An Independent Review of the Transputer Initiative that was published as part of the Proceedings of the Closing Symposium at the University of Reading in March 1992 (IOS Press).
The SERC/DTI Transputer Initiative was designed to introduce the new parallel technology based on the transputer rapidly both to industry and academia. The Regional Centres supported by the Initiative provided centres of expertise and low cost facilities for industry. The Academic Loan Pool enabled permanent members of Higher Education Institutions to acquire, on a short term loan basis, representative examples of commercially available transputer hardware and software. This paper reviews the aims, methods, and achievements of the Academic Loan Pool of the SERC/DTI Transputer Initiative. The conclusions generally confirm the reaction of one participant: I think it was a terrific success. Two papers for an 8 month loan of £2000 worth of kit followed by 3 research projects, value in total at about £222,000 has got to be good value. It also stimulated inter disciplinary work.
The development in the mid 1980's of the transputer offered the UK a unique opportunity to exploit parallel programming technology. Although other, more prominent approaches to parallelism were very active, the transputer was unique because it was designed specifically to give state of the art performance for a new parallel programming language, OCCAM (a trademark of the INMOS group of companies). Occam was in turn designed specifically to support a well researched model of parallel computation (based on Hoare's CSP model[1]). Further, the new technology offered, through parallelism and fast link technology, extensible performance.
These benefits came, however, at a cost. To obtain them it was necessary for system designers to absorb the radically new, and not always comfortable, methodologies associated with parallel programming. Although the development of compilers for conventional languages such as C and Fortran provided alternatives to the radically new occam language, users had to recast their solutions in terms of communicating sequential processes. It was clear that without a significant effort to help potential users through the learning curve, the UK technology had little chance of being widely exploited. It was also clear that to be effective, both industry and academia should be targetted.
In October 1986, a working party of SERC's Computing Facilities Committee recommended that a 3 year programme entitled Engineering Applications of Transputers be established for the joint benefit of the UK Academic Community and UK industry. The programme, which was part funded by DTI, was approved in several stages and, in its final form, will last 5 years. It began operation in April 1987, and is due to end in March 1992. The expected spend over the 5 year period is shown below.
SERC/DTI TRANSPUTER INITIATIVE ACTUAL SPEND |
Total (£K) |
---|---|
Capital: Loan Pool | 500 |
Recurrent: Loan Pool | 150 |
Major Capital: Regional Centres | 750 |
Recurrent Centres | 500 |
Software Library | 200 |
Manpower: Coordination | 615 |
Development Contracts: | 285 |
TOTALS | 3000 |
The expected spend on the Loan Pool is less than originally proposed, both in absolute terms and in relative terms.
The objectives of the Loan Pool, and arrangements for running it, were set out in a memo dated April 1987 and widely circulated to the academic community. The Loan Pool began operation in mid 1987 and closed to new applications on 15th March 1991.
The operation of the Loan Pool had three major objectives:
The Transputer Initiative provided resources for establishing a Loan Pool of commercial transputer products and for distributing this pool to research groups at Higher Education Institutions on a short-term loan basis. It is important to note that a successful application to the Loan Pool provided hardware and software only: no manpower was provided by the Initiative. |
A pool of commercial transputer products, consisting of hardware and software, was established. This contained a mixture of pure occam systems and systems supporting conventional languages such as C and Fortran. Hardware consisted initially of Meiko, INMOS and Transtech products, and a number of PC's to act as hosts. During the Initiative a small number of other systems, such as the Helios operating system, were acquired. Some items were generously contributed for support and for direct use in the Loan Pool by manufacturers including Acorn, Apollo, Gnome and Pacific Parallel. Annexe A contains a list of Loan Pool equipment which was compiled during November 1991.
The Loan Pool was operated by the co-ordination team at Rutherford Appleton Laboratory (RAL). The effort required to run the Loan Pool ranged from 0.5 my/y to nearly 1 my/y. Much of the day to day management of equipment was time consuming, tedious and at times frustrating. Delays in delivery from manufacturers often led to delays in shipping a collection of Loan Pool equipment to a user site. It is a tribute to the efforts of Fran Childs and (from 1989) Linda Reed that around 80% of the Loan Pool equipment was, through the lifetime of the Loan Pool, with users rather than idling either at RAL or at the manufacturer.
A short application form (2 sides of A4) was designed and widely distributed to UK higher education institutions. Applicants could request the loan of appropriate equipment for a period of up to four months. Successful applicants could request at most one extension of the loan for a further period of up to eight months. Note that the only resource available from the Loan Pool was equipment and or software: requests for manpower, travel and other costs were not possible. Hence Loan Pool applicants had to be enthusiasts, willing to put their own time and effort into exploring the new technology.
