House of Lords - Science and Technology

Select Committee on Science and Technology Second Report

CHAPTER 8: NEW OPPORTUNITIES FOR THE UNITED KINGDOM

8.1 Against the background set out in the preceding Chapters, we now turn to consideration of the opportunities open to the United Kingdom and ways in which their exploitation could be better facilitated. This Chapter deals with research and development, but the points here are complemented by the discussion of skill needs and bringing products to market in, respectively, Chapters 9 and 10.

8.2 UK industrial success in the area of microprocessing will depend critically on appropriate ideas and developments coming through from the global research community. However, the flow is not one-way. Research and, more obviously, development can often be at its best when industry keeps the researchers and developers closely in touch with their developing vision of the future market. This Chapter discusses changes to the current level and balance of research activity to improve its relevance to UK industry and its potential to deliver economic benefit over the twenty years on which this Inquiry has focused.

Research into devices and manufacturing technologies

8.3 As noted earlier, the United Kingdom no longer has any major ownership of semiconductor manufacturing facilities. Some foreign companies have semiconductor manufacturing facilities here but, as Dr Cowburn said (p 80), these are production facilities and little involved in the development of new technologies. State of the art fabrication facilities have concentrated in the Far East for sound economic reasons and it is unrealistic to contemplate securing a substantial UK share of mainstream CMOS semiconductor manufacturing. In addition, the lead time for developing incremental improvements of the current CMOS device technologies means that new research to this end is unlikely to yield major economic benefits.

8.4 Once CMOS reaches its physical or economic limits (as discussed in paragraph 4.17 onwards), any opportunities for further progress in microprocessor manufacturing technology will have to come from alternative technologies. There is significant research being undertaken in the UK (and, of course, elsewhere) into alternatives to CMOS. As noted in paragraph 2.7, it was Professor John Enderby's suggestion that the UK's work would benefit from greater co-ordination and focus that led to the establishment of this Inquiry.

8.5 That message was repeated in the written evidence from The Royal Society[71] (p 221) which also noted that many alternatives to CMOS were currently being investigated. That was borne out by all that we heard during our Inquiry, as was the fact that there is not yet an identifiable winner in the race for an alternative technology to CMOS. This suggests that funding should be blue sky in nature, and spread across a range of promising candidate technologies until their potential becomes clearer. We therefore recommend that, within the funding allocated for research into computing devices and manufacturing, the DTI and EPSRC should place the emphasis on radical developments in microprocessor technologies where there is a greater prospect that UK industry could play a significant role long-term. Further, they should also ensure that this work has appropriate co-ordination and focus.

8.6 As noted by Dr Cowburn (p 80), the recently-launched UK Research Council's Basic Technology Research Programme is a good example of an appropriate funding model. However, with a success rate of only 2.5% (eight proposals were funded from 229 submissions) many innovative ideas will not have been supported.

8.7 None of the foregoing is to say that there is no part for the United Kingdom in the global effort to enhance mainstream CMOS manufacture. There will be many opportunities, particularly for niche contributions. One example is Exitech[72], a successful UK manufacturer of leading-edge development equipment for the production of semiconductors.

Nanotechnology

8.8 Shortly before our evidence session with the Minister and Dr John Taylor on 3 July 2002, the DTI published a substantial report by its Advisory Group on Nanotechnology Applications[73]. This covered a range of topics, some of which were directly or peripherally relevant to the future of semiconductor technology. As foreshadowed by the Minister (Q 514), the Government formally accepted the broad thrust of the Group's recommendations[74]. Given the generic importance of nanotechnology for the development of microprocessing as noted in paragraph 4.43, we were disappointed that the report had very little to say on microprocessing technology and see this as an omission. Subject to that, we generally welcome the broad-ranging report, and particularly the key recommendation highlighted by Dr Taylor (Q 468) that two research institutes be established to cover this major new area.

