simulation@uflorida.cis.ufl.edu (Moderator: Paul Fishwick) (10/04/90)
Volume: 18, Issue: 2, Wed Oct 3 14:03:45 EDT 1990
+----------------+
| TODAY'S TOPICS |
+----------------+
(1) Role of CS in Computational Science
(2) RE: Public Domain Simula (msg 1)
(3) RE: Public Domain Simula (msg 2)
(4) DISS: Parallel Simulation of Digital Circuits
(5) REQUEST: Network Simulators
* Moderator: Paul Fishwick, Univ. of Florida
* Send topical mail to: simulation@bikini.cis.ufl.edu OR
post to comp.simulation via USENET
* Archives available via FTP to bikini.cis.ufl.edu (128.227.224.1).
Login as 'ftp', use your last name as the password, change
directory to pub/simdigest. Do 'type binary' before any file xfers.
* Simulation Tools available by doing above and changing the
directory to pub/simdigest/tools.
-----------------------------------------------------------------------------
From: steve@hubcap.clemson.edu ("Steve" Stevenson)
Date: 1 Oct 90 13:54:36 GMT
To: comp-simulation@gatech.edu
Subject: Submission for comp-simulation
Responding-System: hubcap.clemson.edu
Path: hubcap!steve
From: steve@hubcap.clemson.edu ("Steve" Stevenson)
Newsgroups: comp.lang.fortran,comp.simulation,comp.theory.cell-automata
Subject: What is comp.sci's role in computational science?
Date: 1 Oct 90 13:54:36 GMT
References: <1990Aug26.034012.25121@tss.com>
Organization: Clemson University, Clemson, SC
Clemson is developing a Computational Sciences program. This program
has courses which are interdisciplinary in nature. While the engineering/
scientific content and the mathematical/numerical content are reasonably
easy to deal with, the computer science content is not. Computer science,
in this context, means software, languages, etc and not hardware.
I am trying to survey attitudes of scientific programmers, whether in
industry or in academia, on prevailing attitudes relative to how well
computer science helps---or hinders---the development of scientific and
engineering codes. These may relate to such things as langauges, software
development methods/tools, algorithms etc.
Please send any comments by e-mail. Flames welcome if they are clean :-)
and even better if they're funny. I would also be interested in horrid
examples and war stories.
steve
--
===============================================================================
Steve (really "D. E.") Stevenson steve@hubcap.clemson.edu
Department of Computer Science, (803)656-5880.mabell
Clemson University, Clemson, SC 29634-1906
------------------------------
Date: Mon, 1 Oct 90 16:32:24 +0100
From: Boris Magnusson <boris@DNA.LTH.Se>
To: fontes@cs.nps.navy.mil, simulation@bikini.cis.ufl.edu
Subject: P.D. Simula Compiler for Mac/MPW
Cc: boris@DNA.LTH.Se
In response to a request for the Lund Simula system for the Mac/MPW:
There is an implementation of the original object oriented programming language
SIMULA for MAC/MPW. It runs on any Mac (Plus, SE II etc) that runs MPW.
This imples 2 MByte and a hard disc.
The current release (4.07) of Simula for the MAC/MPW is Public domain. It comes
with a Hypercard reference stack for both the language and the interface to
the Toolbox. The quickest way to get hold it is to download it from
rascal.ics.utexas.edu where you find it in under directory:
mac/programming/simula
-rwxrwxr-x 1 19 ftp 251008 May 23 1989 lund_simula_4pt07.sit_bin
-rwxrwxr-x 1 19 ftp 156800 May 23 1989 simula_4pt07_reference_stack.sit_bin
-rwxrwxr-x 1 19 ftp 95360 May 23 1989 sinterfaces.sit_bin
login as "anonymous" under ftp to download the files.
SIMULA is a standardized language. The Swedish Standard (SS 63 61 14) should
be possible to get trough ANSI. The Lund Simula implementaion, which follows
the standard, has been operational since 1981 and ported to various mini
computers and workstations. For Unix based machines the system includes an
interactive, source level debugger.
