Visit to Celoxica

Host: Colin Mason

Colin Mason is (according to his business card) "general manager and representative director" of Celoxica's [»] Japanese office. He had taken me to the CEATEC JAPAN 2001 exhibition three weeks earlier [»], so we were already quite well acquainted. I had also met up with him at an IEE evening meeting in Tokyo (!) the previous week.

Context...

Quoting from Celoxica's website...
 

Celoxica was founded in 1996 under the name Embedded Solutions Limited (ESL). In 2000, the company changed its name to Celoxica (derived from the Latin for 'rapid' (celox) and 'silica' to suggest "rapid to silicon") to better reflect its focus on enabling the rapid design of applications in hardware and offering a fundamentally new approach to design. The company employs over 100 people and is headquartered in Abingdon, UK, with additional offices in Campbell, California, USA; Yokahama, Japan; and Singapore.

Colin told me he came to Yokohama to set up Celoxica Japan in November 2000, having spent around half his time in 2000 researching the Japanese market. He managed to get the company up and running within a couple of months. There are now a total of seven local employees within the company, and Colin said they have already sold both hardware products and software licences to a number of major Japanese companies.

Celoxica occupies an open plan office within the British Industry Centre in Yokohama [»], which is an incubator for small UK companies like Celoxica who want to do business in Japan. There are currently around fourteen such companies within the Centre, which is situated on the 11th floor of the West Tower in the smart Yokohama Business Park estate near Tennocho station.

About Celoxica's products...

Quoting from their website again...
 

A new approach to hardware design Describing complex functionality in silicon is becoming harder and more time-consuming as chip circuits grow smaller in size and larger in number. Many of today's high-level tools struggle to address design complexity without compromising the user's control over structure and timing. Other high-level design approaches can create uncertainty by automating the partitioning of parallel and sequential code to generate output that is often large and slow. Celoxica's design methodology balances abstraction with the need to maintain low-level control. Its design suite is built around Handel-C - a sequential programming language based on ISO/ANSI-C that features a simple expression of parallelism. Handel-C is a powerful language that lets users take concepts quickly to netlist without an HDL step while maintaining the manual control needed to fully optimize designs. By introducing software methodologies of debugging and libraries, iterative refinement is both rapid and limitless giving design engineers the time and creative freedom to explore the best design alternatives. A reconfigurable electronics company Rapidly changing customer demands combined with long product development cycles create a tremendous risk for electronics product manufacturers in today's market. Through the use of off-the-shelf reconfigurable devices, manufacturers do not need to commit to mass production, lowering the overall cost and risk of investing in new ideas. Celoxica's technology reduces development time by exploiting reconfigurability so that products can be quickly created and easily reconfigured in the field to meet market pressures as new features are demanded or new standards become available.

Colin explained that there are three main approaches to high level design of digital circuits/systems. In some companies they predominantly use VHDL [»], in others they use Verilog [»], while in others they are starting to use one of a family of C-based approaches. Although Handel-C is derived from C, it is significantly different to other C-based languages in that it supports design using concurrent processes. Other C-based languages are purely sequential (like C itself) - designers express their designs as sequential algorithms and the tools then attempt to automatically parallelise as far as possible. However, Colin said that auto-parallelisation has proved to be a very difficult problem, and the tools that currently exist don't produce solutions that are particularly good. By contrast, Handel-C enables designers to implement parallel algorithms directly, and if this is done intelligently it can result in very high performance systems.

Colin discussed several different C-based languages, some proprietary. He said that Japanese companies in general often prefer to develop their own EDA tools, and that C-based approaches are relatively popular in Japan. He said that NEC recently announced that they intend to use C-based methodologies for a significant portion of their digital logic design in the future. He pointed out that NEC is one of the leading companies in Japan, and this statement may influence others. I looked this up later on the web, and found an EE Times Online article about this [»] which states that NEC's "announcement of Cyber is believed to be a trial balloon to gauge the industry's interest". Cyber is NEC's in-house C-based approach.

About the staff at Celoxica Japan...

Apart from Colin, all the employees in the office are Japanese. I was curious to know how he managed to find staff in Japan, given that Japanese employees are notoriously loyal to their companies (on the whole). Apparently the three engineers and one of the salesmen were head-hunted by recruitment agencies, while the other salesman and the marketing and admin staff were recruited through existing contacts. The atmosphere seemed very congenial, and I imagine it must be a nice place to work.

At least two of the staff had previously worked for foreign companies in Japan, and I was told that employees of foreign companies are often inclined to change jobs more frequently than employees of Japanese companies.

Forthcoming Celoxica tour of Japanese universities...

