Heard at DAC: The question on everyone’s mind
Ron Wilson - June 25, 2010
One of the favorite pastimes at the Design Automation Conference each year-other than scoring the best hand-outs on the show floor-is trying to divine the subject of the buzz: that background level of conversation between attendees. Some years the subject is an unsolved design problem. Some years it is a new product, or a category of new products. Some years it is speculation on a business deal. This year-narrowly edging out speculation over just how many of Apple’s first million iPads Globalfoundries gave away on the show floor, and the far more sinister question of what Carl Icahn intends to do to Mentor Graphics-the question of the week was what on earth Cadence means by EDA360.
The question lingered despite a surplus of data. EDA360 was all over Cadence’s booth. Words like ecosystem were in every message Cadence people presented at the conference. And of course there is that twenty-something-page “manifesto” on Cadence’s Web site. But somehow none of these sources quite resolved the question.
Clearly EDA360 has something to do with collaboration, and with pulling together IP, tools, and services to create a complete design environment for increasingly under-capitalized design teams. It would be easy to see this emphasis as simply an attempt to put the go-it-alone attitude of Cadence’s previous administration behind it. But there may be more to it than that.
I suspect that the roots of EDA360 lie deeper: not in a new view of how design teams work, but in a vision of how the industry itself is changing. It is widely, if quietly and with some unease, noted that there is profound structural change sweeping the electronics supply chain. Systems OEMs, in the view of Cadence chief marketing officer John Bruggeman, increasingly draw their revenue not from hardware but from selling applications and content. Apple is an obvious example, but with only a little strain you can make the same argument for Cisco, or even a robotics maker. Silicon is becoming a necessary but invisible part of the platform, not the source of differentiation. This attitude, in turn, subjects the semiconductor branch of the supply chain to intense pricing pressure, starving every link in the chain of revenue. Every level in the chain must savagely reduce costs, and some may simply vanish into virtual-or real-reintegration. “There are levels in the chain that won’t make sense,” Bruggeman warns.
In this vision, systems development becomes not a pipeline within a company, but a multiway collaboration between teams with many diverse points of expertise, each bringing their own tools and skills. And the objective changes as well. The design is no longer guided by an SoC requirements document, but rather by a set of software APIs, constrained by power, form factor, and the performance needs of applications that are being developed concurrently with the hardware. “Hardware development becomes API-first, not architecture-first,” Bruggeman says. “The reference for verification is not the requirements document, but the user’s experience.”
This may be the frame of reference in which EDA360 was conceived: a disintegrating development chain occupied not by a monolithic fabless semi company, but by a dynamic ad-hoc network of vendors and contractors. A project guided not by a fixed requirements document, but by the continuously shifting results of system-level simulations of the user experience. And all the pieces held together by a web of interoperating tools and libraries. If that is the idea, then indeed we should be discussing EDA360, for it is a very challenging idea indeed.