Applications were initially processed by the Transputer Applications Management Group (TAMG). Subsequently, in order to ensure rapid processing, applications were dealt with by the coordinator using postal or E-mail advice from TAMG members. The objectives of the Loan Pool (A, B and C above) were taken as the main guideline in judging applications. In general, most of the initial demand came from groups who were unfamiliar with transputer technology and who wanted to explore its potential for future research. Such applicants were required to attend an appropriate training course at the Initiative's expense before the loan was provided.
Few applicants applied directly to develop software.
Successful applicants were required to observe a number of conditions, including the following:
The table below shows the total number of Loan Pool applications over the lifetime of the Initiative, together with figures for extension applications. Notice that out of 165 successful applicants, 106 requested extensions, suggesting that most applicants found the technology sufficiently interesting to warrant further effort.
Initial | Extension | |
---|---|---|
Number of applications | 220 | 106 |
Number of successful applications | 165 | 87 |
Success Rate | 75% | 82% |
Many applications were rejected because they were from groups who were already well provided with the new technology. Others were rejected for lack of a clear programme of work.
Early in 1991 the coordination team carried out a survey of user reactions to the Loan Pool. Questionnaires were sent to all successful applicants whose loans finished before March 1990: that is, all those whose loans had finished at least one year earlier. 96 questionnaires were sent out, and 49 responses were received. The responses received are analysed below. Some replies contained unmarked responses: these are counted in the ? column below.
QUESTION | Yes % | No % | ? % |
---|---|---|---|
Are you still using transputers? | 88 | 12 | 0 |
Was the loan beneficial to your research project? | 96 | 4 | 0 |
Do you believe that the Loan Pool concept is a useful one | 96 | 2 | 2 |
Did the loan lead to a transputer based grant application? | 67 | 33 | 0 |
Did the loan lead to an equipment purchase? | 78 | 16 | 6 |
Did the loan lead to any publications, conference papers or theses? | 67 | 4 | 29 |
Several loans produced more than one publication: about 54 publications were mentioned in the replies received. In terms of generating transputer based grant applications, which was one of the primary objectives of the Academic Loan Pool, the survey results suggest that about two thirds of the loans achieved this objective. From comments made in the replies received, a significant number of these grant applications either succeeded, or were alpha rated.
Many positive general comments were received. A typical comment, indicative of the success of the Loan Pool, was I believe this was a good Initiative and certainly brought transputers into the academic community far quicker than would have been possible otherwise..
From the replies received, some people found that the rapidly decreasing price/ performance ratio of high powered conventional workstations satisfied their requirements better than the newer parallel technology. It seems likely that many of those who did not respond to the questionnaire fell into this category. Even this sort of result is a worthwhile outcome of a loan, for it prevents potentially large grants being awarded to work with inappropriate equipment.
A secondary goal of the Loan Pool was "To encourage software developments of general community interest and relevance. To this end, it was a normal condition of a loan that any resulting software would be made freely available to other users via the Transputer Initiative's Software Exchange Library. This Library was set up and maintained at one of the Transputer Initiative's Regional Centres, initially at Sheffield and later at Liverpool. It was operational by 1988, although at that time it had little software to distribute. The main sources of software were transputer industries and academia.
The degree to which this objective was achieved can be judged from the following extract from the Software Exchange Library, which shows all the software items contributed by the Academic Loan Pool. Column 2 shows the library reference number, and column 3 gives the number of the loan which contributed the item.
TITLE OF SOFTWARE ITEM | CODE NUMBER | LOAN NUMBER |
---|---|---|
Neutron Scattering Detector Simulation | A2 | 039 |
Fluid Dynamics Algorithms | A3 | 028 |
Neural Net Simulation | A4 | 033 |
Mandelbrot programs | A5 | 014 |
Stripcode programs | A6 | 014 |
Digital Filter Algorithms | A8 | 054 |
Finite Element Codes in occam | A9 | 043 |
Target Tracking Algorithms | A10 | 027 |
Model Reduction Using Con. Fractions | A11 | 083 |
Rank Annihilation Processor Ring | A12 | 083 |
Frequency Response Pipeline | A13 | 083 |
Ordinary differential equation Solver | A14 | 020 |
Finite Element Model of Soil Water now | A15 | 064 |
Image Analysis Algorithm | A17 | 056 |
General Relativity | A18 | 128 |
Image Processing/ Mandel | A19 | 101 |
FFT Library | A20 | 095 |
Tridiagonal Block Solver | A21 | 087 |
Symmetric Matrix Problem Solving | A22 | 106 |
Process Simulation | A23 | 097 |
Kron's Method of Tearing | A28 | 053 |
Load Balancing Evaluation | U6 | 044 |
Sieve/ Whetstone Benchmarks | U7 | 055 |
About 17% of completed loans generated contributions to the Software Exchange Library. As there are only 21 other items listed in the library, Loan Pool contributions, at 23, account for over 50%. Of course simply appearing in a library list is no guarantee of usefulness, and there is no demand at all for some of the more obscure contributions to the library. However, several of the Loan Pool contributions have proved very popular and one (the FFT library, code no. A20) is in fact the most popular item in the whole library.
