poulsv@daimi.dk (Poul Svendsen) (09/13/89)
We are about to initiate a project concerning the implementation of a functional languages on a transputer network. We would then like to hear anything about work done in this area, references etc. Please email to poulsv@daimi.dk, and I will summarize to the net.
poulsv@daimi.dk (Poul Svendsen) (10/11/89)
Functional languages on transputers.
Thank you for all the answers, comments and references we got.
Below is a summary (... means delete text )
===========================================================================
From: geoff <geoff@gec-rl-hrc.co.uk>
At GEC's Hirst Research Centre we have been working on the
parallel implementation of lazy functional languages as part on
ESPRIT Project 415 (Parallel Architectures and Languages for AIP
- A VLSI-Directed Approach). Our current status is that we have
implemented a simple functional language on a simple transputer
network, and can demonstrate executing code. I will now give a
brief description of some of the important points of our work.
...
We compile programs that have been annotated with the evaluation
transformers determined by the abstract interpretation into
parallel G-machine code.
The parallel G-machine code is then macro-expanded into
transputer machine code.
...
A number of people have worked on the project over the years :
David Bevan, Geoffrey Burn, Rajiv Karia, Hugh Kingdon,
David Lester and John Robson.
Burn, Kingdon and Lester are currently working on the project.
As with all good things, the project is coming to an end at the
end of October.
If you would like to contact us, the three people above are
currently working at:
GEC Hirst Research Centre
East Lane
Wembley
Middx HA9 7PP
UK
================================================================
From: cspw.rucs01@f4.n494.z5.fidonet.org (cspw rucs01)
I have a small functional language called RUFL (Rhodes University
Functional Language) which is a lot like the language in Bird and
Wadler's text book. My compiler produces fixed-machine FLFM code
which is an extension of SECD code. (See Takeichi, New Generation
Computing 5(1988), 377-391.)
The current status of the project is that I generate native
code modules for an IBM PC, (which can then be linked with LINK),
or can generate FLFM code which I then interpret. Both the
compiler and the interpreter are written in C, and have been
ported to run under Helios on the transputer (but only in
single-processor mode at this stage).
...
Requests to
Department of Computer Science
Rhodes University
Grahamstown 6140
South Africa.
=========================================================
From: "Lyndon While" <while%inmos-c%uunet@unido.UUCP>
...
An IED project just starting at Southampton University and Imperial
College. Hugh Glaser is leading the Southampton part and is probably
the best person to contact; his address is hg@ecs.soton.ac.uk
You could also contact Mike Reeve at Imperial College on mjr@doc.ic.ac.uk
He is also connected with the above project.
==========================================================
From: Paul Kelly <phjk@doc.ic.ac.uk>
The functional programming section at Imperial College, London, in
collaboration with the Department of Computer Science at
Southampton University, have a couple of projects concerning transputers:
(1) FAST:
M.J. Reeve, S.M. Cox (Imperial), H. Glaser (Southampton)
This project aims to produce a high-quality sequential compiler for
the language Haskell, targeted on the transputer, and will employ
the declarative configuration language described in [1] chapter 5,
and perhaps other approaches, to control process distribution,
partitioning and interconnection.
(2) Transformation for Parallelism.
J. Darlington, A.J. Field, P.G. Harrison and P.H.J. Kelly.
This project is based on the Imperial College transformation-based
programming environment, together with the associated theoretical
work funded under the FLAGSHIP project. The objective is to use
the machine-supported transformation of functional programs to
address the problem of deriving parallel variants of an algorithm
to suit differing target parallel architectures.
...
Paul Kelly
Dept of Computing, Imperial College, 180 Queen's Gate, London SW7 2BZ,
UK.
+44 1 589 5111 x.4993
[1] Paul H.J. Kelly "Functional Programming for Loosely-coupled
Multiprocessors". Pitman/MIT Press 1989.
===================================================================
From: markv@acm.Princeton.EDU (Mark VandeWettering)
Organization: Princeton University, Princeton NJ
...
Other places to check are the conference proceedings
from the ACM's SIGPLAN "Lisp and Functional Programming Conference".
Springer-Verlag also publishes proceedings for smaller conferences on
Computer Architectures and Functional Programming, and graph reduction.
====================================================================
We hope that this will be a help to someone.
poulsv@daimi.dk