Oral history of British science
Furber, Steve (Part 5 of 5)
The British Library Board acknowledges the intellectual property rights of those named as contributors to this recording and the rights of those not identified.
Legal and ethical usage »
Interviewee's office, University of Manchester
Furber, Steve, 1953- (speaker, male)
Lean, Thomas (speaker, male)
Part 5 [1:55:40][Interview Five: 18 December 2012] Comments on ARM chip story: testing of first ARM chips in 1985; anecdote about testing tiny power consumption of early ARM chips, important in later uses; low energy consumption originally intended to allow chip to housed in cheap plastic packages; Apple desire to use ARM in Newton PDA [Personal Data Assistant] in early 1990s; importance of low power applications and Nokia in 1990s; [3:30] intention for ARM as part of four chip set, SF responsible for memory controller chips, Mike Muller I/O [Input/Output] controller, Tudor Brown video controller; [04:55] Remarks on ARM memory controller: simplification of memory controller by SF, anecdote about David Wheeler at Cambridge University telling SF about early use of technique on 1960s Manchester Atlas computer; limited use of ARM1 in development system; improvements in ARM2, used in Acorn Archimedes computer. [07:57] Remarks on Archimedes: high performance compared to contemporaries; commercial limitations with not being IBM PC compatible; SF involvement in Archimedes design, closeness of chip-set to end product. [11:15] Remarks on changes in Acorn company: situation early in 1980s working closely with technically ambitious Hermann hauser; financial difficulties, 1984 Olivetti takeover; Olivetti stronger financial management and misunderstanding of ARM; limited commercial success for ARM initially, VLSI licence of ARM design, use of ARM in Radius Graphics Accelerator; flat business in late 1980s; search for way of separating ARM from Acorn, limitations of royalty based business; improvements of ARM processor, advantages of adding cache memory for making a faster Archimedes; ARM2aS used in Hermann Hauser's Active Book tablet computer; Andy Hooper drawing SF attention to Manchester vacant post as Professor of Computer Engineering at Manchester, anecdote about leaving university as a research fellow and going back as a professor in 1990; Apple joint venture with Acorn over ARM in 1990 after SF left; [18:45] expansion of ARM in late 1990, in Swaffham Bulbeck under Robin Saxby; Robin Saxby's idea for up front licence fees; growth of ARM into high value company. [21:30] Remarks on huge success of ARM: low corporate turnover by high profitability; large numbers of chips in use; GEC Plessey benefiting from Apple Newton; value of Apple name to ARM, despite limited success of Newton; Robin Saxby and team hard work in success; importance of 1995 Nokia contract to ARM; development of business outside mobile phones; anecdote comparing ARM and Intel chip numbers; influence of ARM around World, use in Apple, Samsung, HTC products and in other applications; [25:50] SF surprise at ARM success; anecdote about Robin Saxby focussing on global success from early stage; anecdote about Robin Saxby turning ARM engineering team into senior managers, talent of ARM team beyond technical skills, such as Jamie Urquhart, Mike Muller and Tudor Brown; SF feelings about leaving ARM before its eventual commercial success. [32:21] Comments on starting work at Manchester University: limited ideas about early role; taking work on European collaboration project with him from Acorn; involvement of existing staff with SF project; limited teaching at first, previous lecturing experiences; familiarity of students with BBC Micro; [37:40] SF lecturing style. [39:15] Comments on Manchester School of Computer Science: history of Manchester in computer engineering; structure of department's research interests, change to multiple research groups; John Gurd and Ian Watson work on data flow machines in 1980s; ICL funding for SF chair; colleagues Doug Edwards, Viv Woods, and Linda Brackenbury; anecdote about unexpected success in applying for EU research funds in 1992; change in importance of funding from EU to UK EPSRC sources; [44:00] EU funding links with industry, many involving ARM; differences between EU and EPSRC funding arrangements. [45:45] Comments on research interest in clockless systems in 1990s: talk by Craig Mudge from Austek, based on Ivan Sutherland work; interest in robust asynchronous circuits, developing ARM compatible clockless processors during 1990s, ARM purchasing rights; ARM-Phillips asynchronous clockless processor; lack of industry interest in clockless processors due to poor support from design tools; 2003 Silistix spin-out to develop tools, limited commercial success; description of use of clocks in computers, problems with power use