On Fri, 15 Aug 2003, Raymond Dalgleish wrote:
> Does anybody have experience, good or bad, with running GCG
> under Red Hat Linux?
>> Does a 32-bit operating system impose any limitations that could
> be avoided with a 64-bit operating system?
Sorry about not responding to this message earlier. We have been running
GCG on a Dell PowerEdge 6650 RedHat Linux 7.2 server with 4 Xenon cpu's
for almost a year now. I have been pleasantly surprised by how well the
system works. I agree that the 32-bit architecture is a restriction and
GCG would undoubtedly run faster on a 64-bit system. However, our new
server is so much faster and more powerful than our old one (a single cpu
Alpha 500), that in practice the difference is not noticeable.
A bigger concern is the necessity for a true server-class RedHat machine.
We ran GCG at this last month's MBL Workshop on Molecular Evolution on a 2
cpu deskside tower. I do not know the particulars of the machine we used,
but it was not a 'true server' class machine and was unsatisfactory for
the course. We had numerous crashes, particularly with SeqLab and X, and
database similarity searches were unacceptably slow. If you're
considering going this route, buy the biggest, baddest RedHat server you
can get! Our PowerEdge has been incredibly stable and plenty fast.
Another small concern is a bug in SeqLab on RedHat. In Editor mode
selecting non-contigous entries does not work with the control key-mouse
button combo that is supposed to work. However, while at the MBL last
month for the course, we discovered an easy workaround - use the right
mouse button rather than the left along with the control key and it works
correctly. GCG is aware of this issue and will work to correct in in v11.
Cheers - Steve
Steven M. Thompson
A C T G stevet at bio.fsu.edu
/--| FSU CSIT / BioInfo 4U
|--/ Florida State University School of
\-/ Computational Science &
/\ Information Technology
/--\ 1st floor DIRAC 150G
|---\ Tallahassee, Florida
/\ 2538 Winnwood Circle
/--\ Valdosta, Georgia