The FPGA community has long perceived the value of their programmable devices in the embedded market. This week, Intel validated that perception for the larger System-on-Chip (SoC) community at the Intel Developer’s Forum in San Francisco. At the Tuesday (Sep. 14, 2010) morning IDF keynote address, Doug Davis, VP and GM for Intel’s Embedded and Communications Group unveiled a device containing two side-by-side dyes in on package. One die was Intel’s recently announced E600 SoC version of the embedded Atom processor. The other die was an unnamed FPGA core from Altera. The Stellarton platform won’t be available until the early part of 2011.
In his very brief presentation, Davis said that a configurable embedded SoC was needed to address the growing demand for more I/O skews or variations and an ever shortening time to add new features.
Mark Plungy, Senior Manager of Public Relations at Altera, expressed the company’s pleasure in partnering with Intel. In addition, he believes that this partnership, “really confirms the value of FPGAs in embedded applications.” Plungy noted that Stellarton is an Intel device that would be sold and supported by Intel.
The specific FPGA device was not revealed by either Intel or Altera. Still, the low-power and low-cost mandate for today’s embedded products suggest that a likely candidate might be Altera’s Cyclone family of inexpensive, low-power FGPAs.
What does this partnership mean to Xilinx, the other major FPGA vendor? That’s difficult to tell. Xilinx announced a partnership with ARM earlier this year. ARM is the current de facto leader in embedded mobile systems which is a market position coveted by Intel.
Perhaps a more interesting question is not which FPGA architecture will win in the evolving programmable embedded space, but what new programming skill sets and tool suites will be needed to develop successful products on these new configurable, embedded platforms.
Shay Benchorin, Director of Marketing and Strategic Alliances for Mentor Graphics Embedded Systems Division, takes a larger view of the issue by noting the growing heterogeneousness of embedded devices. He explained that today, it’s common for FPGAs to contain processor cores. Now, it will be equally likely for an embedded processor to find an FPGA nearby in the same package and that FPGA contains multiple cores. The hardware heterogeneity is paralleled by the software heterogeneity in which multiple operating systems might be running on different cores within the same embedded package.
The need for configurable embedded devices is as much about merging processor and FPGA functionality as it is about satisfying the growing software demand for multiple operating systems and ever-changing feature sets.