Crafty 22.0 Win32 build available

[Jim Ablett]

Crafty 22.0 by Prof. Robert Hyatt
ftp://ftp.cis.uab.edu/pub/hyatt

Win32 Intel compiler 10 p.g.o build
Non-Smp only
http://www.zshare.net/download/788668066888c1/
Mirror:
http://mihd.net/xlg94c


Hi Bob, first thanks very much for the update. Unfortunately
I'm having some trouble with the new code. Smp/thread code
under Windows seems not to be working correctly when compiling
MSVC++\Intel. Setting to more than one cpu will cause a crash.
Also having trouble compiling x64. Can compile and run unoptimised
build (non-smp), but adding any optimizing compiler switches to build
causes a crash when exe is run.

regards,
Jim.

[AdminX]

That's Great, Thanks Bob and Thanks Jim. I hope the 64 bit version will drop soon.

[Dr.Wael Deeb]

Is there a problem compiling a smp 32 win version

[Jim Ablett]

Hi Wael,

Is there a problem compiling a smp 32 win version

Yes, there are source code problems. I think 2 threads (mt=2) is ok, but
smpmt=2 will cause a crash. Might have to wait till v22.1

regards,
Jim.

[Dr.Wael Deeb]

Hi Wael,

Is there a problem compiling a smp 32 win version

Yes, there are source code problems. I think 2 threads (mt=2) is ok, but
smpmt=2 will cause a crash. Might have to wait till v22.1

regards,
Jim.

Thanks Jim,will wait then

[Jim Ablett]

Hi Wael,

Is there a problem compiling a smp 32 win version

Yes, there are source code problems. I think 2 threads (mt=2) is ok, but
smpmt=2 will cause a crash. Might have to wait till v22.1

regards,
Jim.

Thanks Jim,will wait then

I just tested SMP in 32 bit version again - seems to be working ok.
Maybe I made a mistake.

Jim.

[geots]

Jim, maybe i should not be complaining because this is a free version, but it looks to me like it has catastrophe written all over it, setting aside the 64 bit and MP/SP issue. I mean i have 5 previous versions loaded, and i have never seen anything like this. In the rc file i put my tb path. I knew it was right- because all my other versions recognize it. But 22.0 would not. I had no idea until Martin in our testing group said he couldnt make the engine work right until he removed the word "exit" from the last line of the rc file. Just on a hunch- i removed the word exit, and now it accesses TB perfectly. Help me out here, please. Was i supposed to know that? How in God's name would anyone know they had to do that. Or that the word "exit" would even have anything to do with TB. Lastly, i have the log set to OFF in the rc file, and i unticked it in parameters, and it is still creating the files faster than rabbits can multiply. And they are not 0kb files either. Do you have any idea how i can get them to stop creating the files. I know all this is not your responsibility- but maybe you can tell me something that will help.

Much Thanks,

George

[bob]

Jim, maybe i should not be complaining because this is a free version, but it looks to me like it has catastrophe written all over it, setting aside the 64 bit and MP/SP issue. I mean i have 5 previous versions loaded, and i have never seen anything like this. In the rc file i put my tb path. I knew it was right- because all my other versions recognize it. But 22.0 would not. I had no idea until Martin in our testing group said he couldnt make the engine work right until he removed the word "exit" from the last line of the rc file. Just on a hunch- i removed the word exit, and now it accesses TB perfectly. Help me out here, please. Was i supposed to know that? How in God's name would anyone know they had to do that. Or that the word "exit" would even have anything to do with TB. Lastly, i have the log set to OFF in the rc file, and i unticked it in parameters, and it is still creating the files faster than rabbits can multiply. And they are not 0kb files either. Do you have any idea how i can get them to stop creating the files. I know all this is not your responsibility- but maybe you can tell me something that will help.

