dev-cpp-users Mailing List for Dev-C++

Hello,

Dne 24.8.2013 1:25, Lostgallifreyan napsal(a):
> (And the --help files I directed to text files say many things about optimisations
> but nothing about what 'n' might be or what sorts of groupings of optimisations might
> be done by those preset optimisers).

That's described in the GCC documentation:
http://gcc.gnu.org/onlinedocs/gcc-4.8.1/gcc/Optimize-Options.html#Optimize-Options

> My old command line was this: (Whole batch file..)
> 
> @ECHO OFF
> PATH=%PATH%;E:\CODING\GCC\BIN
> E:\CODING\GCC\BIN\WINDRES.EXE -Ocoff -IC:\WINDOWS\DESKTOP\TEST C:\WINDOWS\DESKTOP\TEST\FRAME.rc C:\WINDOWS\DESKTOP\TEST\FRAME.o
> E:\CODING\GCC\BIN\GCC.EXE -O3 --fast-math -IE:\Coding\gcc\include -IC:\WINDOWS\DESKTOP\TEST -mwindows C:\WINDOWS\DESKTOP\TEST\PHASEMOD.c C:\WINDOWS\DESKTOP\TEST\FRAME.o E:\CODING\GCC\LIB\libcomctl32.a E:\CODING\GCC\LIB\libwinmm.a -oC:\WINDOWS\DESKTOP\TEST\PHASEMOD.exe
> PHASEMOD.EXE

It looks like you don't strip debugging info from executable. You should
add -s option to the GCC command or call strip.exe on the build executable.

Regards,
Michal Molhanec

Hello. While posting about another issue on the MinGW mailing list, I was asked why I was still using v3.4.5, which was the latest I knew of, a slight update to the one I had in Dev-C++. Someone suggested that v4.5.2 might work so I went humnting, and did slightly better:

http://sourceforge.net/projects/orwelldevcpp/files/?source=navbar

I unpacked the entire .7z file from http://sourceforge.net/projects/orwelldevcpp/files/Compilers/MinGW/MinGW%204.7.2.7z/download into a sub directory, added a path to its bin directory to my path, and it works on W98 SE. So far, so good, but the output is slower than that of v3.4.5, when people kept telling me it would be faster!

Is there something wrong, or something I else I should be doing? I ask because the install is huge compared to the old v3.4.5, the compiled file size has grown, and the demand on the CPU, having been 67% with v3.4.5 compiles, now chews 91%! It doesn't seem to matter what -On I set 'n' to, and -Ofast didn't help either. (And the --help files I directed to text files say many things about optimisations but nothing about what 'n' might be or what sorts of groupings of optimisations might be done by those preset optimisers).

My old command line was this: (Whole batch file..)

@ECHO OFF
PATH=%PATH%;E:\CODING\GCC\BIN
E:\CODING\GCC\BIN\WINDRES.EXE -Ocoff -IC:\WINDOWS\DESKTOP\TEST C:\WINDOWS\DESKTOP\TEST\FRAME.rc C:\WINDOWS\DESKTOP\TEST\FRAME.o
E:\CODING\GCC\BIN\GCC.EXE -O3 --fast-math -IE:\Coding\gcc\include -IC:\WINDOWS\DESKTOP\TEST -mwindows C:\WINDOWS\DESKTOP\TEST\PHASEMOD.c C:\WINDOWS\DESKTOP\TEST\FRAME.o E:\CODING\GCC\LIB\libcomctl32.a E:\CODING\GCC\LIB\libwinmm.a -oC:\WINDOWS\DESKTOP\TEST\PHASEMOD.exe
PHASEMOD.EXE

Whatever I do I'd like it to be making a (hopefully) smaller, and a (definitely) faster program. :) I can thrash thrpough my code trying to optimise better any time I like (and I certainly will, as I did before), but for now I need to trust GCC to do better than me, as I have been advised to do, and so far it's not doing it. It's always possible that things I did that help v3.4.5 really do not help here, but that's another foray that had best start with the basic question above..

Crow.