Sophie Wilson

Sophie Wilson interviewed by Alan Macfarlane 21st May 2017

0:05:11 Born in Leeds in North Yorkshire in 1957; my mother's mother lived down the road so we could see her when we wanted; when I had things like a shocking asthma attack at school, when both parents were at work, then Gran could be summoned; in fact her house was closest to the bus stop so you could drop in there before going home if you wanted to, or if there were critical family illnesses and you had to be quarantined off then you'd go and stay with Gran for a while; so I had German measles with her, one of my sisters had measles with her; for all my life Gran was retired; she was a Magistrate, a companion, she took French at university being the first one to go to university at all, and she did travel a lot; my parents were both teachers; Mum taught physics at a girls' school, Dad taught English at a college of further education; they are both quite independent of thought - they are both living but my Grandmother died at ninety-six - my parents are self-sufficient; one of the ways that 1957 was an epoch ago was that I lived in a world built by people I knew; the house was furnished with stuff my parents had made, Dad building the furniture and Mum the soft-furnishings; when we were young Dad built a car, built boats, for my Mum's physics lab at school, we the family jointly built all the equipment; that is inconceivable nowadays; nothing is made by people you know; Dad was as violinist, had been in orchestras, music in the house was mostly Baroque; it was only when we individually got our own personal radios as we were growing up that we could listen to anything else; we didn't have a television in the house until the moon landings which we saw on a tiny black and white Sony Trinitron portable tele, having been invited downstairs by Mum to watch the critical moments which were late at night; as a result, all the children and our parents read voraciously; we would go to the Library and each get out our maximum allocation of books which I think was four each - twelve children's and eight adult's books - and we'd take them all back next week, all read; I remember reading all the Chalet School stories one after another; for birthday presents we would each get an Arthur Ransome book, so an awful lot of 'Swallows and Amazons' right up to 'We didn't mean to go to Sea', which I understand they have just made into a film which is a bit strange; loved books which are made into films rarely work for us; Dad had quite eclectic tastes; if we were very good we could read some of his books; he had things like T.H. White, Tolkien, as well as an awful lot of English literature; he wasn't into C.S. Lewis I think because of the religion, though Dad was a lay preacher; I did go on to read all of Lewis's writings including 'Perelandra' and 'Out of the Silent Planet' rather than 'Narnia'; that was something that only children read so not something that Mum and Dad had ever read 7:19:23 I'm not sure that I have an early concrete memory any more; I am very aware of stuff that's a real memory versus stuff that's a memory of a memory; for childhood things are only memories of memories of memories now; I don't have any feeling that I was actually there for an event, like, for example, I had my thumb squashed in the hinge of a telephone box; I know it happened but I have absolutely no memory of it at all; I know where I went to school, but no memory of it at all; I know that there was a flood and that we walked to school in wellingtons for a while, but no memory of that at all, and so on; the difference between things I know happened and things that I can actually remember if fairly stark; I'm not even sure that teenaged memories have any first-person quality to them, but are more memories of memories; between six and eight I had no hobbies but reading; Gran used to send back postcards from interesting places so there was a stamp collection, but it wasn't very good; we had Lego, Meccano, and there were a lot of jigsaws which I did; unfortunately I had all my jigsaws out and assembled when Mum did the chip-pan fire and affected the house a lot; on the plus side I got help filling in the insurance claim for my jigsaws to buy some new ones; on the minus side they were all covered with sooty gunk and had to be thrown away; we had huge numbers of books; of the three children, we get progressively more short-sighted from oldest to youngest; I am the eldest, and the youngest sister is really badly short-sighted; I was just about good enough to get away without glasses for quite a while; my left eye is much less short-sighted than my right eye which is quite profoundly short-sighted; my right eye is -5, left eye used to be -2 but is getting towards -3 now, and I have a lot of astigmatism as well; but basically, if you sat at the front of the class you didn't have much trouble seeing the board and so on, so I wasn't even aware that I needed glasses until a conversation with my Mum; again, this is not a first-person memory any more, just a memory of a memory where we were going down the street and Mum said we would stop at a particular shop; I asked where it was and she pointed to a sign, and I said "what sign"? I was probably something like eleven or twelve at the time, so quite old
