Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!sun-barr!cs.utexas.edu!uunet!mcvax!hp4nl!philapd!ssp1!roelof
From: roelof@idca.tds.PHILIPS.nl (R. Vuurboom)
Newsgroups: comp.arch
Subject: Re: Apollo 10,000 R.I.P.
Summary: ...and another one bites the dust.
Message-ID: <131@ssp1.idca.tds.philips.nl>
Date: 7 Jun 89 16:29:34 GMT
References: <5125@pt.cs.cmu.edu>
Organization: Philips Telecommunication and Data Systems, The Netherlands
Lines: 68

In article <5125@pt.cs.cmu.edu> lindsay@MATHOM.GANDALF.CS.CMU.EDU (Donald Lindsay) writes:
>
>HP has announced that the Apollo 10,000 will be phased out.
>
>Since new machines are being invented at an amazing rate, I suppose
>it's inevitable that some vanish. It's just a real shame when camels
>live and stallions die.
>
When I was at the Hannover Messe a couple of months back the DN10000VS was
on display. I thought it was the most impressive mini displayed at the fair.

Come to think of it our new mini was being displayed so better make that
"second most impressive mini" :-).

I picked up a technical sheet they were giving out.

Here's some data:

The system is a 64 bit architecture. The central X-bus as they call it
can take 150 Mb/s. You can put up to 4 cpus on it, 128 Mb of physical 
memory and a graphics subsystem. It can connect through a special
interface to a 32-bit 12Mb/s VMEbus _and_ a 16-bit 3Mb/s pc/at bus.

cpu

apollo prism risc multiprocessor
64 bit architecture consisting of a 32x32 integer register file and 2 ecl
floating point processors each(?) consisting of a 32x64 or 64x32 floating
point register file.
128 kb instruction cache and 64 kb data cache both with 64 bit line size.

memory

8-128 kb

graphics subsystem

1.1 million 3d vector transformations/s
65 million pixels/s by plane fills

graphics screen

1280x1024
70 hz noninterlaced
0.26 mm pixel size
(the graphics quality was the best I'v ever seen)

digital signal processing

3.9 ms for a complex fourier transformation of 1024 points (5.3 ms in
double precision) per cpu

performance

15-30 vax mips (1 cpu)
36 double precision mflops peak (1 cpu)
1.5 million 4x4 matrix transformations/s
27000 drystones/s (1 cpu)
5.8 double-precision linpack megaflops

The sheet keeps talking about "1 cpu" and not "per cpu" so I suspect that
a multiprocessor version was not yet working.

I think its a shame that hp has decided to kill this one, I guess the good
do sometimes die young.
-- 
Roelof Vuurboom  SSP/V3   Philips TDS Apeldoorn, The Netherlands   +31 55 432226
domain: roelof@idca.tds.philips.nl             uucp:  ...!mcvax!philapd!roelof