Chris Sullivan, a senior employee of Celoxica in the UK who oversees Celoxica's worldwide University Program [»] amongst other things, is due to come to Japan for two weeks at the end of November. They want to visit as many universities as they can during his stay, in order to promote Celoxica and its technology within academia. Colin pointed out that Celoxica is itself a spin-off from academic research, and that Oxford University actually owns part of the company. Partly because of this background, he said that they are very keen to introduce their technology into as many universities as possible. There are now over 600 universities worldwide using Celoxica's products in one way or another, however relatively few of these are Japanese - hence the forthcoming tour.

Colin's thoughts/experiences as regards education in Japan...

During our journey to the CEATEC JAPAN 2001 exhibition three weeks earlier Colin and I had discussed a range of subjects, and he told me some very interesting things relating to his experiences in Japan. I am reporting them here for convenience (and with his approval).

He told me that he worked for Mitsubishi Electric for nearly 14 years, originally in the UK but then in Japan for five years. He had taken voluntary redundancy less than two years ago, and subsequently joined Celoxica. I asked him about his experiences in Mitsubishi Electric Japan, particularly with regard to his impression of new recruits.

He said that, as a general observation, newly recruited graduates from good universities in Japan do not necessarily have as much in-depth subject expertise as equivalent graduates from good universities in the UK or the USA. Often they are recruited having studied subjects unrelated to their future roles. He said that one year Mitsubishi recruited a group of American graduates and some of the managers commented that they were quite astonished to find that the gradutes were already capable of chip design and familiar with the very latest tool technology.

Colin said he had the impression that many students enjoy quite a relaxed time during the four years in which they are undergraduates, and it often takes a while for them to adjust to the demands of industry. On the other hand, he understands that Masters students generally do work very hard at university, and he added that some companies will usually only consider applicants with postgraduate degrees for the most demanding technical roles.

Like other companies, Mitsubishi Electric has a rigorous training programme for new recruits that includes character building exercises and group activities. There are technical courses too, although much of the in-depth technical training is "on the job".

High school education is much too focussed on rote learning, Colin thinks, and this carries over into the university environment. He understands that students rarely challenge or even ask questions of their professors, although he added that this culture is probably beginning to change now.

Colin subsequently added "the rider to all of the above that these are only his observations up till now, and as always for the foreigner in Japan, every day still brings a constant flow of new surprises no matter how long you live here and as such it is dangerous to draw too many conclusions from each individuals experiences. In other words he reserves the right to be shown to be wrong!".

Takahiro's comments...

At the CEATEC JAPAN 2001 exhibition I also talked with Takahiro Ohara who is one of the engineers at Celoxica Japan. He was looking after the Celoxica stand that day. I was interested to find out something about his background, and he also told me some interesting things which I am again reporting here for convenience (and with his approval).

Takahiro told me he studied electronic engineering at Tokyo Denki University. ("Denki" is the Japanese word for electricity.) I asked him for his candid view about whether students at university generally work hard. He said some don't, but engineering students generally have to, especially in the more prestigious universities. The penalty for not working hard enough, he said, is that you have to repeat a year. He estimated that at Tokyo Denki about 20% of the students have to repeat a year at some stage, while at Tokyo Rika University (which specialises in science) perhaps between 40% and 50% of the students have to repeat a year.

Takahiro told me that in his first job after graduating he was a pachinko (fruit/games machine) engineer. A year later he joined a subsidiary of Fujitsu, and he stayed with them for five years before leaving to join Celoxica. I said that I had the impression most people who join a big company like Fujitsu stayed there until they retired. He said that's true for many people, but things are changing. One of the reasons he decided to leave was because he wanted to spend some time travelling in Africa, but there was no way he could do that while employed by Fujitsu. At Celoxica he can take a holiday more or less whenever he chooses, but at Fujitsu he couldn't. His holiday entitlement at Celoxica is the same as it was at Fujitsu - 20 days/year - but he said that at Fujitsu he wouldn't be able to take more than 10 day (at most) in one go.

It was his opinion that there are many jobs available for electronics engineers in Japan, in spite of the recession. Any relatively young engineer leaving a large company should be able to find employment fairly easily with a smaller company, although re-joining a large company may be out of the question.

I asked Takahiro whether he was involved in a training programme when he joined Fujitsu, but he said no. For one thing, because of his previous experience he wasn't strictly a "freshman". He also said there was generally a lot of work to be done, so most new recruits had to start working on projects straight away. However, he said that there was a selection of short courses available internally that he could choose from.

Acknowledgement

I am indebted to Colin Mason for checking this report, correcting some errors and supplying some additional details and text.