11:38:12 I think my first school would have been Panel Primary School, but there might have been something that was better than a kindergarten and worse than a primary school - Grosvenor House School - before that; I have suffered from asthma since I was two and bronchial asthma affects particularly in childhood; nowadays they are brought under much better control but they were terrible; the third year at grammar school I lost a whole term through a bad bronchial asthma attack; after that I wasn't taken to see my father's parents any more because they lived in Mansfield, Nottingham, and the area was thick with coal dust at the time; it's much nicer now but then it was sufficient to have a whole term off school; as a result I don't know anything about the mechanics of engines, the physics of engines; I did the 11+ exam and just scraped through as borderline; unlike today people weren't aware of what was going to happen; one day I was taken to a strange place, a school I had never been to on the other side of town, did the 11+ exam, no practice, nothing; I got into a grammar school which was probably a good one; having parents who are teachers tends to expose flaws in a school's teaching; I am not exactly academically gifted; the streaming in the school wasn't great for me in any way; I ended up in a A3 stream which is sort of the wrong place; it meant that I could coast for a lot of subjects for a lot of the time; with a mother who is an expert in physics and a father, an expert in English, all the kids were mathematically inclined - it must be something in the genes; my intelligence is different from, qualitatively worse than, whatever, but I have a better memory, a near-photographic memory; I can't remember a page of Tolkien in detail, but near, so you get the gist; but memory is a big advantage in doing exams, and with memory and a little logic you can get through most exams quite well; so it isn't until exams get difficult and are more than just regurgitating the facts that you have already been told that one's lack of ability becomes obvious which for me happened in the second year at Cambridge; probably my ability with maths was nothing more than remembering an awful lot; so it was a grammar school with streams but the streams were based on the overall level, so I was in A3 rather than A1; A1 and A2 got extra subjects, Latin and things like that - I didn't need Latin to get into Cambridge by that time; I was OK at most subjects, 'O' Levels you could just do them, particularly the maths exam which was little more than just regurgitating facts; the teacher was aware of this and we were banned from leaving the room in a three-hour exam before two hours were up, and I think I did the paper in a little over an hour; I spent the next hour memorizing digits of Pi, and precisely at the two-hour mark half the class stood up and walked out; I never liked chemistry so although I was expected to go off to school and choose maths, physics and chemistry for 'A' levels, I ended up choosing maths, further maths and physics, also doing the 'S' level papers for those as well; that wasn't too bad; then I took the Cambridge entrance exams with a lot of coaching from a Cambridge maths graduate who was my teacher at the time, and did get in; of teachers, the maths teacher being a Cambridge graduate was probably the best all-round combination of being a good teacher and having command of your subject; there are other teachers who either had command of their subject or were good teachers, but rarely both, and mostly they concentrated on the A1/A2 stream and I didn't get taught by them 20:24:21 On religion, I was not Confirmed; I don't think there was a great deal of belief in the household and we weren't brought up with any great emphasis on beliefs; I can't remember when Dad stopped being a lay preacher, but he did stop, joined a country dance band as a violinist and did country dancing, so we got dragged off to those; religion was never really important for me, I don't think I ever did believe; I would classify myself as between atheist and agnostic; on music, I heard a lot of Baroque music which remains with me; Dad tried to get us to play the violin, which was probably torture for him; he has perfect pitch and I don't think that any of the children do; Liz, the youngest sister, stuck it out the longest, but we all gave it up so there is no generating our own music any more, apart from him; I used to listen to a lot of music but got tinnitus from the nineties and that has gnawed away at my appreciation for music, particularly classical music, when the dynamics and sound isn't right and there is additional screeching which has nothing to do with the chorus; I still like listening to a lot of music, but more and more it is just the spoken word; I was never sporty; having to participate in organized games at school came to a sad end when they sent us out on the first day at grammar school for a cross-country run; I had a massive asthma attack and the games teacher had to find my Grandmother to get me taken home; after that I was on excused organized activities, walking round the playing field in good weather; it might have been better to have more organized exercise but I didn't; I walk now and have done so most of my life; walking is good because there is something to do and it takes your mind off the problem; I have this theory that to solve a problem you need to assemble everything into your head so that you know what you are trying to do; then you need to let the machinery crank; if you want new novel solutions to problems that haven't been done before then it requires a bit more of being de-focused; you have got the thing in your head, you start examining it and you thought there was a single solution and it turns out there is a big forest of solutions, and you need to select something somehow; walking is a good way of doing something while that process goes on without spending sleepless nights thinking about it; the other sort of thing is when you are already doing something and there is a tiny detail yet to be worked out; walking is not as good for that; sleepless nights are good at that one; drinking and so on are fine for the first type of problem which require the mind to be a bit de-focused; the second type of problem where you just need to work out some final detail, that's when the sleepless night really helps; you can run through the whole thing through your head, mounds of machine code running through your head; when BBC Basic was about to go off to ROM I'd spend lots of nights with all the code running through my head seeing what it did in my head and where the problems were; so there are two different types of thinking going on there; I can't believe drink helps with doing good coding; I had a bad experience with alcohol very young and haven't touched a drop since
