Crafty 23.1 reports score to Arena 2.0.1 from its point

[beachknight]

of view, not white's. The problem is apparent when Crafty plays black and its scores are the inverse of white's. The " Values absolute (always from white side )" switch is on in Arena. There's no change if I turn it off.

Dave

And the result is:

http://tech.groups.yahoo.com/group/Ches ... ssage/3541

Best,

PS: Arena switch is on (Values from white side)

[bob]

Here a better descripion of my problem.
I changed the linux entry in the Makefile to

Code: Select all

linux:
        $(MAKE) target=LINUX \
                CC=gcc CXX=g++ \
                CFLAGS='$(CFLAGS) -Wall -pipe -march=native -O3 -mtune=native \
                        -fno-gcse -mpreferred-stack-boundary=4' \
                CXFLAGS=$(CFLAGS) \
                LDFLAGS='$(LDFLAGS) -lpthread -lstdc++' \
                opt='$(opt) -DTRACE -DINLINE64 -DCPUS=2' \
                crafty-make

and build crafty by make linux.
Using hash=2048M gives
White(1): time surplus 0.00 time limit 10:00 (+0.00) (10:00)
depth time score variation (1)
Segmentation fault

What am I doing wrong?
kind regards
Bernhard

OK, some quick questions.

(1) this is 64 bit hardware, correct? IE intel core-something, anything that is X86_64 or EMT64 hardware.

(2) if so, do you have the current source from my ftp box?

I just ftp'ed the crafty-23.1.zip file to our 8-core cluster, compiled it as a 64 bit app, and did this:


ferrum$ ./crafty
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].

Crafty v23.1 (1 cpus)

White(1): hash=4096M
hash table memory = 4096M bytes.
White(1): book off
book file disabled.
White(1): e4
time used: 20.91
time surplus 0.00 time limit 30.00 (+0.00) (3:00)
depth time score variation (1)
11 0.35 0.13 1. ... Nc6 2. Nf3 Nf6 3. Nc3 e6 4.
Bb5 Bb4 5. O-O Bxc3 6. dxc3 Nxe4

As you can see, it runs just fine. Assuming you have the source from the ftp box dated November 20 of this year.

[bob]

of view, not white's. The problem is apparent when Crafty plays black and its scores are the inverse of white's. The " Values absolute (always from white side )" switch is on in Arena. There's no change if I turn it off.

Dave

And the result is:

http://tech.groups.yahoo.com/group/Ches ... ssage/3541

Best,

PS: Arena switch is on (Values from white side)

What is that odd +M0 in white's output? Could that make Arena think white won? I had problems with Arasan producing wrong results, I had to make my referee program on the cluster ignore its claims to solve this.

[BBauer]

Here a better descripion of my problem.
I changed the linux entry in the Makefile to

Code: Select all

linux:
        $(MAKE) target=LINUX \
                CC=gcc CXX=g++ \
                CFLAGS='$(CFLAGS) -Wall -pipe -march=native -O3 -mtune=native \
                        -fno-gcse -mpreferred-stack-boundary=4' \
                CXFLAGS=$(CFLAGS) \
                LDFLAGS='$(LDFLAGS) -lpthread -lstdc++' \
                opt='$(opt) -DTRACE -DINLINE64 -DCPUS=2' \
                crafty-make

and build crafty by make linux.
Using hash=2048M gives
White(1): time surplus 0.00 time limit 10:00 (+0.00) (10:00)
depth time score variation (1)
Segmentation fault

What am I doing wrong?
kind regards
Bernhard

OK, some quick questions.

(1) this is 64 bit hardware, correct? IE intel core-something, anything that is X86_64 or EMT64 hardware.

(2) if so, do you have the current source from my ftp box?

I just ftp'ed the crafty-23.1.zip file to our 8-core cluster, compiled it as a 64 bit app, and did this:


ferrum$ ./crafty
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].

Crafty v23.1 (1 cpus)

White(1): hash=4096M
hash table memory = 4096M bytes.
White(1): book off
book file disabled.
White(1): e4
time used: 20.91
time surplus 0.00 time limit 30.00 (+0.00) (3:00)
depth time score variation (1)
11 0.35 0.13 1. ... Nc6 2. Nf3 Nf6 3. Nc3 e6 4.
Bb5 Bb4 5. O-O Bxc3 6. dxc3 Nxe4

As you can see, it runs just fine. Assuming you have the source from the ftp box dated November 20 of this year.