8.9 A March 2002 report of the Foresight Programme's Defence and Aerospace National Advisory Committee on Electronic Devices and Materials[75], recommended the establishment within the United Kingdom of a state of the art semiconductor fabrication facility for both standard CMOS and other semiconducting materials at an estimated cost of £100 million. The Royal Academy of Engineering's memorandum (p 101) noted the potential advantages of employing nanotechnology techniques in the production of microelectronic circuits. The Institute of Physics endorsed this view (Q 175).

8.10 Because of the significant commonality between nanotechnology and likely future computer technologies, we recommend that the DTI should include the exploration of alternatives to CMOS in the remit of at least one of the proposed nanotechnology centres. Further, we recommend that the DTI should consider integrating into that centre the semiconductor processing facility recently recommended by the Foresight Defence and Aerospace Advisory Committee.

8.11 Many developments in nanotechnology and microsystems will, as noted by Scottish Enterprise (p 227), yield devices that require close integration with microelectronic systems to provide their control, interfacing and signal processing functions. Similarly, future computing devices may require integration onto CMOS substrates. The support of these new technologies will require access to a microelectronic design and architecture capability that might best be provided as one role of the national centre proposed in paragraph 8.32.

8.12 A case in point is the National Microsystems Packaging Centre[76]. This is being planned by the Northwest Development Agency to provide high-technology companies with customer-focused R&D in the joint development of advanced microsystem-based[77] products. The development of such products will obviously involve a significant element of microelectronics design. Crucial as that design will be to the success of the products, it is likely to have negligible academic research value and thus be of little interest to university microelectronics design research groups. However, a national centre for SoC design could readily accommodate such a service within its operational remit.

Research into design and architecture

8.13 While, as noted above, the United Kingdom no longer has any significant ownership of semiconductor manufacture, we do have major strengths in integrated circuit design and architecture. ARM Limited, based in Cambridge, develops microprocessor architectures for embedding in SoCs, and has established a world-leading position in the business of licensing these designs. A majority of the mobile phone handsets sold world-wide incorporate microprocessor technology licensed from ARM, and the same microprocessor cores are also used (exploiting the universal property of programmable computers) in a very wide range of electronic products including digital cameras, disk drives, printers, portable music players, and so on. Sales of products based on ARM's technology exceed a million units a day.

8.14 Another leading 32-bit embedded processor core, the ST20[78], was also designed in the United Kingdom in the 1980s. The expertise developed during the development of the ARM and ST20 cores has, as noted by Professor May (p 89) led to the establishment of several other UK-based groups developing novel processor architectures currently in design.

8.15 Other UK design companies such as Globespan Virata, ARC and Amphion Semiconductor[79] control significant IP in the field, and these represent major success stories. It is important to note that the UK strength in design and architecture is principally based in a multitude of small companies. ARM Limited — one of the largest and most visible — has well under a thousand employees. Many other successful companies are much smaller.

8.16 The United Kingdom has a small but world-class academic research community in the areas of design and architecture, although the various groups are not well co-ordinated and links with industry are variable. As listed by the IEE (p 44) and Professor May (p 92), notable strengths in computer architecture exist at least in the universities of Bristol, Cambridge, Edinburgh, Loughborough, Manchester and Sheffield. There is scepticism in some circles that microelectronic system design is a valid academic research topic. However, the view that such design is simply an aspect of product development better left to commercial organisations fails to grasp the fundamental problems posed by complexity in these systems. As tellingly noted by the IEE (Q 112), the ITRS uses the word "crisis" only once in its several hundred pages — to describe the problems facing the industry in designing the systems that improving technology makes possible.

8.17 EPSRC noted in its memorandum (p 20) that

"A design methodology revolution is required… Much greater connectivity is needed between universities themselves and industry. The United Kingdom is well placed, with ARM at the forefront, to take a lead provided that there is increased support for the work on design methodology and advanced System on Chip design."

8.18 We fully endorse the message that the United Kingdom has a major opportunity to contribute to, and benefit from, progress in microprocessor design and architecture if it can expand and co-ordinate its research and industrial activities in this area.