26:58:15 At Cambridge I went to Selwyn College; it was a really bad time; the first year wasn't too bad and I got a 3rd; I didn't really know what a 3rd meant; it was obvious that Selwyn wasn't a brilliant college to be at; there were only six mathematicians there; it's a bit isolated from the centre of Cambridge which is sort of nice as I liked walking to and from lectures, walking through King's College or something like that; backgrounds, what people are interested in, when you have only got six people, the chances of a good overlap are quite slim; I felt very isolated coming from a northern grammar school; the second year was much more miserable because by then it was obvious that I didn't know much about maths and couldn't do a lot of it, even with quite a lot of help very kindly given by the other Selwyn mathematicians; at the end of that year I failed the maths course which was pretty bad; I had to go and see the Senior Tutor to work out what to do next, a little bit nerve-wracking; "So you have been through two years and failed quite badly as I understand it, so what's to become of you"? He basically cast around for things I could do; I could do all of this stuff that we had built for Mum's school physics lab, oscilloscopes and things - I could make things; perhaps fortuitously computers were beginning to get a bit more recognition, and most importantly of all there was a one year course; there weren't that many subjects that had one year courses, so I got signed up for the computing one year course; it was a one year course that counted towards my final degree with enough weight that if I passed it I might get a degree of merit rather than nothing - I ended up with a 2:2; this was probably in 1978; there were loads of lecturers in loads of different subjects; you take a large enough basket of things and pass them well enough; several things happened; obviously the first thing was that Selwyn having not many students doing computer science, in fact just me, didn't have any director of studies support so they found an external guy to take over tuition and review my progress, a guy called Paul Bond; mostly he was very tolerant because all of this was running alongside the other thing that happened, the founding of Acorn which occurred before my final year, and I first made the prototype for the Acorn System 1 during that Summer when we were in discussions about what areas might be my future; mostly Paul was tolerant to the fact that if the work set said 'Using any computer...' by which they meant any computer in the Computer Laboratory, I would go and use an Acorn computer; so he would get things in computers he barely understood
33:12:21 I got involved in Acorn through Hermann Hauser who came and found me; he wasn't looking specifically for me but wanted somebody who knew about low power electronics, which it so happened I did; so we have to go back to my age; as a June birth I was quite young at school generally; in both primary and secondary school the date you are born really does determine a lot of your capabilities because those people born in September are effectively nearly a year older than those people born June to August; fortunately for the world there are very few people born in those months compared with September, October, leading up to the big peak in February when there are lots born; anyway, the younger ones have a lot of trouble by being less developed; so at the end of secondary school I was still quite young; if I had gone to university I would have been very young, however taking the entrance exams for Cambridge meant that I had a whole term of being coached for the exams, took the exams in September-October, and then had nothing to do; so I was encouraged to go and get a job; I got a job with ICI Fibres Research in Harrogate without having any idea of what I was going to be doing there; ICI Fibres had put me in this small electronics research place with an odd group of technicians building things; that was OK because we were all good at building things due to the experience of building things for Mum's physics lab; I built them a few things just following the circuit diagrams I was given; I got told off by the Union rep that I was building them too quickly; that got me a little bit noticed and so when it came to building things that hadn't been fully designed yet they began to ask my opinion for things; so I started helping them design things which is only a small extension from building them in the first place; things that are incompletely designed, it's like building something with a few missing bits which you can fill in; so that was fine too; then they started wanting more complex things and I started reading up on the subject; I got the RCA CMOS Databook, I must have got it in 1974-5, though this copy was for 1972, for knowing about logic, parts and so on; that was good because I could introduce them to new families of logic; they were using logic called RTL which is different to today's RTL - resister transistor logic - and TTL had problems on factory floors because it's operating voltage was too low; so the noise in the electrical supplies in the factory made it very unreliable; so they didn't want to use TTL but here was this new stuff CMOS which had very high noise immunity; parts at the beginning were operating of 15 volts and CMOS requires about a third of a rail, so essentially had 5 volts noise immunity where the TTL circuits were operating off 5 volts and required about 0.6 of a volt to switch and had almost no noise immunity at all - 0.6 volts on the ground rail was very susceptible to noise; so I introduced them to all this stuff and I built them some circuits; that was great; I joined the Cambridge University Microprocessor Society towards the end of my second year, it did not exist when I arrived, and I wrote some articles for that which, not surprisingly, were about CMOS logic which was what I knew; thus, when Hermann was looking for someone to build him