[Beowulf] g77 limits...

Andrew M.A. Cater amacater at galactic.demon.co.uk
Thu Feb 23 00:26:13 PST 2006


On Wed, Feb 22, 2006 at 02:35:52PM -0800, Joel Jaeggli wrote:
> On Wed, 22 Feb 2006, Kozin, I (Igor) wrote:
> 
> >
> >>>archives because it is so time dependent an answer.  In particular, can
> >>>an x64 box with a modern linux kernel and a modern g77 allocate 4-8 GB
> >>
> >>isn't "modern g77" a bit of an oxymoron?
> >>(given the existence of gfortran, I mean).
> >
> >Oh, I was totally confused by the combination "g77" and "allocate".
> >
> >Robert, try g95. Although it is not part of the gnu stream but
> >it seems more stable at the moment. It works stand alone.
> >Allocating (nearly) 4GB in a single array worked for me
> >(unfortunately no more physical memory is available).
> >However in terms of performance you may lose considerably
> >compared to commercial compilers.
> 
> g95 is included as of gcc 4 you can try it out on fc4 if you're so 
> inclinded...
> 
> [joelja at twin ~]$ gcc -v
> Using built-in specs.
> Target: i386-redhat-linux
> Configured with: ../configure --prefix=/usr --mandir=/usr/share/man 
> --infodir=/usr/share/info --enable-shared --enable-threads=posix 
> --enable-checking=release --with-system-zlib --enable-__cxa_atexit 
> --disable-libunwind-exceptions --enable-libgcj-multifile 
> --enable-languages=c,c++,objc,java,f95,ada --enable-java-awt=gtk 
> --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre 
> --host=i386-redhat-linux
> Thread model: posix
> gcc version 4.0.2 20051125 (Red Hat 4.0.2-8)
> You have mail in /var/spool/mail/joelja
> 
> 
<rant mode on> And in Red Hat Enterprise 4 - only there, they used a
snapshot dated December 2004 - a full few months before _actual_
release :( Be very, _very_ careful with Red Hat and compiler versions:
I wouldn't be at all surprised to see Fedora Core 5 release in three
weeks or so compiled with GCC "4.1" throughout - 4.1 isn't released yet.
They do employ some of the core GCC maintainers but can still produce
proprietary-type problems by releasing code prematurely that should probably
still be in house - anybody else remember GCC 2.96??  All IMHO</rant mode off>

Folk at work are more concerned over whether GCC 4.x is fully C99
compliant (things like // comments) but we did try some Fortran -
only to be brought up short by non-compliant manufacturer-specific
"extensions" and tweaks. The joys of non-portable code that's
"portable between our machines from one generation to the next but
to nobody else's" so to speak :(

Andy



More information about the Beowulf mailing list