Oyo-Buturi International Interview few books on to the subject. One was Den- work? shi-Keisanki (obi: “Electronic Computer”) Dr Shima: I did it for about four months. Dr Masatoshi Shima was part of a talent- by Shigeru Takahashi, which outlined the The next development in my career oc- ed group of engineers who in 1971 de- system, architecture, instruction set and curred as a result of my being lucky or as we veloped the world’s first microprocessor, microprogramming of computers; almost say in Japanese unmei (obi: fate or destiny). the 4004. In this interview, Dr Shima everything concerning computers. Another OBI: What do you mean by that? sheds light on some of the critical events book I read was about logic. It was written Dr Shima: Well, although the transistor leading up to the development of the by Professor Udagawa. I read both of these was invented in 1947, it was not commer- technology that revolutionised the elec- books avidly and then began to design the cialised until 1951. The commercial use of tronics industry and society as a whole. circuit boards that go into a calculator. This the transistor then led to a new era, namely, process involves connecting ics with wires the “era of the circuit”. That is to say, if you OBI: You studied chemistry as an under- and designing complicated wiring patterns. could fabricate a circuit by putting together graduate but then joined a com- a transistor, a resistor and a di- pany working on calculating ma- ode, you could construct and de- chines. Why did you change velop a “system”. Furthermore, fields? in 1961 the first commercial ic Dr Shima: I studied chemistry at was developed. This led to the university because I liked it and “era of logic”. It was at the begin- was interested in the subject. But ning of the “era of logic” that I I graduated in 1967 when the first started working on the de- Japanese economy was weak. velopment of calculators in in- There was not a lot of demand dustry. So I was lucky that the for chemistry graduates, so I had tools for development were al- to consider other fields of em- ready available when I started ployment. It was at just this time work. Incidentally, it’s interesting that an “electronic computer” to note that there were periods of was installed at Tokyo University. exactly 10 years between the The machine was supposed to be commercialisation of the transis- able to calculate the structure of tor in 1951, the ic in 1961, and compounds. It was in stark con- the microprocessor in 1971. trast to the mechanical calcula- In 1968, one year after I grad- tors that I had been using. The uated from university, my com- idea of an electronic computer pany decided to relocate its was very intriguing, so, after con- R&D to Tokyo. So I, too, re- sulting a friend who was working turned to Tokyo. there, I joined Busicom Incorpo- OBI: Did the move affect the na- rated. Busicom manufactured ture of your work? calculating machines and other Dr Shima: Yes. In 1968 Japan such office equipment. had won a major share of the OBI: What kind of work did you world calculator market, mostly do when you first joined Busi- based on (oem) agreements. Bu- com? sicom also carried out oem-relat- Dr Shima: I first learnt how to ed business. Their major concern write a program. I did not have was that most of the calculators any previous experience in pro- were made using “hard-wired gramming but I managed to pick-up the There is also the inter-board wiring. If logic” which would be difficult to use in basics quite quickly. I did that kind of work there are, say, ten boards then they all have the longer term. Our technical manager for about six months but I found that the to be physically wired up. There were two wanted an alternative to this technology applications side of the work, such as busi- main problems to be overcome in this and, because I had some relevant experi- ness applications, did not interest me work. The first was how to transfer the sig- ence, I was asked to study the introduction much. nal with a minimum of losses. The second, of the “programming” approach instead of OBI: So what did you do instead? was how to devise a process whereby the the “hard-wired” approach. “Stored pro- Dr Shima: Well, Busicom had a manufac- wiring up could be undertaken according gramming logic” as it was called. So I had turing subsidiary working on calculators to a set of instructions by people who were the task of defining the instruction set nec- and I asked to be moved there. So, in Oc- often not qualified engineers. This work re- essary for the development of calculators; tober 1967 I moved to a place called Ibar- quired a lot of new thinking and many as- the programs and instruction set would aki in Osaka and became involved in work pects of the ideas were original. I believe then be given to hardware engineers for im- related to the development of calculators. that no matter what the task may be, it is plementation. Since it was also necessary to OBI: What was your approach to this new always important to try to do something connect a printer to the final version of the field of development? creative and original in your work. calculator, I had to take that into account Dr Shima: I decided to begin by reading a OBI: For how long did you do this kind of as well. That was in the Spring of 1968. OBI: How did you approach the new real problem with the designs. America I met Ted Hoff. He said he going project? OBI: So what did you do? off on his vacation- a big surprise for me. I Dr Shima: I succeeded in designing a sys- Dr Shima: I continued to try to explain the then met Federico Faggin, who had started tem based on “stored programming logic” designs to him for a few months but still at Intel only about a week before. He said and tested it with a printer. It worked very without making any progress. So I then de- he had not been told anything about the well. Then, towards the end of 1968, a ma- cided to talk about the programming of project and asked me to explain it to him. jor Japanese company announced that it desk top calculators, instruction sets, such Another surprise. When I asked why they had developed a calculator using only four as the binary instruction sets and the deci- had not hired more people for the project lsi chips; truth to tell, it was actually five. mal instruction sets, flow charts and so on. and I was told that they had not been able This caused quite a stir in industry because Thankfully, Ted Hoff understood all this find any suitable people. So I decided to at the time no Japanese companies had the and he agreed with some of the new ideas participate in the development myself once technology to make high performance lsi and approaches that I had proposed. He again. chips. The result was that almost all the pointed out, though, that computers based There were four chips and I decided to major companies went to the usa to find on the decimal instruction sets are more develop the 4004 microprocessor by my- partners to fabricate ics. complicated than those based on the binary self. The others parts were to be developed OBI: And what did your company do? instruction sets. by Federico Faggin. Dr Shima: The company’s Osaka branch, It was towards the end of August 1969, OBI: How long did this stage take? which worked on small pocket-sized calcu- that Ted came into the room with several Dr Shima: There was already a working lators, chose mostek. Tokyo, which pieces of paper in his hands. He looked processor in December 1970 when I left for worked not only on calculators but also on quite excited about something. He said Japan via Europe; we then entered the last business machines such as billing ma- that he “had come up with an idea”. phase of designing the calculator. The final chines, approached Intel. OBI: What was his idea? product was ready in March 1971. OBI: How did you choose those two com- Dr Shima: There were “three boxes”. One OBI: How did you feel after this achieve- panies? of the boxes was a 4-bit alu for “making ment? Dr Shima: The decision was based partly micro-commands out of macro com- Dr Shima: I was more deeply moved by the on information provided by consultants mands”. The other two boxes were the completion of the calculator development but, more important, the Busicom presi- “stacks” and “registers” which could be project than by the microprocessor. I think dent of the day had a very good “eye” when used as “address pointers”... So the 4-bit it’s because the original goal was to develop it came to judging potential business part- alu would use “micro commands” for pro- the calculator. The microprocessor was ners. He said that the key requirements for ducing “macro-commands by program, us- merely a means to that end. So, if you were a successful project would be reliable top ing micro-commands”. That was the idea. to ask me why the microprocessor was management and expertise in silicon gate I think that this could be considered to be- born, I would say it’s because we had an ap- technology, instead of metal gate technolo- ing the “birth of the microprocessor”.