what he called an electronic pocket book, what we would now think of as the organiser function of a PDA or Google Calendar on your smart phone, he knew he wanted something that lived in his pocket, was low-power and ran off batteries; so he looked for somebody who knew about low-power logic and people pointed their fingers in my direction; so Hermann came and found me and wanted me to build him this electronic pocket book; that was sort of OK because I had been designing things anyway; I had my folder of designs of my own circuits so I agreed to do it and would go and visit him occasionally to keep him updated on progress; Hermann was located in 6 King's Parade, not very far away from the Maths Dept. so easy to drop in; he had set up a design consultancy called Cambridge Processor Unit which is Hermann's idea of a profound joke; they had won a contract from Ace Coin Equipment in Wales who manufactured electronic gaming machines, fruit machines, and they wanted to replace the existing controller which was electro-mechanical with a fully electronic one; Hermann had found other people to do that and I wasn't involved in doing this the first time round; they had built a dual processor, National Semiconductor SCMC powered device to run the first fruit machine; it was going to be subjected to some tests for rejecting the interference caused by Piezo electronic lighters and obviously being built of ENMOS circuitry, which the National Semi-conductor device is, it was actually quite susceptible in much the same way as TTL is, in fact I think it had lots of TTL in it as well; so it obviously needed some big patch on the outside so they brought me in for no reason that I can think of as somebody who could design something to fix it, and what I did was to design a very wide spectrum radio receiver that would notice that there was all this electro-magnetic interference coming and essentially press the reset button quickly; I did that out of the bits and pieces that were in their back room, which they liked; anyway, Hermann wanted progress on the pocket book so I was showing him circuit diagrams one day and I had taken my whole sheaf of designs with me and he wanted to know what the other circuits were; in there were all my other designs, and one of those was a personal computer like the recently launched Science of Cambridge MK14 which was National Semiconductor SCMC powered and there was a lot of commonality between Hermann's people and Chris Curry's people because they were living together essentially; that was Science of Cambridge; Chris was in the process of levering himself away from Clive Sinclair but hadn't split off entirely - Science of Cambridge is one of the Sinclair radionics companies; so I showed him my version of that and Hermann got very interested in the design, asking questions like whether it would work - of course it will; so during this Summer where I was negotiating with the Senior Tutor I also built the prototype for that - the Acorn System 1; so this was before I started the computer course; so while I was doing the one year Computer Science course I was also helping bootstrap Acorn into existence which may account for the 2:2 since a lot of time was spent at Acorn; the conversation with Hermann was along the lines of "here's the design for the display part of your electronic pocket book". "Will it work?". "Yes of course it will, it's like this design which I've done before". "What's that?". "That's my version of the MK14". "Will that work?". "Oh yes". "Can you build it?"; so he paid for me to stay in some extremely grotty room round near the fire station to build that, and of course it did work; CPU Ltd was the foundation that became Acorn, but that became Acorn's first Acorn machine, the Acorn System 1; Acorn was a trading name for CPU Ltd in the consumer space; so I designed Acorn 1 and also 2,3,4...; if I'd known you were interested in that I would have brought the prototype machine, I have it at home; so this was during the Summer; during the course I built much more stuff; the Christmas term and going home for Christmas I wrote the first Basic interpreter; I think running into Christmas I must have written the assembler-disassembler editor facilities; initially I called it BASIL but it became Acorn Basic, Acorn System Basic; the assembler-disassembler editor was called ADE; so that was all going on alongside the course
47:30:02 Before I even did the exams I had the job offer from Hermann; he came along to the graduation ceremony; so I have never done a job interview in my life; he took my parents out to the tea rooms at Granchester and they discussed money on the lawn; you won't be surprised to learn at a start-up it was remarkably impecunious; I was paid £100 a month in the first year which isn't very much; you can tell it was not very much because my first pay rise I had my salary doubled, so from £1200 a year I was earning £2400; that had never happened again but the next step was £3600, then £4800, so not exponential; you look back and can scarcely believe how on earth you can possibly live on £100 a month; basically if you didn't earn very much you didn't do very much; it did bring to an end habits of buying books and things because I didn't have enough money; I got a room in my Supervisor Paul Bond's house, and gradually inveigled him into Acorn's mysterious world; by the time we got the BBC contract Paul would be employed by Acorn to write the operating system; I designed it but he and Jon Thackray actually wrote it; Hermann is congenitally unorganized and late, that's why he wanted an electronic pocket book; when you are travelling with Hermann it is the most unpleasant feeling in the world; you are on the aeroplane, sitting next to an empty seat, and the pilot is announcing to everybody that they are sorry but are just waiting for one last passenger; Hermann always said he gave us two things that INTEL, Motorola etc. didn't give their microprocessor designers - he gave us no money and no resources, and hence we built something that became