GRAPE newletter vol.2 (05 Feb 2001)

Dear Colleagues:

This is the second issue of a GRAPE newsletter that we intend to send
out every few months, to inform you of new developments.  We will post
this information on our web site as well.  Note a new feature: `reports
from the field', where we will provide information about GRAPE systems
installed in various places in the world.  We start off with Marseille.

If you prefer not to receive email versions of this newsletter, please
see below for instructions as to how to unsubscribe.

       +------------------CONTENTS:-----------------------------+
       | 1) GRAPE-6 boards for sale, from 130 to 1000 Gflops    |
       | 2) Another Gordon Bell prize awarded: to two GRAPEs!   |
       | 3) World wide map of GRAPE installations               |
       | 4) GRAPE-5 cluster installed at NAO-Japan              |
       | 5) IAU Symposium 208, Tokyo, July 10-13, 2001          |
       | 6) Reports from the field: GRAPEs in Marseille         |
       +--------------------------------------------------------+

1---------1---------1---------1---------1---------1---------1---------1

 GRAPE-6 BOARDS FOR SALE, FROM 130 TO 1000 GFLOPS

Single board versions of the high-precision GRAPE-6 hardware are now
available.  The base configuration consists of:

   --- GRAPE-6 processor unit with one processor module, containing
       four GRAPE-6 chips per module (130 Gflops peak speed)
   --- a host interface card

One processor unit can house up to eight processor modules. The peak
performance of one unit with eight processor modules (32 chips) is 1 Tflops.

The commercial version is manufactured by

  Hamamatsu Metrix
  4-10-8 Shin-toda
  Hamamatsu City
  Shizuoka Prefecture, Japan
  Tel +81-53-428-8400
  Fax +81-53-428-8401
  Contact: Minoru Hibino m_hibino@metrix.co.jp

You can contact them directly if you live in Japan.  If you would like
to order GRAPE-6 hardware from overseas, please contact:

  Tansei Intertek, Inc.
  Takadanobaba,
  Shinjuku-ku, Tokyo 169-0075
  TEL +81-3-3366-8585
  FAX +81-3-3363-6649
  Contact: Kazuyuki Nagasaka nagasaka@post.email.ne.jp

They can provide you with price quotes and delivery schedules.

2---------2---------2---------2---------2---------2---------2---------2

 ANOTHER GORDON BELL PRIZE AWARDED: TO TWO GRAPES!

On Nov 9, 2000, the GRAPE-6 was awarded the Gordon Bell prize for
performance.  The calculation submitted was a simulation of a triple
black hole system embedded in the center of a galaxy simulated by
768,000 stars. The sustained performance achieved was 1.349 Tflops
on a prototype 8-board GRAPE-6 system.

This year, GRAPE-6 shared the performance prize with its brother, the
MD-GRAPE2 machine developed by Toshi Ebisuzaki's group in RIKEN. They
achieved 1.34 Tflops for the simulation of melted salt (NaCl).

After both GRAPEs had become finalists for the prize (a `subway series'
for Tokyo!), the jury decided not to do a recount, but to split the prize.

The full GRAPE-6 configuration is expected to become ready around the
end of 2001,  with a peak speed of 100 Tflops.

3---------3---------3---------3---------3---------3---------3---------3

 WORLD WIDE MAP OF GRAPE INSTALLATIONS

Tomohiro Sensui has updated the world wide map of GRAPE locations.
There are currently 34 places which have 136 boards installed.
See the GRAPE website; click on "Around the world"

4---------4---------4---------4---------4---------4---------4---------4

 GRAPE-5 CLUSTER INSTALLED AT NAO-JAPAN

The GRAPE-5 cluster of NAOJ, the National Astronomical Observatory in
Mitaka, Tokyo (see Item 3, Newsletter No. 1) was delivered in December
2000, and is now in full service for Nation-wide use by Japanese
astrophysicists.  The cluster consists of 16 GRAPE-5 boards (40 Gflops
each), each with a 2-CPU Alpha host computer (833 MHz 21164x2, 8MB
Cache, 2GB memory). The 16 hosts are connected by a 16-port Myrinet
switch.  More details can be found at:

  <http://www.cc.nao.ac.jp/~kokubo/muv/>

This page is in Japanese only, but from the illustrations and tables
you can get an idea about the overall setup.

5---------5---------5---------5---------5---------5---------5---------5

 IAU SYMPOSIUM 208, TOKYO, JULY 10-13, 2001

The full program schedule is now available for IAU Symposium 108, to
be held in Tokyo, July 10-13, on `Astrophysical Supercomputing Using
Particles'.  See our website

  <http://grape.c.u-tokyo.ac.jp/iau208.html>

6---------6---------6---------6---------6---------6---------6---------6

 REPORTS FROM THE FIELD: GRAPES IN MARSEILLE

 by Lia Athanassoula (email: lia@paxi.cnrs-mrs.fr)
 and Albert Bosma (email: bosma@batis.cnrs-mrs.fr)

 web site <http://www-obs.cnrs-mrs.fr/dynamique/000dyn1.html>

At Marseille Observatory (Marseille, France), we started the
construction of an environment for N-body simulations at the end of
1989, with the installation of NEMO software on a Sparcstation.  We
got some help with this from Piet Hut and Peter Teuben, and the first
visitor, three months later, was Jun Makino.  He introduced us to the
GRAPE plans, and one of us visited Tokyo in 1991, to start work on
GRAPE-3.  These were only the beginning of a long and very pleasant
stream of visits, which stimulated a fruitful collaboration between
the Tokyo and the Marseille group.  We acquired successively a
GRAPE-3A (handwired), a system with 5 GRAPE-3AF's, a GRAPE-4, and now
have also 6 GRAPE-5 boards and a baby GRAPE-6 board. Our GRAPE-3
systems are linked to Sparc and Ultra stations, and the others to
Alpha stations, DS20s and XP1000s. Our systems were used for some of
the tests of the new PCI interface boards.

Our simulations cover mainly the field of galaxies and galaxy systems
(with a small sprinkling of simulations on planet formation). A large
number of simulations were devoted to understanding the interaction
between a disc galaxy and a small spherical companion. Subtopics
include the fate of the companion, the heating of the disc, the
formation of a bulge, the formation or destruction of a bar and the
origin of ring galaxies. We worked on compact groups, where we found
that it is possible to prohibit the merging within a Hubble time,
provided the mass distribution is predominantly in a common halo with
a large core. We have also looked at the formation of BCMs and cDs
in large groups of galaxies, found conditions which favour their
formation, and set constraints on the distribution of matter in the
common halo of the cluster. We are now embarked on a large study of
the secular evolution of barred disk galaxies, to understand the
formation of the bar, and to construct indicators of the properties
of the dark halos. We also worked out ways to determine the ``best''
softening parameter, thereby optimizing the performance of the
GRAPE-treecode.

Our scientific plans in the future include more work in isolated and
interacting galaxies, including a gaseous component. On the technical
side we would like to couple our GRAPE-5 systems together, to be able
to tackle problems necessitating a larger number of particles.

0---------0---------0---------0---------0---------0---------0---------0

           Piet Hut and Jun Makino
           (submissions to: <piet@astrogrape.org> or <grape@astrogrape.org>)


 HOW TO UNSUBSCRIBE

Just send an email message to <grape@astrogrape.org>

with Subject:unsubscribe

00--------00--------00--------00--------00--------00--------00--------00