An important condition of a loan was the delivery of a final report. The extent to which this condition will be satisfied at the end of the Initiative is not yet clear: out of some 165 loans, 45 are active at the time of writing. Of the remaining 120, some 79 have produced final reports. This leaves 41 final reports which are now overdue. Some of these (about 10) are due to abnormal causes, such as personnel changing jobs.
It seems likely that the remaining 31 represent cases where conventional technology proved more attractive, either because of the availability of more expensive workstations to the investigator, or because of the learning barrier associated with parallel programming on the transputer.
The bulk of the reports received to date are available as 9 volumes listed in Annexe B, and it is these volumes which form the basis of the analysis in the rest of this section. They contained reports for 71 loans. All the statistics given in the remainder of this section are based on this population.
A crude classification of the loans was performed and the results are shown below, ranked in order of frequency. One striking fact is that the first 5 topics account for over 70%. The relatively low figures for the Database and Neural Net categories surprised the author.
Application type | No. reports | Reports % |
---|---|---|
Numerical Analysis | 15 | 21 |
Signal Processing | 13 | 18 |
Generic | 9 | 13 |
Simulation | 8 | 11 |
Graphics | 5 | 7 |
Image Analysis | 4 | 6 |
Robotics | 4 | 6 |
Neural Net | 3 | 4 |
Combinatorial Optimization | 2 | 3 |
Computer Aided Design | 2 | 3 |
Database | 2 | 3 |
Education | 2 | 3 |
Fractals | 1 | 1 |
Music Synthesis | 1 | 1 |
In this section we review the achievements reported under some of the more common applications categories, using the available Loan Pool final reports.
About two thirds of the reports described work on either Finite Element or Computational Fluid Dynamics problems. Remaining topics included Monte Carlo techniques, Ordinary Differential Equations, and Linear Algebra.
The majority of loans in this area were awarded to departments of Engineering, Applied Maths or Physics, and the investigators were unfamiliar with the new technology. After an initial learning phase, over two thirds of the Numerical Analysis loans resulted in demonstrated speedups and about half of these resulted in publications. There was some excellent work here, which clearly involved considerable manpower input from the institutions concerned and in some cases showed deep awareness of the issues involved in exploiting parallelism. Two examples of the work done with Loan Pool equipment in the numerical area are described in [3,4].
Nearly all the loans under the Signal Processing heading involved Control and Instrumentation applications. It seems natural to examine parallelism as a way of improving the performance of real time control and instrumentation subsystems. However, as some of the investigators discovered, achieving dramatic performance increases in this area is hard and with a naive approach it is not difficult to experience slowdowns.
Most of the loans were awarded to Engineering departments and initial loans were devoted to familiarisation. About one third of the loans resulted in clear performance benefits and about the same number of loans led to published papers such as [5,6].
This category includes those working on tools such as compilers for dynamic languages, techniques and theories for analysing and exploiting parallelism. Not surprisingly, Computer Science departments accounted for most of the loans under this heading.
The general effect of the Loan in this area was to make people look rather hard at the occam language and its model of concurrency in the context of dynamic parallelism. Some interesting and sometimes provocative critiques of the occam language, the transputer and its operating systems [7] emerged from this work, together with theses on load balancing such as [8].
Most of the reports in this area came from departments of Chemical, Electrical or Process Engineering, with a few from Applied Maths/Physics departments. Work with Loan Pool equipment tended to involve taking a known published technique such as [9], and exploring its implementation on transputers. Most loans resulted in speedups, given enough effort. However, the difficulty of using the occam/transputer methodology was frequently noted, a fact that was frequently communicated to the manufacturers by the Initiative.
Computer Science Departments benefitted from most of the loans in the graphics and image processing areas. Good speedups were obtained in many cases. However, from the final reports available, there was not much evidence of publications emerging from the graphics/image analysis loans.