successful; Hermann is a hands-on, involved person who likes to understand what is going on, so at all the really key moments he has actually been quite highly involved; if you think of Acorn, when it got the BBC Micro contract it was quite a decently-sized company, there's a lot of stuff going on in such a company; Hermann as a company founder had some weaknesses; he spent the first year going and doing accountancy courses in the evenings just to learn enough to run the books, even so he found the time to be massively involved in all the technical stuff; like the Acorn system one year, he had to have the whole circuit diagram explained to him, and was involved in all the steps going forwards; in designing the BBC Micro he was involved in the decisions, calling us all into his office to hold our discussions including the key ones where it became a dual-processing machine; for designing the ARM, Steve Furber and I and Hermann used to walk down from Acorn to the local pub, discussing where we were with Hermann every lunchtime and walk back again, so he managed to be hands-on even for that; even though Steve and I didn't know what we were doing, and it was probably obvious we didn't know what we were doing, Hermann had actually worked with us all our careers; he recruited both Steve and I out of university, Steve having stayed to do a PhD first, so Hermann knew that we had never designed a microprocessor but was prepared to believe in us that we could, and we didn't even know that we could; of course, for building the BBC Micro computer he firmly believed that we could do it when we knew that we couldn't, and it was his belief that won out
53:25:14 There was not a call from the BBC for the computer but Chris Curry had got wind through his contacts in London that the BBC wanted such a thing and had been employing for over a year, various companies - one called NewBrain which was part of Sinclair Radionics at the time - to build a computer - and they hadn't been getting anywhere; Chris convinced them to come up and see the prototype of Acorn's next computer; at the time Acorn had the System range and the Atom - the Atom being another Chris Curry invention; we had got all the way up to System 4 I think, I'm not sure we had built System 5; Systems were range machines with lots of individual card in them - cards in a rack - so the overall System launches were quite rare, there were a lot of cards and you could build a double-decker rack of cards with a lot of functionality in them, and we had a lot of software and so on; we had been discussing ever since the Atom was moderately successful, which Chris Curry and Nick Toop had built in a back room, and we'd supplied lots of software to it grudgingly as we didn't think it was a very well-designed machine; so we had been discussing what a successor computer should be and had discussed like forever but nobody really knew what they wanted; Chris Curry wanted another home computer like at Atom; Andy Hopper, newly a Director of Acorn, wanted a scientific computer, Hermann wanted a business computer, I just wanted everybody to agree; so I came up with this idea that we would design a computer that was a dual-processor machine and so it could be made in parts and operable in parts - the part that everybody calls the BBC Micro computer is actually the IO processor of the multi-processor machine; so the concept is that you design that bit and then you add - originally it was going to be the National Semiconductor 16032 renamed 32016 - and that would be a scientific computer, or you would add originally an H286 and that would be a business computer for Hermann; that gave some semblance of agreement as to what we would do, but we had nothing and this was the nothing which Chris Curry had invited the BBC up to see; so with little more than a concept in our heads and obviously the knowledge of having designed all the System range boards, so we'd built video controller boards, memory boards, disc controllers, all sorts of boards for the System range; Steve and I had not been involved in arranging the BBC meeting as we had other jobs to do; Hermann rang us each up on the Sunday evening before the proposed Friday visit to let us know that this was going to happen; Hermann's line was a bit of pleasantry and chat with me thinking what does he want, why is he ringing me up late on Sunday evening? The conversation goes: "We've got the BBC coming up on Friday to see the Proton prototype", we had a code name for it; I said: "What prototype? There isn't one, that's impossible". Hermann said: "Well, no, but we'll have built one by then". "No, that's impossible". Then Hermann's master stroke: "Well Steve thinks that it's possible, will you give it a go?" "I think it's completely impossible but it Steve thinks its possible I will give it a go"; obviously he had said exactly the same thing to Steve; there was no hiding the deception because when we arrived at work on Monday morning it was immediately obvious what had happened, but anyway we would give it a go; so we had to get our ideas together pronto, design the IO processor part, so we got our ideas down and had a circuit diagram by Wednesday; in parallel we got all the parts ordered; now that was quite hard as we had quite advanced ideas and wanted to move on from what we had done before; so we needed a 6502 twice as fast as the existing ones, we needed a memory system four times as fast as the existing ones, in the System range you had both a static memory board and a dynamic memory board, we wanted to build a dynamic memory system four times faster; Steve had been reading the books and had seen something just launched, so we ended up the Hitachi rep. bringing by hand the first 4K x 16 dynamic rams of the speed grade we needed that were in Europe; that was quite sobering and we tried our best not to break them; there wasn't time to get a printed board circuit made; the traditional way of building prototypes was this prototyping solution called Verowire where you have small Eurocards and you wrap