Re: Octave 3.3.54 snapshot available for ftp

[Fredrik Lingvall]

On 11/22/10 10:04, John W. Eaton wrote:
> On 21-Nov-2010, Fredrik Lingvall wrote:
>
> | I have problems building this one (and 3.3.53 too). It fails here:
> | 
> | <snip>
> | 
> | It's configured with:
> | 
> | ./configure  LDFLAGS="-lgfortran" F77=gfortran CFLAGS="-O3 -pipe
> | -mtune=core2 -march=core2 -mfpmath=sse -msse2 -msse3 -mssse3 -msse4.1
> | -msse4.2 \
> | -fomit-frame-pointer -funroll-loops -fprefetch-loop-arrays -pthread"
> | CXXFLAGS="-O3 -pipe -mtune=core2 -march=core2 \
> | -mfpmath=sse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2
> | -fomit-frame-pointer -funroll-loops -fprefetch-loop-arrays -pthread" \
> | FFLAGS="-O3 -pipe -fdefault-integer-8 -mtune=core2 -march=core2
> | -mfpmath=sse -msse2 -msse3 -mssse3 -msse4.1 -msse4.2 \
> | -fomit-frame-pointer -funroll-loops -fprefetch-loop-arrays -pthread" \
> | --enable-64 --prefix=/usr/local/ --with-blas=/usr/local/lib/libBLAS.so
> | --with-lapack=/usr/local/lib/libLAPACK.so --enable-shared
> | 
> | on a Gentoo amd64 machine (core i7 MacBook Pro).
>
> I checked in a couple of changes that should help when compiling with
> --enable-64.  I'm also attaching a draft of the info you will need for
> building a mostly functional version of Octave with this option.  Note
> the advice about -fdefault-integer-8.  You need to set that in
> F77_INTEGER_8_FLAG and not FFLAGS.  Otherwise, the files in
> libcruft/ranlib will be miscompiled.  Although there is code in
> configure.ac that attempts to move -fdefault-integer-8 from FFLAGS
> to F77_INTEGER_8_FLAG, it does not seem to be working correctly for me
> now.  Help debugging that problem would be welcome...
I will test this ASAP.

I will also try to make the Gentoo installer (emerge) to build the
suitesparse, arpack etc. libs according to your instructions.
> Help improving the attached document would also be welcome.  I suppose
> it should go in the manual now, but I have not had time to convert it
> to Texinfo.
>
> Also, are you sure you need --enable-64?  Are you working with arrays
> that contain more than 2^31 elements?  For double precision arrays,
> that is 16GB.  How much memory does your MacBook have?  u
I only test build this on the MacBook (which has 8GB) but I intend to
run it on a 32GB 8-core (Westmere) machine.

/Fredrik