Four final reports were on robotics applications. Two had generated publications (10,11] and one final report was the length of a short book.
One particularly interesting loan report, in the area of neural network/simulation, described the use of transputers to model the behaviour of nerve circuits in the human retina. Following initial work with a single transputer, successful runs were reported firstly on an array of 16 T414 transputers, and secondly with a much larger (1024) array of T212 transputers.
The primary objective of the Loan Pool was To assist in the preparation of high quality Research Grant applications involving transputers. With few exceptions, final reports received showed that at least one funding application resulted from the loan and further reported either success or unfunded alpha rating.
The direct generation of good science was not a primary objective of the Loan Pool. However, from the final reports received it is clear that the Loan Pool stimulated some excellent work. Two objective criteria of quality are:
The table below shows the results of grading the final reports according to these objective criteria. Grades 4 and 5 correspond directly to meeting one or both of the criteria and indicate good science. Grade 3 indicates substantial progress in absorbing parallel programming techniques for the transputer. Grade 2 indicates that a reasonable familiarity with the new technology was achieved. Grade 1 indicates a poor attempt to exploit the loan equipment. Only publications reported in the available final reports were counted. Thus publications arising after the submission of these reports are not included, and the table below will undoubtedly undervalue the scientific output of many loans.
Grade | Number achieving grade |
Interpretation of grade | % |
---|---|---|---|
5 | 10 | Speedups demonstrated AND publication of results | 14 |
4 | 21 | Speedups demonstrated OR publication of results | 30 |
3 | 27 | Promising single transputer or theory results | 38 |
2 | 7 | Familiarisation with transputer technology | 10 |
1 | 6 | Poor attempt LO exploit loan | 8 |
Over its lifetime, the Academic Loan Pool spent approximately £650K. From this must be subtracted the book value of the equipment at the end of the Initiative, which is approximately £200K. Hence the cost of the Loan Pool was approximately £450K and this expenditure generated 165 loans, at an average cost per loan of approx. £2.7K.
As reported in an earlier section, some 10 of the 165 loans ended because, for example, the investigator moved to another institution or abroad. Of the loans which have completed and submitted final reports, 71 have been graded in the previous section. The second column of the table below shows the result, with grades 4 and 5 grouped under the heading good science. 45 loans are still active, and 8 have recently submitted final reports. Column 3 in the table shows the result of grading these 53 ungraded loans in the proportions shown in the second column. This leaves 31 completed loans which should have submitted final reports. Column 4 in the table assigns all 31 to the lowest level of achievement. Column 5 sums the previous columns, to give expected levels of achievement when the 155 normal loans have all completed.
Column 6 shows the cost of achieving a given level or better. For example, the total number of loans which achieved promising or good science was 101. If the whole cost of the Loan Pool is borne by these 101 loans, the cost per loan is £4.4K.
Level of loan achievement |
Graded Final Reports |
Est Remaining Reports |
Loans with no reports |
Expected final number |
Cost per loan achieving this level or better |
---|---|---|---|---|---|
Good Science | 31 | 23 | 0 | 54 (35%) | £8,312 |
Promising | 27 | 20 | 0 | 47 (30%) | £4,442 |
Familiarisation | 7 | 5 | 0 | 12 ( 8%) | £3,964 |
Technology rejected | 6 | 4 | 31 | 41 (27%) | £2,903 |
TOTALS | 71 | 53 | 31 | 155 |
All these costs represent remarkably good value at their respective levels of achievement. The cost at the Good Science level appears extremely good value. However, the costing does not of course include the time of the investigator, which - as far as the Initiative is concerned - is provided free by the institution. Nor does it include the overheads.
The SERC/DTI Transputer Initiative was designed to introduce the new parallel technology based transputer rapidly both to industry and academia.
A number of regional centres supported by the Initiative provided centres of expertise and low cost facilities for industry.
A centrally operated Academic Loan Pool enabled permanent members of Higher Education Institutions to acquire, on a short term loan basis, representative examples of commercially available transputer hardware and software.
This paper has reviewed the aims, methods and achievements of the Academic Loan Pool of the SERC/DTI Transputer Initiative. The conclusions are summarised below:
Dr. Raymond Fawcett, who ran the Loan Pool for much of the Transputer Initiative, provided me with key documents from the files, and detailed answers to all my questions. The quantitative analysis of the achievements presented here would not have been possible without his assistance.
The Transputer Initiative owes much to the drive and enthusiasm of Dr. Mike Jane, who also provided essential information.
