Hermann 16,5/20 1111111101100111111=
06: LittleThought 15,5/20 1=100111111111111001
06: Gaia 15,5/20 11110110=11011111011
08: Counter 13,5/20 1=111==1001=10=10111
09: AliUCI 13,0/20 11110000111101110110
10: Bison 11,0/20 0=010=10111001100111
11: Feuerstein 10,0/20 =01101010011=0101=0=
12: Amundsen 9,0/20 11=001000010100101=1
12: GreKo 9,0/20 10100110001001111000
14: Gibbon 8,5/20 01010001=01011010010
14: Clarabit 8,5/20 000111000=111001=00=
16: BBChess 4,5/20 001010=0011000000000
17: Bikjump 4,0/20 11000000001000000010
18: Lime 3,0/20 10000000=01000000=00
19: FluxII 2,0/20 00000100001000000000
19: Roce 2,0/20 00000000001000000010
21: Vicki 1,5/20 0000000000100000=000
I calculated the estimated ratings for this with the bayesian elo tool and the difference came to 48 rating points. Which of course is quite huge.
Anyway I am going to assume that strength difference is linear so that an improvement in one will result in an equal improvement in the other, that way I don't have to recompile all the time. :)
1 comment:
I have created a simple benchmark in Mediocre by adding to the end of entry method:
Board board = new Board();
board.setupStart();
long be = System.currentTimeMillis();
Engine.search(board, 11, false, 0, 0, 0, false);
System.out.println("Done in: " + (System.currentTimeMillis() - be) + " millis");
System.exit(0);
PC: AMD Athlon 64 X2 Dual-Core 3600+, 1024 mb of RAM
OC: Ubuntu 8.04
Standard engine parameters. Starting position.
I found the results very interesting...
Natively compiled with GCJ from bytecode: Done in 16813 millis. NPS ~ 33 677
Sun HotSpot JVM: Done in: 8679 millis. NPS ~ 65 239
Cacao JVM: Done in: 3940 millis. NPS ~ 143 709 (!)
So, as you can see, your engine do very well under cacao jvm in Linux. Don't know about Windows though.
Post a Comment