Critter 1.2 update + new homepage

[Don]

Hello Richard

thanks a lot for thin new version.
It is verry nice that you are still developing this great engine!
Do you have any "idea" if it is stronger than Critter 1.01?
Any results?

best regards, Clemens

It is way stronger!

Well, one cannot ignore the preliminary results by IPON, so possibly there is some undiscovered bug holding it back in matches?

I noticed that too.

The difference in strength as super fast time controls is almost ridiculous and that is what I'm basing my estimate on. But Critter seems to have a scaling issue that is more extreme than most programs so perhaps we are seeing that here. I guess we will know soon enough.

[abik]

To get a rough idea of the strength of this new engine, I conducted a few quick matches with other engines using a random book opening play, one second-per-move, 32MB hash, and 8MB tablebases cache with the 3- and 4-piece Nalimov, Gaviota, Scorpio, Robbobases endgames on SD card. More games upcoming. To see Critter in action, see these screenshots.

Code: Select all

Critter 1.2
 - RobboLito 0.085e4l     3.5 - 6.5  +2/-5/=3   35.00%
 - Komodo32 2.03 AB       5.0 - 5.0  +4/-4/=2   50.00%
 - Stockfish 2.1          6.0 - 4.0  +4/-2/=4   60.00%
 - GNU Chess 5.07.170.7b  7.0 - 3.0  +7/-3/=0   70.00%
 - Gaviota v0.83          8.0 - 2.0  +7/-1/=2   80.00%
 - Rotor 0.6a             9.5 - 0.5  +9/-0/=1   95.00%
 - Toga II 1.4.1SE        9.5 - 0.5  +9/-0/=1   95.00%
 - RedQueen 0.9.8        10.0 - 0.0 +10/-0/=0  100.00%
 - BikJump v2.1P         10.0 - 0.0 +10/-0/=0  100.00%
 - Chess for Android     10.0 - 0.0 +10/-0/=0  100.00%

[abik]

Mr. Bik, is there any chance we will get a "Chess for iPhone" application in the future? I really miss something like your Chess GUI on the iPhone.

Although I appreciate your kind request, I have no iOS plans

[Damir]

Thanks a lot for this new engine Richard. Please keep up the good work. There are very few of original engines out there and yours is one of them.

[mclane]

Hello Richard

thanks a lot for thin new version.
It is verry nice that you are still developing this great engine!
Do you have any "idea" if it is stronger than Critter 1.01?
Any results?

best regards, Clemens

It is way stronger!

i have also very strong results from critter 1.2 so far:

35: Critter_1.2_64bit [t1] 40,5/49 so far.

this is 40/5 time control.

al results so far:
http://schachcomputerwelt.foren-city.de ... .html#2045

[Laskos]

Hello Richard

thanks a lot for thin new version.
It is verry nice that you are still developing this great engine!
Do you have any "idea" if it is stronger than Critter 1.01?
Any results?

best regards, Clemens

It is way stronger!

Well, one cannot ignore the preliminary results by IPON, so possibly there is some undiscovered bug holding it back in matches?

I noticed that too.

The difference in strength as super fast time controls is almost ridiculous and that is what I'm basing my estimate on. But Critter seems to have a scaling issue that is more extreme than most programs so perhaps we are seeing that here. I guess we will know soon enough.

Here is an example of almost totally useless ultra-fast TC test, 1.5s + 0.03s (games in ~6 seconds)

Code: Select all

Games Completed = 780 of 20000 (Avg game length = 5.912 sec)
Settings = Gauntlet/1MB/1500ms+30ms/M 10000cp for 1000 moves, D 200 moves/PGN:K:\Seagate Backup\LASTXP\H\Downloads\LittleBlitzer-2.51\swcr.pgn(3395)
Time = 5127 sec elapsed, 126341 sec remaining


