Crab 1.0 Beta (32-Bit)

[look]

Crab 1.0 seems to be the basis for a very strong program... but by now, the results are a bit surprising:

Quad 2,33 $M Ram w32
Gui Fritz 12
Time control 2 + 2 seconds.
Book: HS Masterbook 2.0
Ponder: Off
DamirRybkaKiller42 using one core. All other engines using four cores.

The authors of Crab 1.0 are top class. We can expect a powerful further development in the future if they decide continue with their project.

2010-06Crab 2010

Crab 1.0 beta - DamirsRybkaKiller42_w32 11.0 - 9.0 +5/=12/-3 55.00%
Crab 1.0 beta - Deep Rybka 4 w32 10.0 - 10.0 +6/=8/-6 50.00%
Crab 1.0 beta - Houdini 1.01 w32 4_CPU 9.5 - 10.5 +4/=11/-5 47.50%
Crab 1.0 beta - Naum 4.2 (x4) 6.5 - 13.5 +2/=9/-9 32.50%

Crab 1.0 is able to get 50% against DR4... but suffers a big lose against Naum 4.2!. DamirsRK42 gets again a very good result: 45% -handicap 1 to 4 cores-.

Regards,

Tom.

Hi Tom

Please dont forget that Crab is based on stockfish with a few changes here and there. So I think you could consider Crab 1.0 as very probably weaker but with a different style of play from SF 1.7.1.

[alfons]

the source did compile with warnings (unused variable ‘Them’) under Linux (Debian Sid, 64 Bit; g++ :4.4.4). the author is asking for cpu-time (in ccc) but leaves no message of his virtual address as it seems that a web-space/simple readme for the engine is missing too . pity.


but interesting thing nevertheless (perhaps the improvements will get merged into maintree?!). will test soon against some ivanhoe...but football comes first (vuvuzela ftw #$!%).

seid bereit!

[look]

the source did compile with warnings (unused variable ‘Them’) under Linux (Debian Sid, 64 Bit; g++ :4.4.4). the author is asking for cpu-time (in ccc) but leaves no message of his virtual address as it seems that a web-space/simple readme for the engine is missing too . pity.


but interesting thing nevertheless (perhaps the improvements will get merged into maintree?!). will test soon against some ivanhoe...but football comes first (vuvuzela ftw #$!%).

seid bereit!

Greetings

If you are interested in testing a beta version, I can upload the source or write differences from stockfish 1.7.1 so you can change the code and compile it. Note that test may take 2 days and it is more likely that it would be weaker than SF.

[alfons]

hi,

you're the author of grab, look?

a *.diff will suffice/but providing patched source is more comfort. are you interested in matches (opponent/time control?) or solving testpositions*?
*which ones?

two days are very ok (I pm you my system specs).

[thorstenczub]

new crab results:

