xr_a_y wrote: ↑Wed Apr 01, 2020 3:48 pm
Using mingw64 (under Windows OS I suppose using cywin ?), the Tools/build.sh script shall be very near to work.
I can indeed confirm that the Tools/build.sh script is working quite well under cygwin64, including the profile driven optimized g++ build.
Maybe a dos2unix on Tools/build.sh and Tools/buildFathom.sh is required depending on your Windows git configuration (some \r can be automatically appended to end lines and cygwin does not like that).
You will also need to set ulimit -s unlimited for massive recursion to work.
Finally I noticed that on my hardware / cygwin version, -flto makes linker crash so I disabled it for when cygwin is detected.
Moreover g++ profile generation with path specified is also an issue with cygwin so I generated profile in-place.
xr_a_y wrote: ↑Thu Apr 02, 2020 3:28 pm
I can indeed confirm that the Tools/build.sh script is working quite well under cygwin64, including the profile driven optimized g++ build.
Maybe a dos2unix on Tools/build.sh and Tools/buildFathom.sh is required depending on your Windows git configuration (some \r can be automatically appended to end lines and cygwin does not like that).
You will also need to set ulimit -s unlimited for massive recursion to work.
Finally I noticed that on my hardware / cygwin version, -flto makes linker crash so I disabled it for when cygwin is detected.
Moreover g++ profile generation with path specified is also an issue with cygwin so I generated profile in-place.
Thank you for your help Vivien. I must admit and you have surely noticed that my knowledge is quite inadequate for this task. I am not a programmer (the last time I programmed in C was 34 years ago), I am only trying to spend my increased free time to learn something useful. I have not altogether given up on compiling Minic but I think I'd better try something easier first. At the moment I am still in the dark as to what git is and how to use a script.
On the bright side, using the command line to activate syzygy usage has done the desired effect.
Gabor Szots wrote: ↑Thu Apr 02, 2020 4:09 pm
On the bright side, using the command line to activate syzygy usage has done the desired effect.
Best regards,
Gabor
It is possible that 2.0 is a very little weaker than 1.54, but I'd be surprised 1.54 is stronger than 2.05.
Any test results you can share ? what were the opponents ? what TC ?
Small sample size is very real. The eexpected elo of these two version is close, you need a few thousand games to get a decent likelihood of superiority.