bits of wire round and solder them which melts the insulation of the wire; it's not very reliable and it's not very quick; so Hermann consulted with Andy and they brought in the fastest gun in the west, Ram Banerjee from the Computer Lab, who was a wire wrap gun expert; we determined to wire wrap the prototype and it took till early Thursday evening for the wire wrap to be built; Ram helped, but we all had to be there; it wasn't like we took a day off and was quite intense; so Thursday early evening we had a prototype but it didn't work; we plugged it into one of our System range machines with what is called an in-circuit emulator, a special board in the System range out of which comes a big cable that plugs in and pretends to be the microprocessor; so now you have a working computer with a connection to the non-working computer so you can write programs on the working computer that exercises functions on the non-working one; so gradually through Thursday evening we could find out what was wrong with each bit of it and fix it; that did take quite a long time because there were quite a lot of mistakes; if you have seen a picture of a BBC computer insides there are about 100 chips in there and the chips have 14, 16, 20, 40 legs, so the number of interconnections is quite high so the chance that your wire-up is a mistake or didn't work properly, there a quite a lot of duff things to find and it's increasingly hard for the more obscure bits; so somewhere around 2am Friday morning there was the machine still plugged into the in-circuit emulator on the System range computer, still resolutely refusing to function; at that stage I said that if they were going to come tomorrow and see a working machine somebody will have to write the software for it and that going to be me, I had better go home now as I am in no state to write any software; so I went off and I came back at 8am having had 4-5 hours sleep which was fairly helpful but not brilliant, so I felt a bit jaded; I found everybody sleeping under the desks in the lab, woke them and they said that the machine now works and that I better get on and write the software; Hermann's story is famous as well which is that he became convinced that it was actually the connection to the in-circuit emulator that was causing the prototype not to work and he convinced Steve Furber and Chris Turner to disconnect the emulator, put a proper microprocessor in there, and indeed the machine burst into life; Steve always says that is extremely unfair of Hermann and was more or less a fluke; anyway, I had a working computer but with no software on it; I portered across some of the Acorn operating system called Acorn MOS, basically enough to initialize the world, then portered in Acorn System Basic in order that I had an easier system than machine code to write the new stuff that was needed; by that time the BBC had arrived but obviously the machine still didn't do anything visible; so I wrote in BBC Basic, typing blind on the keyboard, enough instructions to configure the video controller to display a BBC machine mode zero screen which was a 640 x 256 black and white display, and having got that correctly programmed and an actual display, again still typing blind, I wrote something that could just display a random walk created graph; so by the time they came in there was a machine there working and displaying; phew; they knew that we had nothing and that we had built it in a week, so they awarded the BBC Micro contract to Acorn; I think that a lot of people were somewhat upset; I don't think that anybody was furious at the stage of the contract being awarded because it wasn't for very much; the contract originally stated that the BBC wanted Acorn to produce 12,000 machines and particularly in Sinclair's world, where he had already brought out the ZX80 and was working on the ZX81 Spectrum, 12,000 machines isn't a noticeable number at all; for Acorn it was quite a big thing to have a portion of our design sanctified with a guaranteed set of sales, but it also wasn't a huge number; it was the biggest single number that we'd ever have but it wasn't huge; the fact that in the first year we had manufactured 50,000 of the things, we thought that they were being a bit conservative but we hadn't predicted that; by the end of life there had been a million of them, those numbers were well in the future and nobody knew; I think it was only two or three years later that people started fulminating about Acorn being given the contract
1:07:51:21 The Domesday machine was considerably later; the BBC micro was built in 1981-2 and the Olivetti takeover was in 1986 and the Domesday project was 1987; the BBC Micro was quite successful; we started building all the second processors and they were quite successful too, so by late 1982 we had built most of the equipment that we had promised; we had a National Semi-conductor 16032 second processor, we had an 1886 one, Torch Computers had built at 68000 one, of course other people could do this stuff; all the leading microprocessors at the time we had either built ourselves or other people had built second processors for the BBC machine, and we could see that the promises for these processors were largely vacuous; so each manufacturer had made big claims for their machines saying how wonderful they were, capable of supporting high-level language concepts; what we found was that it wasn't true; when you looked at things a four megahertz 6502 and 8mhz 16032 and 8mhz 68000 all ran at much the same speed no matter the massive differences in their architecture; we theorized that the reason for this was for memory systems; we could design far better memory systems but on an 8 MHz 68000, for example, it took four clock ticks to run a single bus cycle and you got 2 bytes back, so it had 4 megabytes per second bus bandwidth; and the 4 MHz 6502 ran a bus access in one clock cycle and ran at 4 MHz and put 1 byte back, so it also had 4 megabytes; when you looked at it through that prism all these machines, none of which had caches