 1.  Critter 1.2 64-bit       	464.0/780 357-209-214  (L: m=209 t=0 i=0 a=0)	(D: r=121 i=43 f=39 s=10 a=1)  (tpm=45.8 d=11.5 nps=1271134)

 2.  Komodo64 2.03 JA         	45.5/130	30-69-31  	(L: m=69 t=0 i=0 a=0)	(D: r=13 i=4 f=9 s=5 a=0)  (tpm=44.8 d=8.1 nps=685212)
 3.  Stockfish 2.1 JA 64bit   	22.5/130	12-97-21  	(L: m=97 t=0 i=0 a=0)	(D: r=11 i=5 f=5 s=0 a=0)  (tpm=26.1 d=8.3 nps=832725)
 4.  Houdini 1.5a x64         	76.0/130	59-37-34  	(L: m=37 t=0 i=0 a=0)	(D: r=16 i=8 f=9 s=1 a=0)  (tpm=46.3 d=9.8 nps=1275922)
 5.  Deep Rybka 4.1 x64"      	55.0/130	31-51-48  	(L: m=51 t=0 i=0 a=0)	(D: r=27 i=18 f=1 s=2 a=0)  (tpm=47.9 d=6.8 nps=45103)
 6.  Critter 1.01 64-bit      	34.5/130	13-74-43  	(L: m=74 t=0 i=0 a=0)	(D: r=30 i=3 f=8 s=1 a=1)  (tpm=38.0 d=10.7 nps=1038192)
 7.  Ivanhoe B47cBx64-1       	82.5/130	64-29-37  	(L: m=29 t=0 i=0 a=0)	(D: r=24 i=5 f=7 s=1 a=0)  (tpm=47.1 d=10.2 nps=987274)

All is apparently fine, no time losses and such, NPS is correct, but the time used per move is screwed up, and the results are weird. All one can say is that Critter 1.2 is probably close to the bunch of second place engines.

Kai

[rvida]

A type of finish that is muted Critter 1.2 is king, bishop and knight against king (KBNk)
Thanks

Hello Richard

thanks a lot for thin new version.

I noticed a problem with endings.
I have used the engine to analyze the finals, but this time he remains silent.
Do I have the gabiota gtb?

Thanks

Thank You for the bug report.
Indeed it seems that just 2 days before release I managed to broke KBN vs K evaluation
The bug went unnoticed because I mostly test with tablebase probing enabled at root (and all 4 piece endings are adjudicated by the GUI anyway).

Code: Select all

position fen 4k3/8/8/8/8/3KN3/3B4/8 w - - 2 1
eval
eval = +M-246697
evaluated by endgame knowledge

In this position when analysing with tablebases disabled Critter hangs at iteration 1 forever.

Now I am not sure what to do. While it is very easy to fix, I think this bug is not critical enough to warrant a product recall and re-release. If no other or more serious bugs are found, I would like to stick with the usual 3-4 month release cycle.

Richard

[rvida]

Btw, the Android version does not seem to support the Gaviota tablebases, but if you want, you can add it on that platform too.

Neither does the linux version. It is on my todo list though. On these platforms there is a complication with my approach of separating tablebase code into a shared library. I imagine there would be a lot of users confused about where to put libgtb.so and what the hell is this LD_PATH variable?
Probably static linking is more reasonable.

[rvida]

It is way stronger!

Well, one cannot ignore the preliminary results by IPON, so possibly there is some undiscovered bug holding it back in matches?

Maybe it just does not like pondering
There is a lot of stuff one could do during opponent's thinking time - to have a snack, go to bathroom, or just think about non-chess stuff such as girls, soccer and so on...

[zullil]

Hi,

Critter 1.2 is available for win32/64, linux and android platforms.
It also got a new home: http://www.vlasak.biz/critter/

Thanks to Emil Vlasák for web hosting.

Richard

A Mac OS X binary would be very much appreciated. Shouldn't be much different than creating a Linux binary---provided you have access to a Mac.