In article <5ccnii$vn at lyra.csx.cam.ac.uk> tjrc1 at mole.bio.cam.ac.uk (Tim Cutts) writes:
> Hmm. Networking GCG is *not* cheap. You need the supoprt staff (as I
> said above), and the hardware you require is expensive.
> Unfortunately, GCG are refusing to support Intel based UNIX systems,
> otherwise we'd be laughing.
There is some hope. For the EGCG (extended GCG) suite, we are looking
to make most programs run on Linux (etc.) by the 10.0 release (but not
for the 9.0 release sadly).
> Parts of GCG are *very* RAM hungry, so
> even if you buy a low-end workstation to run it on, you still need
> lots of memory. Our current machine has 128 MB RAM, and this is fine
> for the main searching programs (even BLAST, which is notoriously RAM
> hungry), but the srsbuild program caused this machine to crawl; this
> one program allocated over 300 MB of virtual memory while rebuilding
> SRS indexes so that lookup would work, and took days to complete. I
> would say that given the size of databases now, your UNIX box tht you
> run GCG on really should have at least 256 MB RAM, particularly if
> there are a lot of users (mole.bio.cam.ac.uk has well over 1000 users)
This is because GCG is still using SRS 4.0. The new SRS 5.0 builds
indices in *far* less memory. Of course, there are some compatibility
issues - but at Sanger I turn LookUp off as it has so much of SRS
missing and just run the native SRS (actually we run both 4.08 and
5.01 at present while we update the rest of the database parsers to
Icarus, but that is a story for the SRS newsgroup :-)
So, it is really more of a GCG problem than an SRS problem.
Oh well, at least GCG 9.0 has renamed (most) SRS commands to "gcg_srsbuild"
and so on.
--
------------------------------------------------------------------------
Peter Rice | Informatics Division,
E-mail: pmr at sanger.ac.uk | The Sanger Centre,
Tel: (44) 1223 494967 | Wellcome Trust Genome Campus,
Fax: (44) 1223 494919 | Hinxton, Cambridge, CB10 1SA,
URL: http://www.sanger.ac.uk/~pmr/ | England
