IP Cores: Rick Clucas
Only temporary obstacles exist

I don't recommend chemotherapy," said Rick Clucas, as we sat down to a late lunch in Edgware, North London. It was only his second week back at the office after an enforced rest of several weeks, he told me.

It turns out that 1998 has been a busy and difficult year for Clucas. In midyear his wife gave birth to their second son-that was a high point-and shortly after that Clucas was diagnosed with cancer.

Clucas' wife and baby are both fine, and so is Clucas himself, after a session under the surgeon's knife and enduring a program of chemotherapy to ensure that his body is clear of any cancerous cells.

And while all that was going on, Clucas was trying to complete the process of taking ARC Cores Ltd., a young company pushing the edge of the intellectual-property revolution, independent from Argonaut Software, the PC games company where he started his career and was technical director.

Clucas, 30 years old, is chief technology officer and co-founder of ARC Cores Ltd.

His company licenses the ARC 32-bit processor core to ASIC designers, claiming its configurability as a key advantage over other processor cores.

The forced rest has given Clucas time to think about the latest trends in the intellectual-property (IP)-cores field in which ARC Cores operates.

How ARC Cores came to be a pioneer in the IP market is a consequence of Clucas' personal history-from schoolboy computer enthusiast, to games programmer to graphics chip designer, and the different perspective he has of chip and systems architecture.

Soft core
When the ARC 32-bit RISC microprocessor core debuted at the Design Automation Conference in 1996, it was unusual because it was being offered in soft, synthesizable form. Plus, its instruction set was extensible.

Clucas observed, "Now everyone is saying they have a synthesizable core. We're finding less people are saying that a hard macro is better these days."

According to Clucas, that's partly because of issues over integrating hard layouts defined in a physical format with the synthesizable flows used elsewhere in the design process.

Clucas thinks that software engineers are driving projects to a much greater extent as design teams become less rigidly partitioned between hardware and software. So the version of the configurable ARC core eventually selected is not necessarily a hardware-engineering choice.

Clucas said that software engineers are selecting extensions in the ARC Metaware C language compiler and simulating their code on equivalent VHDL models generated by the Configuration Wizard. Once that is working to the team's satisfaction, the processor choice has been made.

"The real vision here is that every product requires a solution based on both software and hardware. By being able to vary the hardware and software, you can make a more efficient solution. We also find this approach is much less work.

"If you select a hardware platform and just write software you end up writing a lot of code just to get 'round the deficiencies of the hardware."

As he says this, it is clear Clucas is thinking back to his days as a games programmer struggling with Zilog Z80 and Motorola 68k processors.

"Less hardware means less software and less power consumption. But if you build up a complex system from components, you end up taking a lot of time." Clucas cites the trend toward combining DSP and CPU cores on a single chip as present-day examples. "Everyone agrees that DSP plus CPU is crazy; what you really want is an integrated solution tuned to the application. What companies are doing now is pulling two cores together and calling it integrated. Some firms do provide integrated support but that's all."

Clucas contends that such dual cores spend a lot of time and computation cycles communicating with each other for the sake of the architecture rather than to the benefit of the end application.

He makes it clear that his ideal is one in which the system requirements drive both the hardware and software parts of the solution.

"A lot of people haven't realized this when looking at system-on-a-chip. They haven't realized that all the rules have changed. If you go in fresh, you do it in a totally different way [from] those people who have been working on it for 20 years."

Doing things in a totally different way just about sums up Rick Clucas.

An unusual history
Clucas may be the chief technology officer at ARC, but he has never been to a university.

"In the late '70s I had built a Sinclair Z81 Spectrum with my father," said Clucas, speaking of a time when he was about 11 years old and experimenting with one of the United Kingdom's early hobbyist computers, available by mail order.

At secondary school Clucas found himself in his element.

"Our school was very early into computers. We had an Altair 8080 system in 1985." So Clucas' homework assignments became the building and maintenance of the school computer system. "Maybe that's why I have had a systems point of view from the very beginning," he said.

On leaving school, instead of going on to a university, he took a job at Argonaut Software, a computer-games company he had been doing some work for while still in school. In fact, at 17 he was the first employee of Jez San, founder of Argonaut, who was himself only 19 at the time, said Clucas.

At Argonaut, Clucas wrote games for the Atari ST and Commodore Amiga platforms and Argonaut did well in those early years. But in 1989 a new force was emerging: Nintendo.

Clucas said, "We wanted to get into the Gameboy market, but we were not a licensed Nintendo developer and therefore could not get a development kit. Nintendo was Z80-based, so we built a ROM emulator for the Gameboy, called RAMboy, which meant we could develop on a 68k-based Amiga platform."

'We were naive'
Clucas makes it sound simple, but in essence Clucas and San bought a Gameboy, cracked open the case, desoldered the ROM, attached a connector and then hacked their way into the system and wrote software to run on it.

"I guess we were naive. Nobody told us you couldn't do that or that it was meant to be difficult, so we just did it."

Argonaut then approached Nintendo and, in effect, said, "This is what we were able to do without any help from you. Think what we could do with a developers' license."

Clucas takes up the story. "Nintendo signed a contract with us for a 3-D game and we had something to show at the Winter Consumer Electronics Show in January 1990. Later that year we told Nintendo that the 3-D graphics would be much better with a graphics accelerator, and that we could design it for them."

So Clucas the computer programmer became Clucas the chip designer, building a three-chip prototype based on Actel FPGAs prior to moving the design to a gate-array ASIC. The ASIC, dubbed SuperFX, was designed to work with Nintendo's 16-bit platform as a graphics accelerator but was included as an extra chip within the game cartridge along with the ROM, starting with the Star Fox game.

After Star Fox was released in 1993, Argonaut formed a multimedia and chip-development subsidiary and started doing consultancy work for companies such as Apple and Philips, while developing a 32-bit configurable architecture for graphics and other applications.

It was that subsidiary and architecture that has evolved to become ARC.

Return to: TimesPeople 98