----- Original Message -----
From: "Rhys Ulerich" <[email protected]>
To: "Richard Haney" <[email protected]>
Cc: <[email protected]>
Sent: Tuesday, September 25, 2012 2:34 AM
Subject: Re: [Help-gsl] Target machine defined by gsl-1.15 ? Or by gcc ?
Hi Richard,
I did a build of gsl-1.15 using gcc.exe (tdm64-1) 4.6.1 under MSYS, and I
used the following command line for configure:
./configure
CFLAGS="-g -O2 -march=native -fexceptions" --prefix=/c/gsl-1.15_64_-g_O2
&> gsl-1.15_configure.out
...
...the config.log has the following line in the "Platform" section
display:
uname -m = i686
I would have expected something different from 'uname -m'. But I know
nothing about MSYS.
Do the compilation messages echoed by 'make' omit the -march=native
flag you supplied?
If not, can you compile a small test binary with -march=nature and -S
that should contain, say, SSE instructions and observe them in the
output assembly.
If so, then I'd dig into the reason behind the weird 'uname -m' output.
FWIW my MSYS shell outputs the same for 'uname -m'.
I build both 64-bit and 32-bit versions of gsl-1.15 (both dynamic and
static) in that shell and I've not experienced any problems .... but then I
also haven't tried supplying the CFLAGS argument that Richard is using.
In fact, I don't provide any CFLAGS arg at all. (I think this means that
'-g -O2' (as specified by configure script ?) is used.)
Cheers,
Rob
