Amy 0.8.7 runs but...

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 05 March 2004 04:57:53:

This is the project I am using, in case someone wants to play with it:
ftp://cap.connx.com/chess-engines/new-a ... amy087.zip
I am pretty sure that a memory overwrite happens somewhere.
The binary version on my ftp site has symbols in it (though built in release mode). That version does not crash. However, when built without symbols it crashes right and left. So I am pretty sure that there is something naughty afoot somewhere.
Maybe someone can spend some time to track it down.


my ftp site {remove http:// unless you like error messages}

[Lance Perkins]

Geschrieben von: / Posted by: Lance Perkins at 05 March 2004 05:50:02:
Als Antwort auf: / In reply to: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 04:57:53:

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

This is the project I am using, in case someone wants to play with it:
ftp://cap.connx.com/chess-engines/new-a ... amy087.zip
I am pretty sure that a memory overwrite happens somewhere.
The binary version on my ftp site has symbols in it (though built in release mode). That version does not crash. However, when built without symbols it crashes right and left. So I am pretty sure that there is something naughty afoot somewhere.
Maybe someone can spend some time to track it down.

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 05 March 2004 06:07:37:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Lance Perkins at 05 March 2004 05:50:02:

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

Runnable or not, something is definitely broken. Some kind of undefined behavior.
If I compile with symbols, it never crashes (even a release build).
If I remove the symbols it crashes every game when the board gets sparse. It does not appear to be EGTB though, because there are still a lot of pieces on the board.



my ftp site {remove http:// unless you like error messages}

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 05 March 2004 06:13:17:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 06:07:37:

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

Runnable or not, something is definitely broken. Some kind of undefined behavior.
If I compile with symbols, it never crashes (even a release build).
If I remove the symbols it crashes every game when the board gets sparse. It does not appear to be EGTB though, because there are still a lot of pieces on the board.

I think the problem is in this routine:
void InitScore(struct Position *p)



my ftp site {remove http:// unless you like error messages}

[Lance Perkins]

Geschrieben von: / Posted by: Lance Perkins at 05 March 2004 06:23:20:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 06:07:37:

I have already played several dozen blitz/bullet games with it (in ThinkerBoard) in different build modes: no debug symbols, with debug symbols (-g), and with debug symbols (-g; then 'stripped' the symbols after linking).
It has gone to KKx and KxKx endings and never crashed.
I don't have egtb, so maybe that could be it.
I also did not build it in MP mode.

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

Runnable or not, something is definitely broken. Some kind of undefined behavior.
If I compile with symbols, it never crashes (even a release build).
If I remove the symbols it crashes every game when the board gets sparse. It does not appear to be EGTB though, because there are still a lot of pieces on the board.

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 05 March 2004 06:27:05:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Lance Perkins at 05 March 2004 06:23:20:

I have already played several dozen blitz/bullet games with it (in ThinkerBoard) in different build modes: no debug symbols, with debug symbols (-g), and with debug symbols (-g; then 'stripped' the symbols after linking).
It has gone to KKx and KxKx endings and never crashed.
I don't have egtb, so maybe that could be it.
I also did not build it in MP mode.

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

Runnable or not, something is definitely broken. Some kind of undefined behavior.
If I compile with symbols, it never crashes (even a release build).
If I remove the symbols it crashes every game when the board gets sparse. It does not appear to be EGTB though, because there are still a lot of pieces on the board.

I enlarged a few of the arrays in score.c by one and the problem went away.
Probably an out of bounds access by 1 or something like that.



my ftp site {remove http:// unless you like error messages}

[Dann Corbit]

Geschrieben von: / Posted by: Dann Corbit at 05 March 2004 06:30:45:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 06:27:05:

I have already played several dozen blitz/bullet games with it (in ThinkerBoard) in different build modes: no debug symbols, with debug symbols (-g), and with debug symbols (-g; then 'stripped' the symbols after linking).
It has gone to KKx and KxKx endings and never crashed.
I don't have egtb, so maybe that could be it.
I also did not build it in MP mode.

Seems to work fine with my mingw build (even with debug "-g" option). Played a few games with it.
untar/unzip the source
cd src
copy ..\config.h.in config.h
notepad config.h
add: #define VERSION "0.8.7"
notepad amy.h
add:
"typedef unsigned __int64 hash_t;"
in same place as:
"typedef unsigned __int64 ran_t;"
del *.o
gcc -O3 -I. -c bitboard.c bookup.c commands.c dbase.c eco.c hashtable.c
gcc -O3 -I. -c init.c learn.c main.c mates.c movedata.c next.c pgn.c
gcc -O3 -I. -c probe.c random.c recog.c score.c search.c search_io.c
gcc -O3 -I. -c state_machine.c swap.c time_ctl.c utils.c version.c
gcc -c -O3 mytb.cpp
ignore the warnings (really harmless)
gcc -o amy.exe *.o

Runnable or not, something is definitely broken. Some kind of undefined behavior.
If I compile with symbols, it never crashes (even a release build).
If I remove the symbols it crashes every game when the board gets sparse. It does not appear to be EGTB though, because there are still a lot of pieces on the board.

I enlarged a few of the arrays in score.c by one and the problem went away.
Probably an out of bounds access by 1 or something like that.

There is still some problem.
I am through trying to find it for today though.


my ftp site {remove http:// unless you like error messages}

[Gábor Szõts]

Geschrieben von: / Posted by: Gábor Szõts at 05 March 2004 13:18:44:
Als Antwort auf: / In reply to: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 04:57:53:

This is the project I am using, in case someone wants to play with it:
ftp://cap.connx.com/chess-engines/new-a ... amy087.zip
I am pretty sure that a memory overwrite happens somewhere.
The binary version on my ftp site has symbols in it (though built in release mode). That version does not crash. However, when built without symbols it crashes right and left. So I am pretty sure that there is something naughty afoot somewhere.
Maybe someone can spend some time to track it down.

It seems Amy can't analyze am EPD position if black is on move.

[Slobodan R. Stojanovic]

Geschrieben von: / Posted by: Slobodan R. Stojanovic at 05 March 2004 14:57:30:
Als Antwort auf: / In reply to: Re: Amy 0.8.7 runs but... geschrieben von: / posted by: Dann Corbit at 05 March 2004 06:13:17:

Compiling this Amy has turned to be like an exciting thriller!
It´s very good... for a change.
Good work.
SL.