[1] C. A. R. Hoare, Communicating Sequential Processes, Prentice-Hall, Englewood Cliffs, NJ, 1985.
[2] Inmos Ltd., occam 2 Reference Manual, Prentice-Hall, Hemel-Hempstead, UK, 1988
[3] J. S. R. Alves Filho and D. R. J. Owen, Using Transputers in Finite Element Calculations: A First Approach, Proc. 7th Occam Users Group Technical Meeting
[4] R. K. Cooper, Successive Over-relaxation on a Transputer Network, CONPAR 88
[5] E. Gul and D. P. Atherton, Some Preliminary Investigations in Target Tracking Using Transputers, IEE Colloquium on Recent Advances in Parallel Processing for Control, Bangor July 1988
[1] C. A. R. Hoare, Communicating Sequential Processes, Prentice-Hall, Englewood Cliffs, NJ, 1985.
[6] S. Gudvangen and G.J.Holt, Evaluation of Fast Fourier and Hartley Transforms on a Loosely Coupled Multiprocessor based on the Inmos Transputer, IEE Colloquium on The transputer in Signal Processing, March 1990
[7] P. Netherwood, HELIOS, Parallelogram Jan. 1989
[8] G. S. Liong, Load Balancing Evaluation for Transputer Based Systems, MSc thesis, Stirling University 1988
[9] R. J. Best, 1989, I Chem E Symposium.Ser.I 14, 157.
[10] D. T. Pham, H.Hu and LPotc, A Transputer-based system for locating parts and controlling an industrial robot, ROBOTICA 1989
[11] F. Naghdy, P. Strickland, Distributed Manipulator Environment: A Transputer-Based Distributed Robotic System, Jnt.J.Computer Integrated Manufacture, 1989
The table below shows the items of hardware and software available towards the end of the Initiative. Meiko, INMOS and Transtech products were available throughout the life of the Loan Pool.
Item | Quantity |
---|---|
Inmos Items | |
B004 single T board | 19 |
B008 motherboard | 15 |
B009 DSP board | 6 |
B404 2Mb TRAM | 32 |
D505-1 (occam toolset for Suns) | 6 |
D511 (C for Suns) | 6 |
D513 (Fortran for Suns) | 6 |
D700D (TDS) | 20 |
D703C software for B009 | 6 |
D705A (occam toolset) | 14 |
D705B (occam toolset upgrade) | 14 |
D7205A (occam toolset upgrade) | 7 |
D71 IC (C compiler) | 24 |
D71 ID (C compiler upgrade) | 14 |
D712B (Pascal compiler) | 24 |
D713C (Fortran compiler) | 24 |
D713D (Fortran upgrade) | 14 |
Transtech Items | |
TSB04 (like B004) | 10 |
TSB44-4 (T4 quad board) | 15 |
TSB48-4 (T8 quad board) | 5 |
TSB07 (frame grabber/graphics) | 7 |
TMB08 (like B008) | 20 |
TIM7 1Mb TRAM | 56 |
TIG-F (frame grabber TRAM) | 6 |
TIGI (graphics TRAM) | 6 |
TIG3 (high resolution graphics TRAM) | 6 |
TIMI! SCSI TRAM | 3 |
TIM14 D-A converter TRAM | 6 |
MCPI000 in Sun board (4x(T8+2Mb)) | 3 |
C and Fortran for MCPI000 | 3 |
Interface software for MCPI000 | 3 |
Meiko Items | |
MK200 in Sun board (4x(T8+2Mb)) | 3 |
CSTools software for MK200 | 3 |
3L Software Items | |
Parallel C ( v2. l.l) | 10 |
Parallel Pascal (v2.0.2) | 9 |
Parallel Fortran (v2.0. I) | 12 |
Parallel Fortran upgrade (v2. I .3) | 9 |
Thug debugger (vl.0.22) | 20 |
Perihelion Software Items | |
Helios | 6 |
Helios debugger | 6 |
Meiko Fortran for Helios | 6 |
Helios X windows | 6 |
Parasoft Software Items | |
Express | 2 |
NEC Hardware Items | |
Multisync II Monitors (for TSB07) | 4 |
Colour monitors (for TSB07) | 2 |
Tandon Hardware Items | |
286 PC's | 23 |
Apollo Items | |
DN 3500 Workstations + TRAMS | 4 |
Acom/Gnome Item | |
Archimedes 420 Workstation+ TRAM | 1 |
Pacific Parallel Item | |
Pac 8 motherboard for Apple Macintosh II | 1 |
The table below shows a list of the loan reports examined in compiling this review. Column 1 shows the volume number in which the report appears.