8.19 It is worth noting that these UK strengths in design and architecture are relevant not only to the current CMOS technology but, in many cases, are also directly transferable to alternative technologies should these be developed to the point where they begin to displace CMOS in some markets. Radically different technologies, such as biological and quantum computing, are likely to require radical new approaches to design and architecture, but technologies that replace the CMOS transistor switch with a smaller, faster and/or cheaper switch are likely to require design and architecture skills quite similar to those currently deployed to make the best use of CMOS. Any investment the United Kingdom makes to enhance its strengths in these areas is likely to be transferable to a new technology that displaces CMOS.

Formal verification

8.20 The United Kingdom is also the recognised world-leader in research into formal verification and semantics[80], and has been responsible for much of the foundation work in this area: page 17 of the International Review of UK research in Computer Science[81] states, "UK researchers wrote the book on semantics." With the increasing complexity of integrated circuits, there is a growing need to find mathematical ways to validate designs.

8.21 Bringing UK strengths in this area to bear on the problem of complex hardware system design would have the potential to make a major contribution to keeping design costs under control. This would help manage the design crisis identified in the ITRS and, if combined imaginatively with UK strengths in design and architecture, could offer considerable opportunity for enhancing the United Kingdom's position.

Enhancing activity in design and architecture

8.22 As the DTI (p 14) and others[82] have noted, the United Kingdom has recognised academic and industrial strengths in design and architecture. For example, we have the lead among EU Member States with 40% of the independent semiconductor design market. The principal challenge is to reinforce and intensify these activities.

8.23 It is reasonable to question whether or not the research community is providing adequate support for industry in this area. The weight of evidence is that, while there are pockets of high-quality research, these are both too few in number and insufficiently co-ordinated to provide solid support for industry. This problem is compounded by the fact that the relevant companies are small and the industry is fragmented, so there is no single large company able to provide a focus for academic research.

8.24 The IEE recommended (p 44) that:

"there should be more recognition given to the high calibre work in systems design and architecture and software development that already takes place in this country's IT sector, which results in research council funding being more closely aligned with the industrial strengths of the United Kingdom".

8.25 The IEE (Q 108) and IoP (Q 167) noted that the balance of funding within the relevant EPSRC programmes significantly favoured device and manufacturing technologies, where the United Kingdom has little industrial strength, over design and architecture, where several UK companies were world leaders. The Royal Academy of Engineering's supplementary memorandum (p 110) also observed that "in terms of EPSRC funding, design has been less prominent [than technology]".

8.26 The Minister (Q 469) noted that this perception was, at least in part, due to the fact that research into devices requires significant capital facilities. The memorandum submitted by EPSRC detailed the annual running costs of their principal capital facilities. These include the Microelectronics Fabrication Facility at Southampton University which costs EPSRC £2 million a year to operate. Although it has a CMOS process, it is not suitable for fabricating the multi-million transistor chips required for research into design and architecture. It is used principally for device and small-scale circuit research.

8.27 There does seem to be a significant mismatch and, accordingly, we recommend that the Research Councils — in particular, EPSRC — should review the funding for research into microprocessor design and architecture to ensure that both its level and its delivery are matched to the strengths and needs of UK industry.

Delivering enhanced funding

8.28 There is then the question of how such funding would be most effectively delivered. One model, as proposed by the BCS (Q 120) — which saw the possibility of doing "something splendid with £100 million" — is the distributed approach. This was successfully adopted for the recent e-Science initiative[83] which is based upon one national and eight regional centres. The merit of such an approach is that it can be established rapidly by engaging existing resources in the leading university groups to create a virtual centre of excellence.

8.29 However, we received considerable support for the establishment of a more substantive, physical, national centre. For example, the Royal Academy of Engineering noted (p 101):

"Major international collaboration would be beneficial… A major initiative, for example on SoC architecture, design and implementation, could be very important in ensuring that the United Kingdom remains at the forefront of such technology. The establishment of a centre of excellence within the United Kingdom would help leverage this type of international collaboration."

