FileName: 20071210-ne-print-ecosystem

PARIS — Open-source hardware is not a novel concept. Because it was seen as a disruptive technology that could build the type of collaboration that Linux brought to the software world, a host of promising initiatives have emerged with the goal of fostering an ecosystem of freely available, usable and reusable open hardware. Years have passed, and the need for viable business models remains.

Although ‘The Open Hardware Certification Program’, ‘The Generalized Open Source Programmable Logic Initiative’ and ‘The Open Hardware Specification’ left a taste of defeat, there are other and more successful open-hardware efforts such as Power.org, a consortium promoting open development for IBM’s Power Architecture, and OpenCores.org, a community developed around open-source hardware intellectual property (IP) blocks.

OpenCores received a new impetus as ORSoC AB (Stockholm, Sweden), a design house that uses open hardware IP blocks, has agreed to take over the OpenCores brand and website. Launched in 1999 by Damjan Lampret, OpenCores’ website generates millions of page views per month according to the organization which also claims more than a million engineers have downloaded IPs from OpenCores in its first eight years of existence.

"In the future, we plan to put a lot of efforts in terms of resources and in finding partnerships to make sure OpenCores continues to live and moves to the next level," stated Johan Rilegård, CEO of ORSoC. "Trustfulness of the website is important for us. We want to control the use of IPs and offer developers more visibility."

"OpenCores’ challenge is to recognize the developments in the semiconductor, EDA and related industries, and to act accordingly," explained Lampret, specifying that time has come to define business models as they are essential to attract new developers, ensure mature IP and get OpenCores to a new level.

Lampret has an idea that he believes could revolutionize open-source IP. "Intel and AMD ask themselves how they can utilize the silicon resources they have on modern processor chips. Most of the silicon is used either for increasing caches or for adding additional processor cores. Very little of that silicon is used for improving single core performance. The next natural step would be to place an embedded FPGA array on a processor chip. This embedded FPGA array serves as application specific acceleration engine. Having one of these FPGA arrays in a desktop processor would dramatically increase the number of open-source IP developers and users, revolutionizing the open-source IP movement overnight."

Although open-hardware is an attractive option, its success is still heavily reliant on a feasible economic model and support. Lampret called for the creation of an ecosystem. On a high level, he said the interests of EDA, FPGA and ASIC vendors as well as silicon foundries have to be united and harnessed for efforts to develop and distribute open-source IP. On a low level, Lampret encouraged the development of enablers such as FPGA systems and open-source or free EDA tools, and to put them in the hands of developers.

"If open-source software got started 25 years ago and it really took off only 10 years ago, I predict that open-source IP will also take off when the ecosystem with the right enablers matures in about 3 to 5 years," asserted Lampret.

Taking the opposite course, Matja Breskvar, CEO of Beyond Semiconductor (Ljubljana, Slovenia), expressed doubts that the open source model could work for hardware. He declared: "The most important piece of open-source software without which there would be no Linux story is the GCC compiler. This provided software developers with free tools to work on their projects."

Due to the lack of free hardware development tools, the need for specialized equipment and know-how and the complexity to develop state-of-the-art hardware unless one works for a large company, Breskvar said it is impossible to reach the critical mass with hardware development.

"Even with suggested business model the success of open hardware is far from guaranteed," added Breskvar. "Other factors such as higher risks, higher costs, patent issues, slow development, etc. of developing hardware make it inherently less suitable for open-source development model."