From berriz:~at~:chasma.harvard.edu  Fri Jan 24 11:23:53 1997
Received: from chasma.harvard.edu  for berriz(+ at +)chasma.harvard.edu
	by www.ccl.net (8.8.3/950822.1) id JAA08658; Fri, 24 Jan 1997 09:39:40 -0500 (EST)
Received: by chasma.harvard.edu (AIX 3.2/UCB 5.64/4.03)
          id AA24925; Fri, 24 Jan 1997 09:39:55 -0500
Date: Fri, 24 Jan 1997 09:39:55 -0500
From: berriz #at# chasma.harvard.edu (Gabriel Berriz)
Message-Id: <9701241439.AA24925 ^at^ chasma.harvard.edu>
To: chemistry %-% at %-% www.ccl.net
In-Reply-To: <Pine.SUN.3.94.970124015104.27505D-100000 %-% at %-% panix3.panix.com> (message from Walter Polkosnik on Fri, 24 Jan 1997 01:53:35 -0500 (EST))
Subject: Quanta, Alphas & Xterms (Was: "Running Quanta remotely" & "Quanta on AlphaStation clones?")




First, on Quanta and X-terms:

   Date: Fri, 24 Jan 1997 01:53:35 -0500 (EST)
   From: Walter Polkosnik <walt $#at#$ panix.com>

   On Thu, 23 Jan 1997, Gabriel Berriz wrote:
   > Another option that has come up is to consolidate all our CPU power on
   > a single server, and to use NCD X-terminals as consoles on people's
   > desks.  Does anyone know if it is possible to view and interact with
   > Quanta (which is running on the server) remotely through an NCD
   > X-terminal?

   Of course, this is the whole point of X-windows. I have run Quanta in
   essentially this manner. You may want to worry about speed of the net
   connections between the server and X-terminals (graphics updating can be
   slow depending on the network).

This sure is a confusing issue.  I recall hearing before something
like what you write here, and I have had some e-mail responses along
the same lines as yours, but below are a couple excerpts from e-mail
sent to me by two different MSI folks in response to this CCL query:

   As far as running Quanta remotely, this can *not* be done using
   X emulation.  We had, at some time in the past, done some work
   running Quanta with DGL, and the core visualization seemed to
   run fine, but it was quite slow, and there were difficulties with
   a lot of the more expert functions.

   ...QUANTA will not work over X, it requires local graphics
   hardware.

(The reason I mentioned NCD in my query is that they have a high-end
graphics-oriented line (HMX), which supports OpenGL, and that I was
hoping would be able to handle Quanta.)

After the letters from MSI, my conclusion at this point is that
planning a computer setup around the possibility of using Quanta96
would be very unwise.  (If anybody has had better luck with remotely
run *Quanta96* than the above would suggest, please pipe in.)

Now, on Quanta and Alphas, by same two MSI people:

   We (MSI) dropped our support of Quanta on the Digital Alpha
   some time ago.  Older versions of Quanta ran fine, but the latest
   release, Quanta96 has not been ported, nor (obviously) tested.

   ...We did support DEC Alpha workstations until recently, but they
   have been dropped.

Bottom line, no Alphas either.

I'm struck by Quanta's "pickiness" (as one of my e-mail responders put
it).  I tend to associate this narrow hardware dependance with the
personal computer world; it seems to me extraordinary (not to say
archaic) by Unix world standards.  I guess I'm an idealist...
Quanta's (and in general MSI's) predominance in this market means that
every computational chemistry and structural biology lab will have to
have, at the very least, one or a few dedicated "Quanta boxes" around
(i.e. good-bye to a simple, homogeneous xterms+server configuration),
or, at the other end, a computer configuration planned around Quanta.
(Though this pessimistic view may only reflects my provincialism.)
For labs that use Quanta heavily, and have no other conflicting
requirement, the latter course may make most sense; for our lab, whose
Quanta use is a minute fraction of total CPU load, this course would
be a case of the tail wagging the dog.

Of course, our lab will continue to use Quanta, there's no doubt about
this, but I personally (thinking/dreaming about my future lab here)
would be very interested in learning about credible alternatives to
Quanta for Unix that are a) more portable across hardware platforms
and/or b) capable to run well over a (reasonably fast local) network.
This is because a) as a consumer, it makes sense for me to reward with
my business those software companies that are committed to the
principle of maximum portability, and thus will be less likely to
force my hand at the time of choosing my hardware; and b) for an
academic computational chemistry lab (which tend to be managed by
graduate students who'd rather be graduating), a configuration
consisting of X-terms attached over a fast local network to two
powerful servers (for minimal redundancy) is the simplest, cheapest,
most commonsensical one.  Quanta appears to go against both these
considerations.  If credible alternative to Quanta as described above
does not exist, well, to all those computational chemistry
Ph.D. candidates unenthusiastic about an academic future, voila a
niche.

Many, many thanks to all those who responded to my queries.  I hope
the above information, and (very biased) opinions are useful to those
considering computer hardware purchases in the near future.

Gabriel Berriz