on them of course, all ran at the same speed because their bus speed was low; we could build memory systems with much higher bandwidth easily, we did all the time, but we couldn't find a microprocessor to use them, and we went all over the world looking for them; at that stage it was an annoyance; we could build a high-performance memory system; at the time we could only build a 16 megabyte one but later on we thought of an additional wrinkle that would push it much faster, but where was the processor that could use that, there wasn't one; that was it for a while; we just remained unsatisfied; so two things happened; Andy Hopper put on my desk the very first publications of IBM, Stanford and Berkeley's RISC work; RISC meant 'reduced instructions set computer'; that is a bad acronym and what it should have said was reduced instructions set complexity; it was a way of designing a processor that was very different and it was all very interesting; we had been out to visit people's microprocessor labs, huge buildings stacked full of engineers all doing stuff; we could see that it was hard because National Semiconductors, I think is was revision E of the 16032, to be renamed 32016 before it even functioned, revision H before it didn't have any bugs that were very obvious, and something like revision K before it was actually shippable because it had lots of bugs, interrupts and so on that they hadn't fixed; so it seemed to be something that employed a lot of engineers and was very difficult to get right, so we didn't really think of designing our own at that stage; then we went to visit Western Design Centre because they had a new microprocessor and we wanted to know more about it; there was no internet so beyond what was printed in the papers the only way you could find out more was going there; Western Design Centre were important because they were the people who designed the 6502 originally although it was manufactured by lots of other people, Rockwell, Synotech etc.; they were designing a new thing which would be called the 65SC8 number 6 eventually and Acorn would go on to use it in the Acorn Communicator; we flew out to Phoenix, Arizona to find out about it; we were expecting Western Design Centre would be like all the other places we had visited - big buildings full of engineers; what we actually found in some back alleys down an unfashionable side of Phoenix were some bungalows staffed by some senior engineers and a bunch of school graduates, and they were designing a microprocessor; we weren't that impressed with the 65SC816 as it didn't seem to reach forward very much but our overwhelming impression as we left there was if they could design a microprocessor maybe we could; so it switched a little switch in our brains from we can't design one of those things to perhaps we can; so then you get to play fantasy instruction set design for a long time and get increasingly convinced that we do understand what we are talking about; that would have been late 1982; with the hat on which said perhaps we can design our own stuff I reread all the RISC papers and began to appreciate that here was a methodology that was amenable to a small team; I think it might have been Berkeley RISC 1, but certainly that had been designed by quite a small team; the Stamford MIPS was obviously designed by more and the IBM 801 was a big building full of lots of engineers; we began to be convinced that if we were going to do anything then it had to look more or less like this new way of designing a microprocessor, but mostly it's playing fantasy set games; so the conversation between me designing the architecture and Steve designing the microarchitecture is basically resolving what instruction is it advantageous to have; if there was an instruction I wanted that he couldn't work out how to implement, we wouldn't have it; if he knew how to implement something that I thought was interesting, maybe we could stretch the design of the instructions to include that sort of capability; there were lots of things in Berkeley RISC 1 and Stamford MIPS in particular that we couldn't work out how to design; at the same time there were lots of things that we could work out how to design and they looked very advantageous; we had a very different way of thinking about how computer instruction sets were used than the other teams, both Stamford and Berkeley, were designing computers to match the compilers they had; we didn't have a compiler so that wasn't a goal for us, but Berkeley RISC 1 and all the Berkeley RISC machines after it which led on into Spark, they were all designed around close examination of the Portable C compiler and that led them to make particular design decisions; Stamford made similar decisions based on better compilers; we had no compilers; we later discovered that we had build a machine that was compiler agnostic; it suited different types of language, different types of compiler quite well; you could write interpreters, both Prolog and Lisp compilers were written for it as well as conventional BCPL C type compilers and it suited them all which was good; all these people, INTEL, National, Motorola. have been trying to build machines that execute high-level languages better; so when the BBC Micro computer was built the number of transistors you could use to manufacture a microprocessor was quite low, so the microprocessor couldn't do very much; that basically meant that you programmed it at its best in Assembly language, and on the BBC Micro computer you got heroes doing amazing things with Assembly language - David Braben and Ian Bell writing Elite; so you had a 1 megahertz 6502 microprocessor with a fairly restricted amount of memory, and they were able to cram a galactic space trading game with 3D wire frame graphics with hidden line removal into that; so one of the interesting things we did quite early on - Hermann formed us into Project A, - Project B was to design the successor to the BBC machine, but Project A was to built the Acorn RISC machine, ARM - was to come up with some