In article <1992Apr18.222928.23373 at bcnews.bc.edu>, grant at darkstar (Grant W. Balkema) writes:
|> We are going to purchase the University of Wisconsin GCG package soon.
|> From what I understand, there is one set that runs on DECVax VMS that
|> is the standard etc. and a newer UNIX port. We will be running the
|> software on either a VAX cluster or if UNIX a SPARC based system. The
|> useage will be mainly via Ethernet connections. I recall some mumblings
|> about Xwindow support, anybody have any comments? I guess the choice
|> would be the UNIX based product, except that the GCG people
|> mentioned that the UNIX based product was a bit buggy.
...
We have both the VMS and the UNIX version running here. The VMS users like
their version but they generally stick to what they know, and dont migrate
easily. X Windows is used only little, because most users have only tek-
tronix or regis graphics. I am a bit concerned about the memory requirements
UCX needs if many users want to use it, and not really delighted about
its performance (speedwise) on the 8840 VAX with >100 interactive
users on it. It is however much better on the 6510, but this one should be
used for computations at this site rather than for graphics.
I had a course recently where I used three SGI systems (a 320 and two indigos)
in addition to a VT1200, and a lot of terminals. The software was running
ok with impressive performance. It was installed only on one machine, while
the others were using NFS to access it. Batches were executed on a distributed
queueing system. Graphics came out smoothly even if the load was very high.
I was sort of depressed on the X-Windows performance of the macintosh
sitting beside an indigo. If your users don't permanently use large screens
the tek4105 on Versaterm Pro via ethernet is much more preferrable than X
on the Mac. I don't want to comment on the VT1200 X terminal because it is not
sold any longer (for good reason). I have used the X-terminal graphics on the
DECstation 5000 and it came out nicely.
I made some tests before we went for a permanent twin-headed system and this
concerned the use of NFS on various stations. As long as you don't use
sequence searches all the time via the network, the loading time for the
GCG UNIX executables via (NFS) network were shorter than the time needed to
load via (DECNET) network on VAXes. So if you plan to have a transparent
network environment where UNIS systems can execute the code locally which
is downloaded from a NFS server the UNIX version seems to be better than the
VMS. You certainly can improve the VMS if you install the software
locally and pay the license, but as far as I understood GCG their policy
is to license per *MOUNTED* software and this is easier on the UNIX world.
Maybe this helps
Regards
Reinhard
The usual disclaimers on hardware software and their vendors apply.
This post is my personal opinion and not a policy statement.
--
+----------------------------------+-------------------------------------+
| Dr. Reinhard Doelz | RFC doelz at urz.unibas.ch |
| Biocomputing | DECNET 20579::48130::doelz |
|Biozentrum der Universitaet | X25 022846211142036::doelz |
| Klingelbergstrasse 70 | FAX x41 61 261- 6760 or 2078 |
| CH 4056 Basel | TEL x41 61 267- 2076 or 2247 |
+------------- bioftp.unibas.ch is the SWISS EMBnet node ----------------+
-----------------------------------------