To your questions
(1) of course it is 64-bit hardware.
(2) I downloded crafty23.1 in dezember
I had a look in chess.h and 2 other places of the sorce code. I changed some "ints" to "size_t" and now it works.

Thanks to Lois Zulli and Aaron Becker.

kind regards
Bernhard

[bob]

Here a better descripion of my problem.
I changed the linux entry in the Makefile to

Code: Select all

linux:
        $(MAKE) target=LINUX \
                CC=gcc CXX=g++ \
                CFLAGS='$(CFLAGS) -Wall -pipe -march=native -O3 -mtune=native \
                        -fno-gcse -mpreferred-stack-boundary=4' \
                CXFLAGS=$(CFLAGS) \
                LDFLAGS='$(LDFLAGS) -lpthread -lstdc++' \
                opt='$(opt) -DTRACE -DINLINE64 -DCPUS=2' \
                crafty-make

and build crafty by make linux.
Using hash=2048M gives
White(1): time surplus 0.00 time limit 10:00 (+0.00) (10:00)
depth time score variation (1)
Segmentation fault

What am I doing wrong?
kind regards
Bernhard

OK, some quick questions.

(1) this is 64 bit hardware, correct? IE intel core-something, anything that is X86_64 or EMT64 hardware.

(2) if so, do you have the current source from my ftp box?

I just ftp'ed the crafty-23.1.zip file to our 8-core cluster, compiled it as a 64 bit app, and did this:


ferrum$ ./crafty
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].

Crafty v23.1 (1 cpus)

White(1): hash=4096M
hash table memory = 4096M bytes.
White(1): book off
book file disabled.
White(1): e4
time used: 20.91
time surplus 0.00 time limit 30.00 (+0.00) (3:00)
depth time score variation (1)
11 0.35 0.13 1. ... Nc6 2. Nf3 Nf6 3. Nc3 e6 4.
Bb5 Bb4 5. O-O Bxc3 6. dxc3 Nxe4

As you can see, it runs just fine. Assuming you have the source from the ftp box dated November 20 of this year.

To your questions
(1) of course it is 64-bit hardware.
(2) I downloded crafty23.1 in dezember
I had a look in chess.h and 2 other places of the sorce code. I changed some "ints" to "size_t" and now it works.

Thanks to Lois Zulli and Aaron Becker.

kind regards
Bernhard

Somehow you did not get the latest version when you downloaded it. The november 20 version has no "ints" that can be changed to size_t, as that was what was fixed. Are you downloading from my ftp site or somewhere else???

[bob]

No, I am not certain that I build a 64-bit compile.
I even dont know how to build a specific 32-bit or 64-bit crafty.
But I I used a -D????64 for compiling and my nps are good, so I think its a 64-bit compile. On Manday I can give you the exact compiler flags and everything else.
Kind regards
Bernhard

If you include the inline64.h then it has to be a 64 bit compiler or it will complain. Still doesn't explain your somehow outdated 23.1 source files however. The current version uses size_t (dated 11/20/09) and compiles/runs fine here with any reasonable hash size up to 16384M which is as large as I can run on the machines we have around.

[BBauer]

Now I guess it was in some way my fault, so I do not know what I did wrong.
When I looked into the sources everythig was ok. And in any case it works now for me.
Sorry for the confusion.
kind regards
Bernhard

[Dirt]

Arena switch is on (Values from white side)

While you've probably already checked for this I wanted to post this information in the subject since it is the easiest way to tell which is which. The bad ones (changed to be from the engine's point of view) are dated Nov 20 (except for the UCI ones which have a 2007 date).

[beachknight]

Arena switch is on (Values from white side)

While you've probably already checked for this I wanted to post this information in the subject since it is the easiest way to tell which is which. The bad ones (changed to be from the engine's point of view) are dated Nov 20 (except for the UCI ones which have a 2007 date).

I use PS build:

crafty-23.1-win64.exe 20.11.2009 21:53

Is there a newer or corrected x64 build?

Best,

[bob]

Arena switch is on (Values from white side)

While you've probably already checked for this I wanted to post this information in the subject since it is the easiest way to tell which is which. The bad ones (changed to be from the engine's point of view) are dated Nov 20 (except for the UCI ones which have a 2007 date).

I use PS build:

crafty-23.1-win64.exe 20.11.2009 21:53

Is there a newer or corrected x64 build?

Best,

\

Yours is certainly dated 11/20/2009 so it may be bad...