8.30 The Royal Academy of Engineering saw IMEC[84] as a good example of a national research institute, but felt it was too late for the United Kingdom to try to get back into semiconductor manufacture: we should concentrate instead on design and architecture (Q 288). Support also came from a number of others, for example British Airways commented (p 191) that:

"Initiatives that bring together industry and academia in more practical deployments of ideas around the emerging areas of highly pervasive and portable computing combined with high bandwidth communications capabilities would be particularly welcome."

8.31 Sir Robin Saxby was less convinced of the desirability of a research institute (Q 368): "I am more for linking what we have got and making it work than [for] creating another 'Great White Hope'." His view was that we needed a common vision to drive research forward and to connect the interested parties: "You need a serious theme to make a difference." Sir Alec Broers also noted the difficulty the United Kingdom historically has had in agreeing the location of a single institution (Q 296).

A national centre For System-on-Chip design

8.32 Computing in all its forms is already of overwhelming importance. As new applications come on stream (by exploiting present technologies, let alone the projected improvements), its significance can only increase. The United Kingdom has, through the design and architecture industry that has emerged comparatively unnoticed from a number of uncoordinated sources, a significant stake in the global industry. We see it as essential to underpin the currently fragmented industry with a clear and dedicated research infrastructure. So that UK strengths in computer design and architecture can be maintained and developed within the global market, we therefore recommend that the DTI and EPSRC should urgently establish a single national research institute for System-on-Chip design.

8.33 The Royal Academy of Engineering helpfully produced an outline proposal for such an institute in its supplementary memorandum (p 110). We were pleased to note that, when we discussed this matter with the Minister, he indicated (Q 469) that EPSRC was indeed looking into such a centre in order to concentrate UK resources into a place where critical mass could be achieved.

8.34 To minimise possible turf wars, any new national centre for these purposes would need to be located off any existing university campus but, to assist in gaining critical mass, perhaps near one of our major national laboratories. Its site would need to have room for a spin-off park, and have good physical communications. It should look to funding from the Research Councils, Regional Development Agencies and EU programmes. It would also be important for industry to be strongly involved in both the setting up and operation of the proposed centre, where the aim should be to establish a new style of institution, learning lessons from the difficulties that have been experienced in previous initiatives.

8.35 For the centre to succeed, it will need to establish and retain a good body of experienced researchers. This has been a particular difficulty for university groups because of the fixed-term contracts commonly used to provide a means of matching expenditure commitment to income. The centre should be able to offer longer-term employment opportunities thereby avoiding these problems and ensuring the continuity of knowledge and experience that it will need.

8.36 The purposes of such a centre should be to:

(a) do development work;

(b) train students;

(d) co-ordinate EU applications; and

(e) take on contract research and development for industry.

8.37 We would also see advantage in setting up this institute with a strong application focus. This would require, as desirable, the institute to be of an interdisciplinary nature. It should draw on academic research for the fundamental insights required to develop useful prototypes, and industry should take those prototypes through to production. Care would, of course, need to be taken in selecting an appropriate focus. Possibilities we advance for consideration are ambient computing (a focus of the next EU programme — see paragraph 7.29) or, as Sir Robin Saxby noted (Q 357), biomedical systems which would mesh in well with other UK strengths.

8.38 There is some overlap between our proposal and the Institute for System Level Integration (ISLI) based in Livingston, near Edinburgh. Opened in 1998, this has, as noted by Scottish Enterprise (p 227), a mission to "support the growth of systems design, SLI, SoC and related activities with world-class research, education and training programmes." It is a joint venture of the universities of Edinburgh, Glasgow, Strathclyde and Heriot-Watt and forms an integral part of Scotland's Alba Centre. Again as indicated by Scottish Enterprise, this "has at its heart the mission to develop Scotland as a leading world centre in microelectronic product design and technology."