Vol | Loan No | Title | Investigator(s) | Institution |
---|---|---|---|---|
I | TR1/002 | EEG Monitoring using Transputers | M.D.Wilson | Hatfield Polytechnic |
I | TR1/005 | Dynamic Molecular Modelling | Dr. R.Bacon | Univ. of Surrey, Dept. of Physics |
I | TR1/013 | Transputer based Shape Analysis | M.S.Nixon | Univ. of Southampton, Dept. of Electronics & Computer Sci. |
I | TR1/014 | Dynamic Processes in Occam | S.J.Tumer and I.A.Horton | Univ. of Exeter, Dept. of Computer Sci. |
I | TR1/028A | Computational Fluid Dynamics on the Transputer | R.K.Cooper | Univ. of Belfast, (Queen's) Dept. of Aerospace Eng. |
I | TR1/028E | Successive Over-Relaxation on a Transputer | R.K.Cooper | Univ. of Belfast, (Queen's) Dept. of Aerospace Eng. |
I | TR1/029 | Design and Implementation of a Graphics Server to Allow Access to the Hercules Graphics Card on the PC Host | G.R.Brookes, A.J.Stewart, J.R.Vaughan | Univ. of Hull, Dept. of Computer Sci. |
I | TR1/033 | Neural Net Simulation using Transputer Nets | L.S.Smith | Univ. of Stirling, Dept. of Computer Sci. |
II | TR1/004 | Load Balancing Evaluation for Transputer based Systems | S.B.Jones, G.S.Liong | Univ. of Stirling, Dept. of Computer Sci. |
II | TR1/034 | Using Transputers in Finite Elements Calculations | J.S.R.Alves Filho and D.R.J.Owen | Univ. of Swansea, Dept. of Civil Eng. |
II | TR1/039 | Recent Experience with Transputer Based Processor Farms | J.M.Carter | Royal Holloway and Bedford |
II | TR1/047 | Use of Transputers in Process Simulation | C.C.Pantelides | ICSTM, Dept. of Chemical Eng. |
III | TR1/053 | Occam and the Transputer | K.Bowden | NE London Polytechnic |
III | TR1/053E | Kron's Method of Tearing | K.Bowden | NE London Polytechnic |
III | TR1/069 | The Numerical Analysis of Electromagnetic Fields | R.D.Woodcock | Univ. of Bradford |
III | TR1/074A | Parallel Scan Conversion Algorithms for Computer Graphics | H.E.Bez, L.Parks, I.Fielder | Loughborough Univ., Dept. of Computer Studies |
IV | TR1/012 | A: Real Time Simulation, Analysis and Control of Electrical Power Systems. B: Concurrent Digital Signal Processing |
M.Irving, A.Purvis | Univ. of Durham, School of Engineering and Applied Sci. |
IV | TR1/026A | Simulation of Nerve Networks | J.B.Willis | Univ. of. Southampton, Dept. of Maths |
IV | TR1/028B | Computational Fluid Dynamics on the Transputer | R.K.Cooper | Univ. of Belfast, (Queen's) Dept. of Aerospace Eng. |
IV | TR1/055 | Performance on a Transputer of an Operations Research Package | J.E.Galletly | Univ. of Buckingham, Dept. of Computer Sci. |
IV | TR1/056 | Parallel Algorithms for use in Image Analysis | I.East | Univ. of Buckingham |
IV | TR1/070 | Monte Carlo Simulation of Amphiphile and Solvent Mixture using Transputer Array | C.M.Care | Sheffield Polytechnic, Dept. of Applied Physics |
V | TR1/015 | Evaluations of Transputers and Occam | J.Herd | Univ. of Heriot Watt, Dept. of Electronic Eng. |
V | TR1/020E | Feasibility Evaluation of Transputer Equipment for Finite Element Software and Solution of ODEs | H.M.Liddell, S.Khaddaj | QMC, Dept. of Computer Sci. |
V | TR1/027 | Target Tracking Algorithms using Transputers | D.P.Atherton, E.Gul | Univ. of Sussex, Dept. of Eng. and Applied Sci. |
V | TR1/035 | Feasibility of Improving Database Performance | D.Nikodem | Univ. of Aberdeen, Dept. of Computing Sci. |
V | TR1/046 | Numerical Solutions of ODE's for Initial Value Problems | D.B.Clegg, A.N.Richmond | Liverpool Polytechnic |
V | TR1/049 | Systolic Designs for Lowpass Digital Image Filtering on a Transputer Network using TDS | S.A.Amin | Loughborough Univ., Dept. of Computer Studies |