slogans to keep us honest; we had a slogan called MIPS for the masses which was to keep us focussed on the mass market; we weren't building a scientific processor that would have a low number of users; Hermann wanted us to have a MIPS per dollar rating that was good, cheap processor, millions of instructions per second; we also had a belief that we wanted everybody to be able to achieve results which were as good as writing Elite but to do that from a higher level language; that of course meant we needed big numbers of MIPS because high-level languages are quite wasteful in machine performance; so those slogans did keep us to a different line and kept us focussed on building things; the low-power bit, well we didn't have access to any tools which would reliably predict power; we wanted to put the processor into a plastic package so Steve became very leery of anything that would use power up, and all that engineering to avoid using power resulted in it being very low power; we simply wanted it because we wanted to build a computer that didn't have expensive heat sinks and power supplies; one of the lessons from building the big BBC Micro was that the places where there were heat sinks, like the video processor were hot and unreliable, and the power supply is a hot and unreliable bit; so if you can reduce the loads on those systems you have a better machine, a quieter machine; we wanted it to be cheap which meant a plastic package, and we didn't want to build an expensive power supply, and we wanted it in small packages as we wanted to build home computers with it, so that precluded exotic and expensive solutions; ARM originally stood for Acorn RISC Machine; when ARM was floated off as a separate company in 1990 they had to delete the word Acorn so for a while it stood for Advanced RISC Machine but eventually Robin Saxby decided it should stand for Arm so Arm RISC Machine; so Acorn RISC Machine dates from the beginning of Project A which was April 1983; back to the slogans; you can think about things better when they've got names as you can discuss them more easily, so it has always been very important, particularly for me to have a name for something; I now work for a company called Broadcom
1:25:24:06 On work, the obvious thing is that nothing complex it built by a single person, and nothing complex is built quickly; I always like to think of it like going into a gallery and looking at an enormous painting on the wall, and asking oneself how could Titian or whomever painted that, it must be stupendous, and a. it took a long time and b. he had a lot of helpers; so a. it took a long time to solve, and b. there were a lot of engineers and so on working on it; for ARM, although as is well-known I did the architecture and Steve did the micro-architecture, I also led a team of people to verify the correctness of the machine for verification and validation, being immensely important; so me, Hugo Tyson, David Seal, Jon Thackray wrote programs in our machine language to demonstrate the thing was an ARM, and of course we had to build the thing; so Steve had the micro-architecture but a team of engineers led by Robert Heaton, Jamie Urquhart, Dave Howard were also there; then we had to also write models of it; so I wrote a model in 6502 machine code. another one in 16032 machine code, Steve wrote a model of it in BBC Basic version of what the logic was going to do; we verified all those models did the same thing; then the software models we used to verify the verification programs and to begin porting compilers; so Lee Smith led a team of people - Harry Oldham, Richard Carry - to write programming tools for it; David Seal, Jon Thackeray started doing silly things like porting LISP and the adventure language for a large adventure game called Acton; meanwhile we were also building a computer, we weren't just building a microprocessor, so ARM was backed up but having an IO controller, a video controller and a memory controller; the IO controller was run by Mike Muller, the video controller by Tudor Brown, the memory controller by Alistair Thomas with more engineers; each participated in each of the other projects, Steve and I getting involved in both of them, so we actually build 4 chips; ARM came back first as we had started it first but all 4 chips came back and we had a 4 chip computer prototype later in 1985; so that is how it actually gets done, a big number of people, that was most of the Research and Development Department of Acorn at the time all designing this stuff
1:29:17:11 Andy Hopper became a non-executive director of Acorn fairly early on and was quite influential in pushing particular technologies, particularly towards scientific computing; in his own research life he did the Cambridge Ring and the Cambridge Fast Ring so he was quite influential when we built our own networks for Acorn, Acorn eConnect; he was indirectly also very influential because to exploit the Cambridge Ring designs he started up a company called Orbis and that had people like Mike Muller in it, and over time Orbis was fused with Acorn, and so Mike Muller is ARM's CTO today; so Andy has been quite influential in bringing advances in the world of technology from outside and into Acorn; Acorn was basically a place which grew things by itself; not exactly not invented here, but more a plethora of things are invented here and we don't have time for other things
1:30:54:04 On advice to potential computer science students, engineers get moderately well-paid, which is good, and they have a very interesting life because engineering is about creating stuff, even if it's recreating stuff because you are fixing somebody else's bugs it's still about creating things; so it suits creative individuals; the stereotypical engineer as portrayed on TV doesn't talk to anybody; engineering is all about communication, you have to talk to people, you have to collaborate in groups; engineering does not suit isolated reclusive individuals unless they are totally and utterly brilliant because engineering is a team sport