Motor                     Punkte    Cr   De   St   Ry   Ti   Fi   Ho   Na   Ko   De   Ho   Iv   Hi   Za   De   No   Sp   Gu   Ha   Pr    S-B
01: Crab-x64_PGO [t1]         52,0/73 ···· =01= ==0= 01== ==== ==0= 0=00 =1== 111= 1=01 11=  111  1111 10=1 =111 11=1 1111 1111 111  1111  1580,0
02: Deep Rybka 4 x64 [t1]     51,5/73 =10= ···· ==1= 1==0 11=0 ==00 =11= 1=11 ==== ==01 ===  ===  =111 11=1 1=11 1111 111= =111 11=  1111  1619,2
03: Stockfish-18-64-ja [t1]   51,0/73 ==1= ==0= ···· 1=10 =1== ==1= ===0 =111 1=1= 1=01 10=  00=  1111 1=11 1111 0==1 1=11 =111 1=1  1111  1631,0
04: Rybka 3_64 [t1]           49,5/72 10== 0==1 0=01 ···· 1011 ==1= 0=11 ==1= ==11 =011 0==  =0   =111 11=1 =11= =111 1101 1111 11=  11=1  1578,5
05: Tinapa1.01 [t1]           48,0/73 ==== 00=1 =0== 0100 ···· ==1= 010= 00== =111 1101 1=0  001  ==== 1111 1111 1111 11=1 1=11 111  1111  1441,5
06: Fire_12_x64 [t1]          47,5/73 ==1= ==11 ==0= ==0= ==0= ···· ==1= =1== =0== 011= ===  ===  1=11 ==1= ===1 1==1 1111 =11= 111  1111  1499,5
07: Houdini102_x64_1CPU       47,0/73 1=11 =00= ===1 1=00 101= ==0= ···· 10=0 ===0 =110 ===  1==  11=1 1==0 1=11 1=11 1111 ==11 1==  1111  1482,7
08: Naum4.2_64 [t1]           41,5/72 =0== 0=00 =000 ==0= 11== =0== 01=1 ···· 1=== =110 1=0  0=0  10=1 101= 1=11 11== =11= 1111 1=   1=11  1275,7
09: Komodo-12-64-ja           40,0/72 000= ==== 0=0= ==00 =000 =1== ===1 0=== ···· =1== =01  ==   1==1 1=== ==11 111= 11=1 0==1 =11  111=  1212,2
10: DeepShredder12UCIx64 [t1] 37,5/72 0=10 ==10 0=10 =100 0010 100= =001 =001 =0== ···· 0=   =0=  1=01 =111 11=1 1=1= 01== 111= 01=  ==11  1165,5
11: Houdini103_x64_1CPU       36,5/56 00=  ===  01=  1==  0=1  ===  ===  0=1  =10  1=   ···· 1=1  =1=  111  =10  11=  111  111  11=  101   1155,2
12: IvanHoe53_64 [t1]         36,0/55 000  ===  11=  =1   110  ===  0==  1=1  ==   =1=  0=0  ···· 11=  ==1  1==  111  111  111  10=  111   1130,5
13: Hiarcs12MPUCI [t1]        29,5/73 0000 =000 0000 =000 ==== 0=00 00=0 01=0 0==0 0=10 =0=  00=  ···· =00= 11=1 1001 =1== 1111 111  1111  764,50
14: Zappa Mexico II x64 [t1]  27,5/72 01=0 00=0 0=00 00=0 0000 ==0= 0==1 010= 0=== =000 000  ==0  =11= ···· 0==1 11== =1=1 1000 11   =1==  823,75
15: Deep-onno-1-2-70-x64 [t1] 24,0/73 =000 0=00 0000 =00= 0000 ===0 0=00 0=00 ==00 00=0 =01  0==  00=0 1==0 ···· =1=0 1=1= 1==1 101  ==11  647,50
16: Now2t                     19,5/73 00=0 0000 1==0 =000 0000 0==0 0=00 00== 000= 0=0= 00=  000  0110 00== =0=1 ···· ==== =110 ==0  01=0  588,00
17: Spike 1.2                 18,5/73 0000 000= 0=00 0010 00=0 0000 0000 =00= 00=0 10== 000  000  =0== =0=0 0=0= ==== ···· 0110 1=1  01=1  497,50
18: Gull 09 x64               17,0/73 0000 =000 =000 0000 0=00 =00= ==00 0000 1==0 000= 000  000  0000 0111 0==0 =001 1001 ···· 0==  1=01  486,25
19: Hannibal1.0ax64           15,0/55 000  00=  0=0  00=  000  000  0==  0=   =00  10=  00=  01=  000  00   010  ==1  0=0  1==  ···· 1=1   428,50
20: ProDeo 1.5                12,0/73 0000 0000 0000 00=0 0000 0000 0000 0=00 000= ==00 010  000  0000 =0== ==00 10=1 10=0 0=10 0=0  ····  314,25

701 Partien von 760 gespielt
Name des Turniers: Arena Turnier
Ort/ Land: ORION8-LATITUDE, Deutschland
Spielstufe: Turnier 40/20
Hardware: Intel(R) Core(TM)2 CPU         T7600  @ 2.33GHz  mit 2.048 MB Speicher
Betriebssystem: Microsoft Windows Vista Professional (Build 6000)
PGN-Datei: C:\schach\arena201\Arena.pgn
Internetseite: 
E-Mail Adresse:

[Hagen]

Has anybody else been able to come up with these numbers?! With Crab?! I find this hard to believe. Crab...better than Houdini 1.03a or even Stockfish 1.8? I'm not taking this seriously. At 40/20?!

Looks like I'll need to take another look at Crab...but I'll use the Shredder Classic interface this time and not SCID.

[DaveD]

new crab results:

Motor                     Punkte    Cr   De   St   Ry   Ti   Fi   Ho   Na   Ko   De   Ho   Iv   Hi   Za   De   No   Sp   Gu   Ha   Pr    S-B
01: Crab-x64_PGO [t1]         52,0/73 ···· =01= ==0= 01== ==== ==0= 0=00 =1== 111= 1=01 11=  111  1111 10=1 =111 11=1 1111 1111 111  1111  1580,0
07: Houdini102_x64_1CPU       47,0/73 1=11 =00= ===1 1=00 101= ==0= ···· 10=0 ===0 =110 ===  1==  11=1 1==0 1=11 1=11 1111 ==11 1==  1111  1482,7
11: Houdini103_x64_1CPU       36,5/56 00=  ===  01=  1==  0=1  ===  ===  0=1  =10  1=   ···· 1=1  =1=  111  =10  11=  111  111  11=  101   1155,2
pre]

These results look strange. There must be something wrong with your settings.

[thorstenczub]

hi dave, what do you believe is wrong?

tournament is not finished. late entry engines have to play some more games...

[BTO7]

What looks a little strange to me is ..if this tournament was at 40/20 that many games looks like it would take a month to play day and night. In Houdini's case 1.03a I dont think has even been out two weeks yet.

BT

[thorstenczub]

Yes. the tournament is old. maybe you don't understand the concept of late-entry engines.

i can add at any time any engine to an old tournament.
concerning houdini, i am not playing with 1.03a but with 1.03. that is a little older i guess than the 1.03a bugfix.