V | TR1/054 | Application of the Transputer to the Aerospace Industry | R.Vepa | QMC, Dept. of Aerospace |
V | TR1/064 | A Transputer based Finite Element Model of Variably Saturated Soil-water Flow on Hillslopes | A.M.Binley, K.J.Beven | Univ. of Lancaster |
V | TR1/067 | Performance and Characteristics of Electro-hydraulic Servo-valves and other Metering Devices | D.C.Pountney | Liverpool Polytechnic |
V | TR1/077 | Transputer Instrumentation applied to Electrostatic Powder Flow Measurement | E.Mills, B.C.O'Neill | Trent Polytechnic, Dept. of Electronic Eng. |
V | TRI/080 | Profile of Multifrontal Algorithm | R.G.Miies | Wales Polytechnic |
V | TR1/083 | Hardware Accelerators for a CAD Control Package | G.M.Megson, S.D.OYoung | Univ. of Oxford |
VI | TR1/021i | Smalltalk on Transputers | E.Miranda, S.Cook | QMC, Dept. of Computer Sci. |
VI | TR1/021ii | Communication and Process Subdivision Overheads in Pipelines | K.Drake | QMC, Dept. of Computer Sci. |
VI | TR1/038 | A Transputer based System for the Manipulation of Multi-media Databases | P.J .Gartshore, J.A.Powell, I.A.Mayfield | Portsmouth Polytechnic |
VI | TR1/041 | Development of a Transputer Statistical Process Control Engine as the Heart of a Factory Network | D.C.Hodgson | Univ, of Birmingham, Dept. of Mechanical Eng. |
VI | TR1/042 | Investigation into the Potential for Applying Transputers to Power System Protection Relaying | R.K.Aggarwal | Univ, of Bath, Dept. of Electronic Eng. |
VI | TR1/043 | Pilot Study of Parallel Algorithms for the Finite Element and Discrete Vortex Methods | P.Bettess, M.J.Downie | Univ. of Newcastle, Dept. of Marine Technology |
VI | TR1/074E | Parallel Scan Conversion Algorithms for Computer Graphics | H.E.Bez, L.Parks, J.Fielder | Loughborough Univ., Dept. of Computer Studies |
VI | TR1/075E | A Parallel Solution of Some Groundwater Flow Problems | H.R.Thomas, C.L.W.Li | Univ. of Cardiff, Dept. of Eng. |
VI | TR1/090 | Several Short Reports by Various Users of Transputer Loan Equipment at Birmingham | J.Bowcock, N.Queen, G.McCauley, P.Borcherds, P.Bertrand | Univ. of Birmingham |
VII | TR1/037 | Concept and Control Strategy of a Flexible Robotic Assembly Cell (FRAC) | R.Seals, S.Ruocco, Mei Tao | Middlesex Polytechnic |
VII | TR1/052 | A Transputer-based System for Locating Paris and Controlling an Industrial Robot | D.T.Pham, Huosheng Hu, J.Pote | Univ. of Cardiff, Dept. of Electronic Eng., Robotics Univ. of Oxford |
VII | TR1/059E | Feasibility Evaluation of Transputer Equipment for Real Time Image Processing | R.L.Rhodes, G.J.Porter | Univ. of Bradford |
VII | TR1/077E | Integrated High Speed Information Router for VLSI Design | B.C.O'Neill | Trent Polytechnic, Dept. of Electronic Eng. |
VII | TR1/079 | The ZEUS Central Tracking Chamber Second Level Trigger | R.Devenish et al., J.Lane et al | Univ. of Oxford |
VII | TR1/086 | Calculation of Fractal Dimension of Closed Loop Shapes for Particle Characterisation | W.B.Whalley et al. | Univ. of Belfast, Dept. of Geography |
VII | TR1/088 | A Parallel Processor Implementation of Algorithms used in the CAD of Integrated Circuits | G.Russell | Univ. of Newcastle, Dept. of Electronic Eng. |
VII | TR1/098 | A Formal Method and an Empirical Metric for Memory Latency in Multiprocessors | D.F.Snelling | Univ. of Leicester |
VII | TR1/100 | Application of Real-Time Image Processing to Dynamic Measurements on the SERC Earthquake Simulator | G.Taylor, G.Stephen | Univ. of Bristol, Dept. of Civil Eng. |
VII | TR1/101 | Feasibility of Using Transputers as a Solution for Real-Time Image Analysis | P.Forte, P.J .Netherwood | Kingston Polytechnic |