and radio interference; description of clockless operation using handshaking; future potential for clockless processors. [51:35] Comments on interest in making computer models of the brain from c1998: asynchronous nature of the brain's systems; exascale computers making recreating power of brain foreseeable; computer to explore problem using a million ARM processors, equivalent to 1% of human brain; development of SF interest in modelling brains, frustration with developing merely faster computers; [55:45] anecdote about reading neuroscience; SF interest in associative memories and their limitations, compared to biological systems such as sight; SF bid for EPSRC Realising Our Potential grant in new area of research, inexact associative memories; anecdote about research leading inevitably to Neural Networks; 'Sparse Distributed Memory' by Pentti Kanerva; SF publications on spare distributed memories; [59:50] key problem of not understanding the brain enough to build networks based on it; SF motivations in wanting to understand the brain; communication between neurons. [1:02:50] Description of SpiNNaker [Spiking Neural Network Architecture] computer architecture and physical construction. [1:05:58] Remarks on hope for research: understanding brain and making new types of computer; brain's ability to cope with component failure; importance and challenges of parallel computing. [1:08:20] Remarks on: interdisciplinary nature of brain modelling work, including psychologists and neuroscientists; debugging SpiNNaker; examples of student use of SpiNNaker for biological models; media reactions to brain modelling, 'IEE Spectrum' article, 'BBC Techlab'; anecdotes about filming Discovery Science Channel 'Through the Wormhole' programme. [1:14:35] Comments on potential for thinking machines: work of Alan Turing at Manchester, continued difficulties despite development of computers from 1948 to c.2000, Turing Test; limited understanding of natural intelligence; Jeff Hawkins theories, 'On Intelligence'. [1:18:35] Remarks on: naming of SpiNNaker; definition and description of neural networks; structure of networks within a brain. [1:22:50] Remarks on daily work on SpiNNaker: phased nature of project leading from concept, through design, testing; teamwork; 'tape out' at end of development process, leading to return of finished chip; risks and potential problems; hunting for bugs. [1:26:55] Remarks on SpiNNaker team: SF leadership role; Jim Garside role on hardware development; Viv Woods organisational work; Dave Leicester; importance of post-docs to core work, Steve Temple, formerly of ARM; Luis Plana, Eustace Painkras, Simon Davidson, Cameron Patterson, Sergio Davies, Alex Rast; role of PhD students Yebin Shi, Jian Wu, Shufan Yang in hardware; new group of PhD students working using chips, such as Francesco Galluppi, Martin Grymel, ARM CASE student Tom Sharp, Evangelos Stromatias; [1:31:20] value of contributions of PhD students to SpiNNaker, comparisons to postdocs. [1:33:00] Remarks on: SF role running group, interaction with PhD students and postdocs; split of time of PhD students between project contributions and thesis; value of PhD publications; Research Excellence Framework [REF] and SF role on REF panel; publication authorship. [1:37:29] Comments on chairing 2011/12 Royal Society report on computing in schools and revamp of ICT curriculum: state of ICT in British school; recommendations to support digital literacy and computer science; teaching of programming at primary schools, such as through Scratch; supporting perspectives from Eric Schmidt's MacTaggart lecture and NESTA Livingstone-Hope report; Raspberry Pi similarities to feeling of BBC Micro. [1:42:40] Remarks on BBC Micro: portrayal of BBC Micro today; comparisons with Sinclair computers; value as an educational product, compared to recent computer education. [1:44:20] Remarks on: frequent visits to London for professional body committee meetings; SF membership of Royal Society, Royal Academy of Engineering, IEEE, Institution of Engineering and Technology, British Computer Society; SF activities as Vice President of British Computer Society and with other learned societies; presentation of electronic industry to government, National Microelectronics Institute [NMI] and minister Mark Prisk. [1:47:15] Remarks on awards: Royal Academy of Engineering Silver Medal; IET Faraday Medal; Finnish Millennium technology prize laureate; CBE; Fellowship of Computer History Museum in USA; lengthy wait before technology impact clear; enjoyments of work today at university, freedom of work compared to working in industry; extensive travel for work, anecdote about visiting San Francisco for a day; [1:53:25] recent grant awards, prospects of retirement; reflections about interview.
Interview with computer scientist, Steve Furber