rtp1@quads.uchicago.edu (raymond thomas pierrehumbert) (11/16/90)
I am rather happy with my DN10K as a high end cruncher (especially
with recent performance improvements in the vector library, about
which I will say more some other time), but we will need to be buying
a rather large number of individual workstations in the next few
years. The path of least resistance for these is the SparcStation,
but I would like to hear other opinions, specifically on how the 400t
series stacks up against it with regard to:
--Price of a color system
--Graphics capabilities
--Availability and price of third party memory, magnetic disks, and
magneto-optical disks
--floating point performance
--availability of "ready to make" public domain software, like the
NCSA stuff
--quality, price and availability of commercial software for graphics,
numerical analysis, database handling (of databases of images and
3D flow fields)
--graphical user interface and class library support (in c++?) for
building relatively portable user interfaces.
In short, the question is why should I buy 400t's rather than
Sparcs? I have an open mind, but if the argument is only that 400t's
are "as good" that won't cut it, as we are more familiar with
Sun equipment, so the sparcs have no learning curve and it is
easier for us to get help. The 400t's have to offer substantial
benefits over a Sparc. Unfair, perhaps, but there it is. As far
as I can see, with OSF/Motif, the user interface issue is the
most compelling advantantage of the 400t's, given that OSF probably
won't be ported to Sparcs (or will it?).
Happy to hear from HP/Ap marketing types as well as technically
oriented users. We run straight BSD unix plus NFS, so the ability
to run Domain is not an issue. Also, we don't have much need for
fancy source code control systems. We do fluid dynamics related
to global climate modeling, and analysis of massive databases of
flow data and satellite data.