hjm@cernvax.UUCP (hjm) (07/01/88)
I'm intending to purchase a Transputer development system in the not too distant future, and I would appreciate the advice of those who are already in the field. What I require is a PC or AT add-on board, an occam-2 compiler, a C compiler and an assembler. The board need only have one T4 on it, but with a reasonable amount of memory (1 MB would be nice). Does anyone have any recommendations for or against any particular products? There seem to be about 6 C compilers available from: Pentasoft, INMOS, Renishaw, 3L, Parsec and Norcroft. Again, comments? Thanks, Hubert Matthews
nick@lfcs.ed.ac.uk (Nick Rothwell) (08/29/88)
In article <8808261949.AA00805@uk.ac.ox.prg> PVR%bgerug51.earn@NSS.CS.UCL.AC.UK (Patrick Van Renterghem / Transputer Lab) writes: >Dear mailing list, > >I am writing a paper for the "largest transputer event on the continent". >There is a small section on development systems and I don't want it to >contain wrong information. So, I have posted it to the net and am awaiting >replies from you. If anything is untrue or missing let me know. > Not many people are > using the other languages, such as ML (Lisp dialect from > Edinburgh, transputer implementation by Jon Kerridge, contact > ml@lfcs.ed.ac.uk), Prolog (Microway), ... so they are probably > still full of bugs. For the record, I did the Transputer ML port, with some assistance from people at 3L. ML is *not* a Lisp dialect, it's a statically-scoped polymorphic language used for (amongst other things) theorem proving and logic systems, and large-scale software prototyping. And why the implication that not much usage -> full of bugs? Transputer people seem to be mostly interested in number crunching applications and Mandlebrot programs, hence the Occam and FORTRAN. C is universal, hence C compilers. ML isn't a number crunching language, so there isn't necessarily the same appeal to transputer costumers. Even so, I've received a large number of requests for all over the world for our ML system. If our ML were "full of bugs", then we wouldn't be sending out several systems a week, would we? I did the transputer port as an exercise to see how fast the transputer was at running a complex, conventional, C program (the abstract code interpreter). It means that people with PC's and transputer boards can have our ML system (a bare PC is too brain-dead to run something as big as ML). This is the text of our announcement about the ML release for Transputers. Standard ML for Transputer Systems ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ We are now distributing a version of the Edinburgh ML system which will run on a transputer board. The distribution consists of the source code of the ML interpreter (written in C), and an image of the ML compiler for the interpreter to run. The transputer ML system is simply our standard ML release ported to 3L's C compiler for the transputer. The ML system runs as a stand-alone application on a single transputer, and doesn't make use of the transputer's multi-tasking or networking capabilities. It requires the afserver program to provide terminal I/O and file access facilities. To run serious ML applications, the transputer should have access to at least 2Mbytes of memory. Currently, we cannot distribute the system as a transputer object binary on 5" disks, although this situation may change in the next month or two. We send out the ML kit (ML image plus interpreter sources) either on 1600bpi magtape or 1/4" tape cartridge. You will need 3L's C compiler, plus some means of transferring files to an IBM PC, to build the system. Our charge for the distribution is 50 pounds sterling, to cover administration and media/postage costs. In addition, we ask that you fill in and return a licence form, so that we can keep records of the sites running ML and the work being done with the distribution. This is to avoid any wasted or duplicated effort if any clients want to modify the system for their particular environment. We will be looking into the possibility of providing the interpreter as a transputer binary file (so that you don't need 3L's C compiler, just the afserver, which is Public Domain), and of providing the distribution on native media (so that you don't need to be able to read magtapes or cartridges). >Patrick Van Renterghem, Seminar Coordinator "Transputers for Ind. Appl." -- Nick Rothwell, Laboratory for Foundations of Computer Science, Edinburgh. nick@lfcs.ed.ac.uk <Atlantic Ocean>!mcvax!ukc!lfcs!nick ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ...while the builders of the cages sleep with bullets, bars and stone, they do not see your road to freedom that you build with flesh and bone.