Host: Prof Hiroto Yasuura
Prof Yasuura [http://kasuga.csce.kyushu-u.ac.jp/~yasuura/] is another influential figure within the field of VLSI design in Japan; he was one of the instigators of the Silicon Sea Belt project (see below), which seemingly has the potential to become a very important alliance in the world of silicon chip design and manufacture.
Prof Yasuura told us that Kyushu University is the third oldest national university in Japan, after Tokyo (the oldest) and Kyoto, and that he had moved to Kyushu University from Kyoto University about ten years ago. According to a brochure he gave us, the Graduate School of Information Science and Electrical Engineering (ISEE) is one of 14 graduate schools within the University. Nine of these are each associated with one of the ten undergraduate "faculties"; the other five are independent. ISEE, established in 1996, is one of the independent graduate schools.
According to the brochure, "several related subjects, thus far pursued independently within the Faculty of Engineering, the Faculty of Science, and the Interdisciplinary Graduate School of Engineering Sciences were integrated to launch the Graduate School". Prof Yasuura said that the structure of the University was changed about five years ago – coinciding with the establishment of ISEE - and he mentioned that there may need to be more restructuring due to the current "big storm" of semi-privatisation which is affecting all national universities.
Close links between hardware and software engineering at Kyushu University
ISEE consists of five departments:
- Dept. of Informatics
- Dept. of Intelligent Systems
- Dept. of Computer Science and Communication Engineering
- Dept. of Electrical and Electronic Systems Engineering
- Dept. of Electronics
Prof Yasuura said that the students in the Dept. of Informatics are mainly computer science graduates while the students in the other four departments are mainly electronic engineers or computer/communications engineers. Although these departments are separate entities from an administrative point of view, Prof Yasuura said that there are cross-department research groups - resulting in significantly closer integration between the hardware and software domains than is usual. Chris commented that he thought this was a very good thing, and that he believed this is the way of the future. (Celoxica’s technology, which involves FPGA design and reconfigurable computing, is of course a prime example of how hardware and software can overlap.) Chris said that while he had seen similar developments at some other universities, particularly within the USA and UK, in general he saw little evidence of integration between the worlds of hardware and software within most universities.
Joint laboratory for students
As far as teaching is concerned, Prof Yasuura explained that he works very closely with two colleagues (Prof Matsunaga and Prof Murakami) in the sense that they have set up a joint lab/ supervision system in which students are free to move around depending on their supervision requirements [http://kasuga.csce.kyushu-u.ac.jp/YMI/]. He said that there are currently around 30 students in this (these?) joint lab(s). From their website it is apparent that twelve of these students are doctoral students, which is very impressive given the numbers of doctoral students I have seen at other institutions (Bush, 2002). He added that it is very unusual for professors to work together in this way, even at Kyushu University. Prof Yasuura and Prof Matsunaga are in the Dept. of Computer Science and Communication Engineering, while Prof Kazuaki is in the Dept. of Informatics.
Contribution to VDEC’s activities
Kyushu University hosts one of VDEC’s nine remote "license servers" (see first Tokyo University visit report). Prof Yasuura said that VDEC is just one example of the way universities have started to collaborate in the setting up of national networks for the benefit of all. He added that private universities were originally not able to participate in VDEC’s activities, but the Ministry of Education recently changed their stance on this – partly in response to pressure from industry with regard to the need to enhance the education of engineers within private universities as well as public universities.
Prof Yasuura said that since VDEC is under-staffed, Kyushu University are also providing assistance in the form of developing standard cells for VDEC’s cell library. He said that they use commercial EDA tools for this, and that they have contributed over 200 cells to date. He added that Kyoto University have also contributed standard cells to VDEC’s cell library, but unlike Kyushu University they have been using EDA tools which they have developed themselves.
The Valen-C language
Prof Yasuura told us something about the research he has been involved in which focuses on System-on-Chip (SoC) design methodologies. He and his colleagues have developed a parameterised soft-core processor as a SoC platform that is customisable in terms of data path width, number of registers and instruction set architecture. They have also developed the "Valen-C" design language [http://kasuga.csce.kyushu-u.ac.jp/codesign/Valen-C/] to enable users to engage in hardware/software co-design targeting their soft-core processor. The user should first describe their algorithm in C, then translate to a Valen-C program. The translation involves explicitly defining the bit-width of each variable, and partitioning the system into hardware and software parts - i.e. circuits and machine code.
Prof Yasuura subsequently added that "our compiler generate an object code for a soft-core processor, parameters (the data path width, the number of registers etc.) of which is specified, to preserve the computation accuracy of the program. We also developed an optimization method of the parameters of soft core processors to minimize chip area and/or energy consumption".
The Silicon Sea Belt project
Prof Yasuura gave us an overview of the Silicon Sea Belt project [http://www.ist.or.jp/lsi/english/] which aims to bring about collaboration between the following centres of silicon chip production in Asia:
- Kyushu (Japan)
- Okinawa (Japan)
- Shanghai (China)
- Hong Kong (China)
- Taiwan
- South Korea
- Malaysia
- Singapore
Prof Yasuura pointed out that over 30% of the world’s silicon chips are currently produced in these centres of activity, which stretch (roughly speaking) along the east coast of the continent of Asia. Shanghai is currently the least developed of these, but Prof Yasuura said that it is developing quickly into a major centre of production. He said that there 30 or 40 fabrication lines are now starting in the Shanghai area, using 0.35 or 0.25 micron technology; some are already in operation, others are still under construction. While this isn’t state-of-the-art, he said he expects them to have progressed to 0.1 micron technology within five to ten years. He pointed out that Taiwan and Shanghai are about as close to Fukuoka as Tokyo is, and South Korea is even closer. He also mentioned that the working language for the Silicon Sea Belt project would be English.