More information on Lund Simula can obtained from:
Lund Software House AB
Box 7056
S-220 07 LUND, Sweden
or from: boris@dna.lth.se
------------------------------
Date: Tue, 2 Oct 90 09:04:21 EST
From: Richard E. Nance - SRC <srcnance@popeye.cs.vt.edu>
To: simulation@bikini.cis.ufl.edu
Subject: Public Domain Simula
Contact the following:
Lund Software House AB
Box 7056
S-220 07 LUND
Sweden
My version is for Macintosh/MPW and is 4.07.
There is a telephone number in one of the Simula Newsletters,
but I just do not have time to look it up.
Dick Nance
------------------------------
To: comp-simulation@mcnc.org
Path: duke!jvb
From: jvb@cs.duke.edu (Jack V. Briner, Jr.)
Newsgroups: comp.simulation
Subject: Parallel SImulation of Digital Circuits Dissertation Available
Date: 1 Oct 90 18:13:34 GMT
Organization: Duke University Computer Science Dept.; Durham, N.C.
My dissertation titled "Parallel Mixed-Level Simulation of Digital Circuits
Using Virtual Time" is now available to researchers. The abstract appears
at the end of this article.
We will also be distributing the sequential version of the simulator
(LDVSIM) via ftp in the near future. The simulator provides mixed-level
simulation of digital circuits at the gate (or higher) and switch-timing levels.
The switch-timing levels are based upon Chris Terman's work (MIT) as updated by
Mark Horowitz and C. Y. Chu (Stanford).
Please send me US MAIL (not email) if you would like a copy either the program
or the dissertation.
My current address is:
Jack Briner
Department of Mathematics
University of North Carolina at Greensboro
Greensboro, NC 27412
Abstract
========
This dissertation studies the simulation of mixed-level, digital circuits on
parallel computers using the virtual time paradigm introduced by David
Jefferson. Virtual time is an appropriate mechanism for parallel simulation
of digital circuits because its asynchronous nature allows varying levels of
event intensity to be handled with minimal overhead. Other models, such as the
conservative models of Chandy and Misra, may require extensive resources to
handle idle portions of the circuit. A model of parallelism, using a greedy
deterministic processor scheduling algorithm, demonstrates that circuits have
significantly more parallelism than previously thought, provided asynchronous
simulation is used.
Previous implementations of virtual time simulation have used periodic
breakpointing, periodic calculation of global virtual time, and fossil
collection on demand. Such maintenance computations are performed better if
done regularly, reducing overhead and balancing the simulation. State, if
saved incrementally on each event, keeps the natural, demand-driven computation
of discrete event simulation. Global virtual time can be calculated frequently,
using a new tree-based algorithm. With a close bound on the global virtual
time, fossils can be collected regularly, reducing memory usage and relieving
the need for complicated state management operations. By keeping all processors
within a specified bounding window of the global virtual time, processors are
not allowed to get far out of synchronization, reducing rollback costs.
Existing models of partitioning are inadequate for virtual time simulation. For
transistor level model evaluations, random partitioning performs satisfactorily
because the ratio of computation to communication is high. For gate level
modelling, where event handling is more frequent, partitioning must reduce
communication costs and also must promote concurrent model evaluations.
However, such partitioning algorithms may be expensive.
An implementation of virtual time simulation, known as PLDVSIM, is measured on
BBN's GP1000 parallel processing system. PLDVSIM is functionally equivalent to
LDVSIM, allowing existing mixed-level simulations to be used without
modification. Circuits from 600 to 32,000 gates are simulated in parallel and
compared against the tuned, sequential algorithm. Speedups in the range of 10
to 25 are achieved with 32 processors.
<Please use US MAIL for requests! I will deny requests by email!>
------------------------------
Date: Mon, 1 Oct 90 16:09:41 -0500
From: Ram Reddy <ramreddy@ux1.cso.uiuc.edu>
To: simulation@bikini.cis.ufl.edu
Subject: Request information about PC Simulation packages
I am trying to simulate the working of a Local Area Network and am looking
for application software to do this. Are there any simulation packages for
the PC that allow you to model the nodes as objects with attributes that
can be changed on the fly ? I mean the nodes take jobs from one probability
distribution vs another ? I would appreciate any help.
Thanks, Ram Reddy.
------------------------------
END OF SIMULATION DIGEST
************************