With the launch of the newly formed Power Forward Initiative (PFI), Cadence Design Systems opens yet another chapter in its long history of driving standardization in the EDA industry. This time, Cadence has set its sights on low-power design. Having identified a need—or depending on your point of view, an opportunity—in the low-power arena, Cadence has gathered an initial set of eight partners to build an industry-standard approach to the design, verification, and implementation of low-power systems-on-a-chip (SoCs).
For Cadence, along with Advanced Micro Devices, ARM, ATI, Freescale Semiconductor, Fujitsu, NEC Electronics, and TSMC, the objective is to spawn an industry-wide collaboration that will take on the myriad challenges of low-power design. The first step will be to create a unified specification format (known as the Common Power Format) with which to express low-power design intent. The members of the coalition hope that the specification will serve as a foundation for an ecosystem that will span all aspects of the design flow through to implementation.
Cadence, of course, has been down this path before. In 2001, it spearheaded the OpenAccess Coalition, an effort aimed at improving engineers’ productivity by building an infrastructure for EDA tool interoperability. The OpenAccess effort centered around Cadence’s donation of its Genesis design-database technology and an application programming interface (API) that would enable third-party tools to read from and write to the database. Under the aegis of the Silicon Integration Initiative (Si2), the OpenAccess Coalition would seek to build industry support for and adoption of OpenAccess in the interest of true tool interoperability.
Upon its announcement, though, the OpenAccess Coalition met with fairly stiff resistance from some quarters within the EDA industry. A number of competitors, most notably Synopsys, expressed concerns about the effort on several levels. The most unsettling aspect of the planned governance of the OpenAccess technology was the establishment of an OpenAccess Change Team that would oversee the development of the technology.
Anchoring the Change Team would be two “architects.” Part of the conditions of Cadence’s donation of the Genesis database technology and the associated API was that one of the two “architects” would always be a Cadence employee (Fig. 1). Thus, it was felt, Cadence would always have the ability to veto proposed changes to OpenAccess that conflicted with its best interests in some way.
This condition clearly rankled Cadence’s competitors again— most notably Synopsys, who, in its acquisition of Avanti, had gained its own world-class design database in MilkyWay. To Synopsys, OpenAccess was anything but “open.” Synopsys was about as likely to acquiesce to Cadence’s dominance of an industry-standard design database then as President Bush is to help Iran develop nuclear weapons today. So a war of words started over OpenAccess, ultimately culminating in Synopsys opening up MilkyWay and attempting to establish a competing standard for interoperability.
Cadence learned quite a bit from the OpenAccess Coalition effort. It’s not as if it was a failure by any means. In fact, the overriding irony of the whole story is that any number of smaller EDA vendors has done a far better job of implementing OpenAccess in their own tools than Cadence ever did. And to the extent that OpenAccess has seen adoption, it has proved to serve just the purpose it was intended to.
So today, in the Power Forward Initiative, we see Cadence once again seeking to take the lead in establishing an industry standard. It seems that in the EDA industry, any such efforts eventually will spark conflict between vendors. But there are some significant differences between OpenAccess as it was constituted at the outset and PFI in its embryonic state.
For one thing, there’s not much to go on technology-wise for any detractors to take shots at. All Cadence and its PFI partners have done so far is issue what it’s termed a “call to action” to the industry to address the leaky boat (no pun intended) that low-power design is at present.
For another, Cadence has to date foregone any effort to ensure that PFI’s technological underpinnings will be under its proverbial thumb. PFI is to be approached as a more traditional standards-building effort, in which the core members will develop a working standard proposal that will be subjected to a membership vote sometime during 2007 (Fig. 2). At that time, feedback from members will be considered and incorporated where necessary. The final version will be submitted to a standards body, such as Accellera and/or the IEEE, by 2008.
In theory, the entire industry has much to gain from the success of the Power Forward Initiative. Cadence so far has managed to draw support from myriad segments of the industry, including large semiconductor makers, a major IP provider, and a major foundry. It’s my hope that the industry will give the effort a chance, see where it’s headed, and then rally around it, not because it’s uniting behind Cadence’s agenda, but rather it’s own.