Can't answer the question about the "exit". I tried it and it works perfectly normally here, although I have never put it at the end of my .craftyrc file. The only advice I have ever given there is that the last line should generally be a blank line, because using notepad to create the file often leaves the last line without a CR/LF on the end, which causes the last .craftyrc command to be combined with the next command sent to the engine. But your problem with the exit makes no sense and it _certainly_ has no effect on endgame table usage in crafty.

If you set "log=off" in your crafty.rc file, crafty absolutely will not produce log files, unless the gui or operator is overriding that. If it does create logs, look in the first log.nnn file to see what is wrong with your log=off command. I play this way on the cluster since I don't want the files either when playing tens of thousands of games, and it does not produce them if this is done correctly.

It would be much better to check what you are doing, before calling this a "disaster" since this specific version has played 50,000 games in the last 2-3 days alone, and we have seen zero problems. Only issue I have found was the smpnice command had one fix that was lost, but this other stuff is simply not related to this version...

I'd almost bet that the "exit" has no cr/lf, and that will certainly wreck things, but that is not a fault of crafty, it is a fault of the notepad/wordpad editor not following the convention of ending each line with a CR/LF...

Much Thanks,

George

[Peter Skinner]

I'd almost bet that the "exit" has no cr/lf, and that will certainly wreck things, but that is not a fault of crafty, it is a fault of the notepad/wordpad editor not following the convention of ending each line with a CR/LF...

This is the exact reason I switched the default notepad to notepad2 or notepad++.

Both allow the cr/lf at the end of the file without a problem. The default notepad/wordpad in Windows is horrible for things like this, thus the reason a nice switch is nice.

Both programs are free; just search on Google. You can even make them your default text editor.

[geots]

Jim, maybe i should not be complaining because this is a free version, but it looks to me like it has catastrophe written all over it, setting aside the 64 bit and MP/SP issue. I mean i have 5 previous versions loaded, and i have never seen anything like this. In the rc file i put my tb path. I knew it was right- because all my other versions recognize it. But 22.0 would not. I had no idea until Martin in our testing group said he couldnt make the engine work right until he removed the word "exit" from the last line of the rc file. Just on a hunch- i removed the word exit, and now it accesses TB perfectly. Help me out here, please. Was i supposed to know that? How in God's name would anyone know they had to do that. Or that the word "exit" would even have anything to do with TB. Lastly, i have the log set to OFF in the rc file, and i unticked it in parameters, and it is still creating the files faster than rabbits can multiply. And they are not 0kb files either. Do you have any idea how i can get them to stop creating the files. I know all this is not your responsibility- but maybe you can tell me something that will help.

Can't answer the question about the "exit". I tried it and it works perfectly normally here, although I have never put it at the end of my .craftyrc file. The only advice I have ever given there is that the last line should generally be a blank line, because using notepad to create the file often leaves the last line without a CR/LF on the end, which causes the last .craftyrc command to be combined with the next command sent to the engine. But your problem with the exit makes no sense and it _certainly_ has no effect on endgame table usage in crafty.

If you set "log=off" in your crafty.rc file, crafty absolutely will not produce log files, unless the gui or operator is overriding that. If it does create logs, look in the first log.nnn file to see what is wrong with your log=off command. I play this way on the cluster since I don't want the files either when playing tens of thousands of games, and it does not produce them if this is done correctly.

It would be much better to check what you are doing, before calling this a "disaster" since this specific version has played 50,000 games in the last 2-3 days alone, and we have seen zero problems. Only issue I have found was the smpnice command had one fix that was lost, but this other stuff is simply not related to this version...

I'd almost bet that the "exit" has no cr/lf, and that will certainly wreck things, but that is not a fault of crafty, it is a fault of the notepad/wordpad editor not following the convention of ending each line with a CR/LF...

Much Thanks,

George

Guessing that you dont care a bit for UCI anyway, im figuring that "zero" of the these games were run in chessbase with the WB to Uci adapters. ( I know, the hell with CB) Its not possible that 1% of the test games were run WB converted to UCI- because if they had been- this version would have been put on hold.