On my machine Dann Corbit's executable has always been the fastest Crafty based on the "bench" command. So I'll wait and hope that he makes an executable.Check Peter Skinner's Crafty site for an executable.
Slater Wold also has a compile.
For those who have to have it now.
Later.
ftp://cap.connx.com/pub/chess-engines/new-approach/crafty-1913.exeOn my machine Dann Corbit's executable has always been the fastest Crafty based on the "bench" command. So I'll wait and hope that he makes an executable.Check Peter Skinner's Crafty site for an executable.
Slater Wold also has a compile.
For those who have to have it now.
Later.
10-Q's to all.ftp://cap.connx.com/pub/chess-engines/new-approach/crafty-1913.exeOn my machine Dann Corbit's executable has always been the fastest Crafty based on the "bench" command. So I'll wait and hope that he makes an executable.Check Peter Skinner's Crafty site for an executable.
Slater Wold also has a compile.
For those who have to have it now.
Later.
The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
are the two compiles comparable, that is do they have the same options compiled in (DETECTDRAW, FUTILITY etc.)? This might have a bearing on speed as well.10-Q's to all.ftp://cap.connx.com/pub/chess-engines/new-approach/crafty-1913.exeOn my machine Dann Corbit's executable has always been the fastest Crafty based on the "bench" command. So I'll wait and hope that he makes an executable.Check Peter Skinner's Crafty site for an executable.
Slater Wold also has a compile.
For those who have to have it now.
Later.
The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
Roger,The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
Hello Norm,
Thanks for the heads up. Have you tested the versions Slater Wold compiled?
Later.
Peter uses the Intel compiler so if your running on a AMD platform his compile will be slower. As for the switches I do not know but the DETECTDRAW will be slower Hyatt tested this (and I know Dann Loves this switch). As for the FUTILITY switch, you can tell if it has been used in the compile by running the bench mark on a single cpu and looking at the total nodes searched. If you look below you will see that Dann's compile searches more nodes and thus he did not use the FULILITY switch. Another difference between Dann's and Peter's is that Peter adds SMP support in his compile. This slows things down if you are on a single CPU.are the two compiles comparable, that is do they have the same options compiled in (DETECTDRAW, FUTILITY etc.)? This might have a bearing on speed as well.10-Q's to all.ftp://cap.connx.com/pub/chess-engines/new-approach/crafty-1913.exeOn my machine Dann Corbit's executable has always been the fastest Crafty based on the "bench" command. So I'll wait and hope that he makes an executable.Check Peter Skinner's Crafty site for an executable.
Slater Wold also has a compile.
For those who have to have it now.
Later.
The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
Robert
*My Compile with out EPD Kit
cl /O2 /Ot /G7 /DNT_i386 /DVCINLINE_ASM /DUSE_ASSEMBLY /D_MBCS /DFUTILITY /FD /MT /nologo crafty.c egtb.cpp
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].
Crafty v19.13
White(1): bench
Running benchmark. . .
......
Total nodes: 83973817
Raw nodes per second: 1217011
Total elapsed time: 69
SMP time-to-ply measurement: 9.275362
*My Compile with EPD Kit
cl /O2 /Ot /G7 /DNT_i386 /DVCINLINE_ASM /DUSE_ASSEMBLY /D_MBCS /DFUTILITY /DEPD /FD /MT /nologo crafty.c egtb.cpp
EPD Kit revision date: 1996.04.21
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin]
Crafty v19.13
White(1): bench
Running benchmark. . .
......
Total nodes: 83973817
Raw nodes per second: 1166303
Total elapsed time: 72
SMP time-to-ply measurement: 8.888889
*Dann's Compile
EPD Kit revision date: 1996.04.21
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin].
Crafty v19.13
White(1): bench
Running benchmark. . .
......
Total nodes: 87625750
Raw nodes per second: 1184131
Total elapsed time: 74
SMP time-to-ply measurement: 8.648649
*PETER's Compile Using Intel 8
CFLAGS = /O3 /G6 /Gr /Ob2 /Qip /Qipo /QaxiW /W0
COPTS = /DFAST /DEGTB6 /MT /DSMP /DCPUS=4 /DNT_INTEREX /DUSE_ASSEMBLY /DVC_INLINE_ASM /DEPD /DFUTILITY
Initializing multiple threads.
System is SMP, not NUMA.
EPD Kit revision date: 1996.04.21
unable to open book file [./book.bin].
book is disabled
unable to open book file [./books.bin]
Crafty v19.13 (1 cpus)
White(1): bench
Running benchmark. . .
......
Total nodes: 83973817
Raw nodes per second: 1076587
Total elapsed time: 78
SMP time-to-ply measurement: 8.205128
*Slater's Compile using Whidbey (.NET 2005)
CFLAGS = /O2 /Ogxtb2 /G7 /arch:SSE /fp:fast /GL
EPD Kit revision date: 1996.04.21
unable to open book file [./book.bin]
book is disabled
unable to open book file [./books.bin
Crafty v19.13
White(1): bench
Running benchmark. . .
......
Total nodes: 83973817
Raw nodes per second: 1253340
Total elapsed time: 67
SMP time-to-ply measurement: 9.552239
I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Hi Roger!The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
Hello Norm,
Thanks for the heads up. Have you tested the versions Slater Wold compiled?
Later.
Your right, there is a problem.... Both Bob and Eugene are look in to this.I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Regards,
Mogens
Should be here:Hi Roger!The bench test with the P. Skinner version gave me 756520 nodes per second, while the bench test with the D. Corbit version gave me 826658 nodes per second. So I will continue to use the DC version.
Hello Norm,
Thanks for the heads up. Have you tested the versions Slater Wold compiled?
Later.
Are those Slater Wold versions available somewhere for the publicum?
And don´t forget Mike Byrne wcrafty_p4 version who is very fast on my very slow computer.
Ómar.
Where can you get the free MS compiler?I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Regards,
Mogens
Where can you get the free MS compiler?I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Regards,
Mogens
http://msdn.microsoft.com/visualc/vctoolkit2003/Where can you get the free MS compiler?I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Regards,
Mogens
Thanks!!!http://msdn.microsoft.com/visualc/vctoolkit2003/Where can you get the free MS compiler?I think there may be something wrong with the Crafty source. None of the public compilations are able to access the tablebases. They find them, but are seemingly unable to use them. Perhaps a bug or two snug in during the alteration of the tablebase specific code.I just tried Slater's CFLAGS with the free commandline MS C++ compiler that uses the Incremental Linker and got the fastest results...... So anyone can get the Free Crafty source and the Free MS Compiler and taylor it to their system for the fastest crafty. Note if you have a Intel pentium system you can use the SSE2 which should increase your speed.
Regards,
Mogens
Return to Archive (Old Parsimony Forum)
Users browsing this forum: No registered users and 28 guests