VII | TR1/104 | Introducing Occam and the Transputer in Courses in Electronics and Computing | K.Pillay | Univ. of Exeter, Dept. of Eng. |
VII | TR1/111 | Transputer-based Prototyping of Object-Oriented Parallel Hardware Descriptions | P.Kelly, C.Hankin | ICSTM, Dept of Computer Sci. |
VII | TR1/117 | The Potential of Transputers for Computing the Hydrodynamics of Offshore Structures | D.Pizer, P.Sayer | Univ. of Strathclyde, Dept. of Marine Technology |
VIII | TR1/022 | Graphics Applications | D.F.Beal, M.Slater, A.Davison | QMC, Dept. of Computer Sci. |
VIII | TR1/060 | Transputer Based Fast Graphics Multiplier | A.P.Boume | Univ. of Hull, Dept. of Electronic Eng. |
VIII | TR1/061 | A Graphical Design Environment for Transputer Systems | I.Hyland, M.Moulding | RMCS, Dept. of Electronic Eng. Development |
VIII | TR1/068 | A Parallel-processing Approach to Computer-Based Music Generation | M.Greenhough, J .Shiulewonh | Univ. of Cardiff, Dept. of Physics |
VIII | TR1/084 | A Multiprocessing Approach to Real Time Control | J .Billingsley, J .E.L.Hollis, F.Naghdy, P .Strickland | Portsmouth Polytechnic, Dept. of Systems Engineering |
VIII | TR1/087 | S/W Development for Testing Methods of Parallel Solution of Tridiagonal Systems of Linear Equations | M.Tomlinson | Sheffield City Polytechnic |
VIII | TR1/088 | A Parallel Processor Implementation of Algorithms used in the CAD of Integrated Circuits | G.Russell | Univ. of Newcastle, Dept. of Electronic Eng. |
VIII | TR1/093 | Transputer-based Readout Controller | R.Belusevic, G.Nixon | Univ. College, London, Dept. of Physics and Astr. |
VIII | TR1/097 | Parallel Computation in Chemical Process Simulation | R.Best | South Bank Polytechnic, Dept. of Chemistry |
VIII | TR1/102 | A Feasibility Study into Implementation of Abacus's Lighting Model (DIM) | H.Baniamer | Univ. of Strathclyde, ABACUS |
VIII | TR1/105 | Adapting the PACE CFD Program to run on an Array of Transputers | A.Manners | Loughborough Univ., Dept. of Transport & Technology |
VIII | TR1/106 | Solving Symmetric Matrix Problems on Rings of Transputers | M.Pidcock, K.Paulson | Oxford Polytechnic, Dept. of Applied Analysis |
IX | TR1/026A | Simulation of Nerve Networks | J.B.Willis | Univ, of Southampton, Dept. of Maths |
IX | TR1/052 | A Transputer-based System for Locating Parts and Controlling an Industrial Robot | D.T.Pham, Huosheng Hu, J.Pote | Univ. of Cardiff, Dept. of Electronic Eng., Univ. of Oxford, Dept. of Eng. |
IX | TR1/062 | Parallel Monte Carlo: A Distributed Memory Approach | M.Wilby, S.Clarke | ICSTM, Dept. of Physics |
IX | TR1/066 | Hierarchical Circuit Simulation on Transputer Trees | Y.Bouchlaghem | Univ. of Southampton, Dept. of Electronics and Computer Sci. |
IX | TR1/095 | Evaluation of Fast Fourier and Hartley Transforms on a Loosely Coupled Multiprocessor Based on the Inmos Transputer | S.Gudvangen, G.J.Holt | Univ. of Newcastle, Dept. of Electronic Eng. |
IX | TR1/112 | Mixed Integer Programming on an Array of Transputers | R.Ashford, R.Daniel | Univ. of Warwick, Univ. of Buckingham |
IX | TR1/116 | The Learned Control of Complex Dynamic Systems | E.Grant, B.Zhang | Turing Institute, Univ. of Strathclyde |
IX | TR1/128 | The Characteristic Initial Value problem on a transputer array | N.T.Bishop, C.J.S.Clarke, R.A.d'Invemo | Univ. of Southampton, Dept. of Maths |
IX | TR1/152 | Investigation into Change in Running Time of Existing Occam Building Simulation Program | N.H.Cole | Univ. of Bristol |
IX | TR1/164 | Multiprocessor Solution of Non linear Equations for Chemical Process Simulation | D.Juarez, C.C.Pantelides | ICSTM, Dept. of Process System Engineering |