The first panel, composed of representatives from AMD, ARC International, IBM Microelectronics and Tensilica, examined the question of where to seek growth in the processor business. This question quickly devolved into two threads, with PowerPC vendor IBM and Intel-bating AMD stalking renewed growth in the server arena and the two embedded core vendors focusing on a very different scenario.

Both David Fritz, vice president of technology marketing at ARC, and Tensilica president and CEO Chris Rowen, cited a wide range of embedded applications as possible "next big things" for their businesses. Fritz cited storage devices and wireless networks as specific areas, while Rowen pointed to the general area of networking/media convergence. But both players offered a similar viewpoint: that increasing the level of abstraction at which their customer could engage with his problem was an important step.

For ARC's Fritz, this took the form of complete platforms of hardware, operating systems software and tools, configured for a particular application space. Rowan of Tensilica advocated flooding the die with processors — 100 to 1,000 on a die — so that each aspect of the design from conceptualization to detailed implementation could be treated as a software problem and dealt with via established programming languages.

Both approaches, though radically different, were expressed as ways to free the chip developer to focus on meeting the needs of the end customer rather than on the details of chip design.

A second panel later in the day took a second question: where is the growth in wireless? This one was addressed by speakers from Analog Devices, Inc., California Micro Devices, Intel, Motorola, and RF Micro Devices.

Once again, the question came back to the end customer. "If you want to know where the growth will be, don't look at semiconductor companies or even systems companies. Look at the market-makers in the consumer space," said Motorola strategic marketing manager Kyle Harper.

Kyle Baker, vice president of marketing at California Micro Devices, added that the typical supply chain reaching from a component supplier to an end user in the cell phone market is about five layers deep. "It is incredibly difficult to see through to the end-user needs," he said. "It is very easy to just focus on the person who gives you the purchase orders."

It was suggested that Intel's decision to only allow the Centrino brand on PCs that purchased everything on a list of Intel components amounted to trying to control end-user preferences from the component end, a claim which Intel director of marketing Mac Agan rejected.

Finally, an afternoon panel debated the question of who will control the middle ground between obvious FPGA applications and obvious cell-based ASIC applications. Moderator Bryan Lewis, Asic analyst at Dataquest Inc., said FPGAs are attempting to move into higher-performance, lower cost areas, while ASIC vendors are formulating ways to reduce turn around time and initial cost without sacrificing performance and density.

Meanwhile various vendors are creating platform chips that can be quickly configured to create a number of fixed macroblocks for a particular application need.