Path: utzoo!attcan!telly!lethe!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!att!linac!midway!quads.uchicago.edu!rtp1
From: rtp1@quads.uchicago.edu (raymond thomas pierrehumbert)
Newsgroups: comp.sys.apollo
Subject: Re: DN10000 disk space
Message-ID: <1990Dec7.034301.29493@midway.uchicago.edu>
Date: 7 Dec 90 03:43:01 GMT
References: <90Dec6.172904est.57368@ugw.utcs.utoronto.ca>
Sender: news@midway.uchicago.edu (News Administrator)
Organization: University of Chicago
Lines: 29


I have looked at a lot of machines, and for floating point performance,
if you get all four processors and especially if you count on the
new 2x processor upgrade, the 10k is a very fine piece of hardware.
To get good performance, you will need to use vector library, or
APollo-coded BLAS calls, unless the new compilers are a lot better
than mine.

However:  (get out your ear trumpets, HP), THE DISKS cripple the   
DN10k.  It is not impossible, but it is difficult to get third party
ESDI disks for the 10k.  HP/APollo doesn't do anything to make this
any easier.  If you buy disks from HP/Apollo, even with University
discount, you will pay about as much for your 2.8GB as SCSI users
pay for 10GB.  If you want to hang other SCSI devices off the 
machine, like optical drives or Exabytes, you need to deal with
a custom software/hardware house, like Workstation Solutions.

I don't know anybody who has cracked the 2.8GB barrier on the 10k. It
is a crippling limitation, and the main reason the 10k is not the
clear choice for the data server for our Cloud and Climate center
here at U of C.

For my 10k, I use NFS and hang SCSI disks of a Sparc330.  You get
a performance hit, a big one, for using the network though.

For the 10k to remain a viable machine, HP/APOLLO MUST RELEASE
a VME/AT BUS SCSI BOARD AND SET OF DRIVERS.  There is no way
this machine will remain competitive otherwise.
.