A company spawned from the carcass of the UK’s defunct Acorn Group Plc has secured first round venture funding of $13m to develop a new generation of high-performance, low-power and more easily programmed digital signal processors (DSPs). Cambridge-based Element 14 is being backed by a venture capital keiretsu led by Europe’s Atlas Venture, and by the Boston, Massachusetts-based Bessemer Venture Partners, in a deal thought to be the biggest ever first round backing for a European silicon venture.
Element 14 emerged from the ruins of Acorn in January when it was created as a new identity for the ailing group’s computer business. It was originally focused mainly on development of set-top boxes to terminate high-speed services such as ADSL, but that part of the business was sold to Pace Micro Technology Plc for a consideration believed to be around 600,000 pounds ($970,000).
Since then, says Element 14 founder and CEO, Stan Boland, the company has worked with investment banker Morgan Stanley to build a new business plan aimed at exploiting the market for power efficient but fast DSPs in a market that analysts estimate is growing at between 30% and 40% compound per year, and which they predict will outstrip the larger embedded system space by 2001.
Element will face some stiff opposition in this arena, not least from established majors like Texas Instruments, today’s DSP leader, and Lucent Technologies. But Boland claims the company’s team of silicon design and software engineers has come up with a uniquely power efficient DSP architecture, which is also significantly easier to program than conventional embedded signalling devices.
Calling on silicon design and software tool expertise inherited from Acorn, and from the fortuitous recruitment of seven leading designers from STMicroelectronics’ Chameleon 64-bit processor project, Element has picked up an ARM (another Acorn spin-off) architecture, already heavily modified by STM and developed it further. According to Boland, the result is a DSP well ahead of commercial products in terms of power efficiency and four to five times ahead of commercial product in terms of operations per second.
Some of these claimed performance achievements stem from Element’s own-design point tools, which Boland said will enable the company to resize transistors on chips, and generally give it the flexibility to put more transistors in the right place, and dispense with redundant transistors. This will produce denser chips, greater on-chip device utilization, and smaller die sizes, said Boland. However, he remained tight-lipped about other aspects of the new architecture, and says these will be made available over time, but not at the forthcoming Hot Chips exhibition.
The $13m funding is now being poured into delivering an evaluation device by the middle of next summer, but before then Boland said he hoped to put in place significant partnerships with relevant tool and software suppliers. These could come from some unusual directions, given that one unique selling point of Element’s approach is the claim that its DSPs will be open to programming using high level languages such as C and C++.
This is a significant departure from common practice, which has seen most DSP builders pursue VLIW (very long instruction word) designs, which have made it almost impossible to program their devices using conventional languages and compilers. Instead, Element has stuck with an LIW (long instruction word) approach, and liberal use of SIMD (single instruction multiple data) technology.
Whichever companies are enlisted to help Element ready its technology for market, they will not be encouraged to take a stake in the company. Boland said he has already turned down several approaches from inside the trade, claiming that the expertise available from the companies wide selection of first round backers should be sufficient for the time being.
As well as Atlas, which led the funding’s organization, Element has picked up the support of Bessemer’s Rob Soni, an acknowledged expert-picker of silicon development investment opportunities, and a contribution from Japan’s Nippon Investment & Finance Co Ltd. Nippon has only ever backed one other high-tech start-up in Europe, and that was ARM, the other Acorn spin-off which is now making waves in the low-power core processor space.
Coincidentally, Boland was heading up Acorn at the time of its ARM divestment, and his involvement in the new Element venture is bound to attract comparisons with ARM, and with that company’s founder and ceo, Robin Saxby. Indeed, one VC executive connected with the Element funding said that there are three major factors behind ARM’s success: Robin Saxby, Robin Saxby, and Robin Saxby but, he added Boland is better.
High praise indeed, but comparisons with ARM may yet prove unhelpful, not least because Boland would not be drawn on whether Element plans to transfer ARM’s intellectual property driven model for the core CPU business to the DSP space. For the time being at least, Boland will only say that his company has enough funding in place to complete the development of its first product, and will seek a second round of funding, probably by the end of 2001, or early 2002.
