There is a major issue here: the size of our available genetic
information. Linux on Intel architectures is currently 32 bit - this
will not change until merced PCs running linux are available. Some of
the fasta files we produce as dumps of GCG now exceed a 2GB filesize
by a significant margin. No matter how nice the VA box, they cannot
address this issue. I presume the same holds for solaris/intel -
correct me if I am mistaken.
I am a strong linux advocate (we have dozens of boxes here, and two
clusters), but perhaps a 32 bit linux port of GCG is not the way to
Just my $0.02 (speaking for myself, not GCG).
Greg <jquinn at shell7.ba.best.com> writes:
> [I posted this a few days ago from a local net site,
> but I couldn't see it showing up, so here goes again...
> apologies if it did go through and this is a second post]
>> Sorry for opening this old chestnut, but from my perspective it's now
> becoming extremely relevant. A lot of the servers and databases I
> administrate are now almost exclusively coming online on Linux boxes;
> these aren't necessarily just old PC's, but dedicated multiprocessor
> jobs, for example from VA Research, costing several thousand dollars, but
> more cost-effective than a sparc box.
> Has anyone had any indication from GCG that they intend to distribute the
> next version in Linux binary form? It really isn't clear to me why this
> is not already an option. Presumably compiling the code on Linux is no
> more of a problem than for (example) sparc solaris (BTW is there a version
> of GCG for intel Solaris?). I've now started to set up a lot of jobs
> using EMBOSS rather than GCG because I can get the Linux binaries with
> access to appropriate databases.
>> I guess this is more of a gripe than a question because unfortunately I
> suspect I already know the answer.
N a t h a n O . S i e m e r s
Division of Applied Genomics
Bristol-Myers Squibb Pharmaceutical Research Institute
Hopewell Building 3B, P.O. Box 5400, Princeton, NJ 08543-5400
siemersn at bms.com