MRWS MICROSYSTEMS MUSEUM NOTES
The MicroComputer Club of Melbourne (MICOM) was the first Computer Based Bulletin Board to go live in the southern hemisphere, initiated by a group with Sysop Peter Jetson and Phillip Freidin to the fore. This occurred during the few years of the early 1980's when I briefly led MICOM as president.
Philip - the previous President- also led the Munchkins, a group dedicated to a variable very long word microcode bit slice architecture machine based on the AMD2900 series bitslice architecture chips. It was called the AFC3200 and was set up in the phase-zero design with a Zilog Z80 microcode set permitting the PAL ASM microassember to be used to assemble fresh microcode. It significantly predated the Orion, which later followed the same AMD2900 bit-slice basis as used for the AFC3200, and went into production in the UK. A fascinating Cambridge Phd (Curzon,1991) describes issues in verifying this class of machine and an Orion in particular, and Bill Clocksin designed an incremental compiler oriented virtual machine (ZIP) for a Prolog (-X) designed to implemented directly in microcode an Orion. Its a pity that this progress from PERQ to AFC3200 to ORION is not better known. Later in the 80‘s Philip was hired by AMD to design the next generation AMD bitslice chip., and was asked by AMD to bring the AFC3200 with him.. although he prtested that it was already outofdate by then
Most of the Melbourne House and Olympic computer gaming companies were then active MICOM members. Every Saturday a large lecture theatre in Burwood (a middle suburb of Melbourne Australia) would fill with a mix of older scientists, young bit warriors and other across the spectrum to the political activist Albert Langer (of the Langer option for voting the way you actually wanted to in a single transferable voting system such as Australias’) to Principal Scientists from CSIRO and University lecturers met to hear about Prolog compilers, bit slice machines, communications... and this was the only venue available, as Universities had simply not yet come to terms with the micro revolution then brewing and got down to teaching computing or information systems....and MICOM had many hundreds of members (I must check my files, I think it went as high as 1500 at one point), many of whom were hungry for knowledge and building their own computers.
This was an era of community based expertise only now likely to be matched again (by the Open Source, Open Data, and crowd sourcing movements .. and after only 30 years even such bodies as the World Bank and the UK Ordnance Survey are at last moving in this direction).
MICOM still exists nearly 30 years later, and meets on the third Saturday of the month at Wadham House by Mt Waverley Railway Station.
Soon the Apple group split off to become AUSOM as did several other groups which have not survived the century, leaving Micom with any of the core founders even 20 years later: their same machines having aged from the bleeding edge to history over the same years. Of course I share all this with the rest.....
The Mainframe era
I have been involved in computer usage since 1961, with the Ferranti Mercury at Oxford University using Autocode. In the 1960s my first wife, Jane (nee Geiringer), and her team, including Susan Mathews, created defence programs at Bracknell for Ferranti FM1600B computer system, which were still in use in various Navys from the 1970s and 1980s to the late 1990s.... some computers are very persistent: perhaps the design of the FM1600B by the Hon Christopher Baily (a classicist) had something to do with it.
At Harwell in the mid 1960s I used the IBM 7030 Stretch system at AWRE Aldermaston (feeding it via card created on an IBM 1620 at AERE). It ran well with the S2 Fortran Compiler written at AERE, which turned the Stretch from a disappointment to a seriously fast machine for the time. Mike Powell allowed me to use the very first version of Powell Fletcher Davidon optimisation code written by him at AERE and one of the first codes in the standard Harwell Fortran Library for my work. Such are the benefits of a critical mass high quality research centre... the S2 (later S3, both written by Chris Pyle at AERE Harwell) compiler had variable array dimensions settable at run time after compiling into binary.. back in 1965. The 1620 required punched cards to be created as input, so the data from each run had to be manually punched and supplied with the binary pack of the compiled version of the program for the Stretch. Not easy after 24-36-48 hours non stop data collection using the Harwell syncrocyclotron - or even longer as the nuclear scattering experiments in which I was engaged usually ran into weeks.
Pehaps it is now (tolerably) safe to say that i did my flying with the RAFVR on Chipmunk T Mk10's at RAF Bicester with the Oxford University Air Squadron AFTER such a series of nonstop days and nights. My RAFVR discharge certificate stated that “Wigan, of exemplary conduct, maintained a low average standard of flying throughout”. Hardly surprising: of course my standard at the two week camps was far far higher, but fortunately nobody asked me why at the time.
The deck of binary punched cards were dropped at a critical time just before my thesis was due, and had to be manually reassembled card by unmarked card, by matching the holes. After a day of hard work all 432 blank binary punched cards (yes i remember them well) were once again in oder, and worked on the 1620 in time for the final analyses required just two days before my thesis had to be completed and submitted.
Here is a recent picture of the 1967 original punched card program deck (in Fortran IV), ready to be complied into a binary deck of compiled code for the IBM 7030 at Aldermaston under the S3 complier (with dynamic arrays).. I am still attached to the British Computer Society Fortran Special Interest Group... and the Fortran language is still used when accuracy and numerical precision is of major concern.... Alick Glennie’s S3 Fortran, with its dynamic allocation of arrays without recompilation (using PRELUDE) transformed the IBM 7030 Stretch into a successful machine (which it certainly was not before he wrote it) - I used i heavily in my DPhil work.
The data was created by an entire room of tunnel diode based early Camac modules arranged to carry out high speed logic sing zero crossing of the signals from the scintillation counters (which were receiving 10**18 events a second) , and to output the counts that we selected using this logic to a printout. This was then prepared for the IBM1620 for preprocessing and then physically carried to Aldermaston to the IBM 7030.This placed a great deal of pressure on accurate coding..
My most memorable encounter with a mainframe era computer was after three years of work with the IBM 1620/7030 systems, with the English Electric KDF9 at Nuffield College, Oxford. My first wife, Jane, was there as Research Assistant to Josephine Klein and Denis Munby in the era when the famous economist Marty Feldstein was on the staff, as was our contemporary and Somerville friendship group historical economics, Sir Roderick Floud (both Rod and I did ground breaking work on family expenditures afterwards, Rod on historical data myself on Australian Family Expenditure data from the very first FES carried out by the Australian Bureau of Statistics))
I had allocated the month of September 1967 to write my DPhil thesis.. but at the beginning of the month was offered 3 weeks on the Harwell synchrocyclotron - so of course I took it and did another major experiment: what scientist would not do the same? As a result I had exactly one day left to write the thesis and hand it in: remember that in 1967 IBM had not even invented ‘word processing’: that was left until 1969.
I think I can make a genuine claim to have created the first word-processed Doctoral thesis, undoubtedly the first at Oxford, in 1967, using a KDF9 MainFrame, a Singer Freiden Flexowriter, and an early printer at Nuffield College Oxford.....
The Micro era
From late 1970's, when computers came within the grasp of individuals, I realised that the control over cumulative information was going to become a personal rather than a solely corporate issue. This indeed has occurred, and knowledge capital has become critically dependent on computers for generation, application, accumulation and communication. The second round effects of personal data space and the shift of international and global corporate competition towards securing intellectual property has followed equally inevitably.\. However the means - computer mediated communications - have also allowed individual global access to niche markets and accelerated the innovation cycle. The seeds of this later revolution were set in the bulletin board systems of the early 1980s as the first harbingers of the individual accessed personal networks. The Internet has only comparatively recently taken over this role as well.
One of the first comprehensive collection of basic benchmarks for early 1980’s mainframes, minis and micros is given by my 1982 paper to the ACS MICSIG In Canberra: a downloadable version (here).
Updated on 11 November 2015