While Prof Yasuura seems to be one of the instigators of the project (he told us that he devised the name, for example), according to the website the project was initiated by Fukuoka Prefecture, which with a population of around five million is the most densely populated prefecture on the island of Kyushu. Kyushu has seven prefectures in all, and a total population of around twelve million – about 10% of the total population of Japan. Over 30% of all the silicon chips produced annually in Japan come from Kyushu, yet until recently very few chip designers were based in Kyushu. According to Prof Yasuura this was the main reason why the Silicon Sea Belt project was initiated. He added that they’re developing the Fukuoka/Okinawa relationship first, and that he already has many contacts in Taiwan and S. Korea.
In spite of the fact that the Internet is enabling easier inter-communication, it seems that with the introduction of SoC technology the traditional separation of chip design from chip fabrication is becoming harder to maintain. This is because SoC technology involves the integration of many different kinds of circuit within one chip, such as DRAM, logic blocks and analogue circuitry. The design of a SoC involves many trade-offs, forcing closer collaboration between many different specialists – including analogue circuit designers, memory design specialists, logic designers, embedded software engineers and process engineers. Also, ever-decreasing time-to-market – Prof Yasuura said that a 4 to 5 month "design window" is now common – and the need for quick feedback from testing also make it more important for the designers to be physically close to where the chips are fabricated.
Prof Yasuura told us that he went on a study tour last year together with the governor of Fukuoka Prefecture to see the Alba Centre [http://www.albacentre.co.uk/] in Scotland, and that they had been very impressed by what they saw there. Chris said he was familiar with it too, and that he knew from past experience how crucial it is to have the right funding mechanisms to foster collaboration rather than competition. He added that the funding mechanisms in England are not the same as those in Scotland.
ISIT – the Kyushu Institute of Systems & Information Technologies
Prof Yasuura told us a little about ISIT [http://www.isit.or.jp/index.en.html], which is sponsored by Fukuoka City Government. He gave us a brochure that states "at ISIT we conduct medium – to long – term research and development with strategic themes. We also coordinate partnerships between major companies and local companies, as well as between industries and academic institutions, for example between companies and universities". ISIT is located in the new Fukuoka Software Research Park, which occupies part of Fukuoka’s waterfront. According to the brochure, six major manufacturers and many local companies now have offices there. Prof Yasuura said that the researchers are backed up by a lot of support staff, and that many of his students work on projects connected with the ongoing research at ISIT.
Prof Yasuura explained that when he came to Kyushu University (around 1991) there was virtually no local job market for graduates of computer science or electronic engineering. Kyushu Electric Power Company [http://www.kyuden.co.jp/english/English-Home.html] was the only local employer in the industry at that time, but its annual intake of graduates was small. After lengthy discussions with academics, and feasibility studies, Fukuoka City decided to create ISIT as a way of stimulating a local hi-tech job market. The belief was that many graduates would like to stay on in the area, given the opportunity, and that others would also be attracted by the relatively high quality of life that Kyushu has to offer in terms of unspoiled countryside and low land/house prices. Chris added that graduates do indeed often like to stay on in the area where they studied; he mentioned that exit surveys from the University of Sheffield (where he used to work) showed that over 60% of graduates would prefer to stay on in the Sheffield area - although the job market doesn’t enable this.
The strategy seems to have worked. The high-profile research at ISIT has attracted many of Japan’s leading high-tech companies to open offices in the area, and Prof Yasuura said that now there is a very healthy job market, with around 50% of Kyushu’s SoC-oriented graduates now finding employment within Kyushu. (Prof Yasuura added that out of an annual total of 160 ISEE graduates, around 20% find employment within Kyushu.) It seems very impressive that the situation should have changed so radically in only ten years.
Fukuoka System LSI College
Prof Yasuura also told us of a new private teaching institution - of which he is the president - that has recently been set up with financial support from Fukuoka Prefecture, Fukuoka City and Kitakyushu City. It is called the Fukuoka System LSI College, and opened on December 3 (2001). It is located within the same building as ISIT. Hitachi, Seiko, Sony and other industrial companies are customers of the college; students come from these companies supported by grants from the Japanese government to promote the restructuring of industry. They now have 70 students for four courses - Analog, Digital, Layout I, Layout II - in the first quarter (December 2001 - March 2002).
Prof Yasuura pointed out that traditional universities in Japan don’t offer re-education courses for mature learners. The new institution offers short courses to people already who already have a good understanding of the general area (usually because they are already employed within the industry) but who want to diversify or update their skills. He said that, for example, someone working in logic design might choose to undertake a course in chip layout or analogue circuit design. The courses range from three weeks in duration to six or seven weeks (involving 130 hours of study).
Prof Yasuura explained that being "private" means it does not come under the jurisdiction of the Ministry of Education (MEXT). The teaching staff for the first quarter consists of 19 professors from 14 universities all over the country, as well as 5 from industry. They have hired teaching assistants from several universities, including Kyushu University.
He said that there are no plans to offer undergraduate degrees at present, but this might be a possibility in future – although in that case the institution’s activities would then presumably become subject to scrutiny from MEXT.
Miscellaneous
I noticed from the literature Prof Yasuura gave us that there appear to be similar numbers of full professors and associate professors at Kyushu University. This contrasted with what I'd seen in previous visits to other institutions (Bush, 2002) - I had the impression (prior to this) that there were generally more full professors than associate professors (or "assistant professors", or "instructors") within Japanese universities.
Acknowledgement
I am indebted to Prof Yasuura for his feedback on this report.
|
|