8.39 However, we see our proposal as distinctively different. Rather than a university-based enterprise, it would be a free-standing centre with both direction and significant funding from industry. Moreover, it would have a significant core of full-time professional staff with terms of employment commensurate with its industrial standing. Above all, it would have an explicitly UK-wide reach.

A national programme

8.40 It is clear that there needs to be a better bridge between academic design and architecture research and the industry. Our proposed national centre would provide a clear physical focus for increased and more coherent activity in design and architecture. Inevitably, it will take a little time to get off the ground. We recommend that the DTI and EPSRC establish a national programme for design and architecture, along the lines of the e-Science initiative, for the three-fold purpose of providing an earlier focus for research and development in design and architecture; wider support for the institute when established; and a framework for potentially useful activities not central to the particular focus selected for the institute.

8.41 To facilitate the early establishment of this national programme and ensure its continuing relevance, it should be developed from and round the nodes of other relevant activity. These might include existing research establishments, DTI's new nanotechnology centres (as noted in paragraph 8.10) and other relevant initiatives — perhaps particularly those emerging from Regional Development Agencies[85]. The opportunities such nodes should provide for bringing together research, academic, industry and finance interests should also assist the development of clusters to help generate greater critical mass for developing the market, as discussed further in Chapter 10.

Short-termism

8.42 It is important that any plans to develop research should, as discussed in paragraph 7.34, recognise the long-term nature of any such activity. Many past initiatives have had unrealistic expectations of early returns on the invested funding. For example, the 1980s Alvey programme[86] was perceived as largely unsuccessful immediately after its termination. However, as Professor O'Reilly noted (QQ 71-73), there is now a much more positive view of the benefits the programme brought, ten years after its conclusion.

8.43 There are numerous examples that demonstrate that, despite the reputation of the microelectronics industry for very rapid product cycles and its emphasis on short time-to-market, fundamental research frequently takes twenty years or more to reach fruition. A case in point is the ARM processor. This was designed in the early 1980s, but its commercial success was not widely apparent until quite recently. We therefore recommend that the DTI and EPSRC should explicitly recognise that any new funding initiative in this area would require sustained long-term commitment.

Summary

8.44 By way of summary, we can do no better than to quote the closing paragraph of the memorandum by the IEE (p 45).

"The United Kingdom is in a fortunate position of being a significant player in certain areas of microprocessor design. It also has a small but important cadre of world-class academic researchers. Current research funding is arguably targeted more at devices than design and would benefit from a closer alignment with UK industrial capability. However, the will is there to create a coherent and competitive academic research capability to support a flourishing UK industry. Better co-ordination and a targeted programme of support to ensure adequate manpower and research capability will help ensure that in the next 25 years the United Kingdom maintains and extends its dominance in embedded and application-specific microprocessor design. This is a significant field and we are significant players."

71 Of which Professor Enderby is the Physical Secretary. Back

72 See Exitech's memorandum on p 194. Back

73 New Dimensions for Manufacturing - a UK strategy for nanotechnology, DTI, June 2002, URN 02/1034. Back

74 DTI Press Notice P/2002/479, 23 July 2002. Back

75 See www.foresight.gov.uk Back

76 www.nwda.co.uk - follow links to "Business Development" and "Innovation", or search for "nmpc". Back

77 Microsystems here are combinations of micron-sized sensors or actuators with electronic functions manufactured on a single chip. Back

78 A derivative of the Inmos transputer, see paragraph 7.4. Back

80 Semantics relates to giving a computer language (for either hardware or software definition) a rigorous mathematical interpretation and using this to reason about the function of a program. Back

81 Edited by Fred Schneider and Mike Rodd, published by the IEE in November 2001. Back

82 See, for example, the comments in the memorandum by the Royal Society of Edinburgh on UK strengths in parallel processing and computing "mavericks" (p 226). Back

83 See the Appendix to Annex 2 in the supplementary memorandum from the DTI on p 14. Back