This is the mail archive of the mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: gcc version 3.3.1 (cygming special)

"chris" <> wrote in message">
> Corinna Vinschen wrote:
> >On Mon, Sep 22, 2003 at 02:36:32PM +0300, Alex Vinokur wrote:
> >
> >
> >>------ Cygwin ------
> >>$ gcc foo.c -o foo1.exe
> >>$ gcc foo.c  -mno-cygwin -o foo2.exe
> >>
> >>------ MinGW ------
> >>$ gcc foo.c -o foo3.exe
> >>
> >>1. What is the difference between foo1.exe and foo2.exe?
> >>2. Is there any difference between foo2.exe and foo3.exe?
> >>
> >>
> >
> >Call cygcheck foo[123].exe.
> >
> >
> Just to add to this. For whatever reasons, I have found mingw
> executables to be slightly, but measurably faster (this was done some
> time ago).

Not always.

Comparative performance tests were carried out
  using the same compiler (gcc/g++/gpp 3.2)
  in different environments (CYGWIN, MINGW, DJGPP)
  on Windows 2000 Professional.

Different methods of copying files were tested :

  ------ C methods ------
  Method C-1   : Functions getc() and putc()
  Method C-2   : Functions fgetc() and fputc()

  ------ C++ methods ------
  Method CPP-1 : Operators >> and <<
  Method CPP-2 : Methods get() and put()
  Method CPP-3 : Methods sbumpc() and sputc()
  Method CPP-4 : Method sbumpc() and operator <<
  Method CPP-5 : Method rdbuf() and operator <<

The results for CYGWIN and DJGPP are consistent, in particular:
  C-methods C-1 and C-2 are faster than C++-methods CPP-1, CPP-2, CPP-3.

Whereas it seems that C-methods C-1 and C-2 on MINGW are too slow, e.g.,
  C-methods C-1 and C-2 are slower than C++-methods CPP-2, CPP-3.

  The summary results are can be seen at


   Alex Vinokur

Unsubscribe info:
Problem reports:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]