Systems-on-silicon approach

I read Michael Markowitz's "Where should your systems grow?" (EDN, Sept 1, 1998, pg 29) with interest. My product line is an FPGA cost-reduction conversion service: We convert Xilinx FPGAs to an ASIC equivalent for customers that need flexibility on the front end of their design but ASIC costs for production volumes.

We get to see the direct fallout of some of the points Markowitz made. All designers start out overly optimistic. They assume that their design will be easy to develop, on schedule, and full of the latest enabling technology. What happens in reality is that, although EDA companies claim their tools can make design simple and silicon vendors claim they can make design cheap, the design turns out to be neither of the above.

Frustrated designers turn to programmable logic to help them design faster and to try out portions of larger devices in silicon. They avoid the re-spin trap and find ways to make 50,000 or 100,000 gates a reasonable sweet spot for development. If the program does ramp to volume production, they can then make the commitment to turn the FPGA into an ASIC. It's a win-win situation for the customer, because designs get to market sooner, and customers can save NRE costs and custom-inventory management issues for those products that are known to be ramping rather than for those products that may be obsolete as soon as they are prototyped. They don't design million-plus-gate devices, but, as Markowitz pointed out, few million-plus-gate devices actually get to volume production.

Our best customers are the ones who go the full custom-ASIC route and determine that all of the time and money they spend makes them less competitive.

—Shelly Davis, Xilinx Inc

Corrections

In "Gloves Off" (Oct 8, 1998, pg 40), the URL we printed for the Web version of the article was incorrect. Please point your browser to www.ednmag.com/reg/1998/100898/21go.htm instead.

Also, in Bill Whitlock's "Subtleties count in side-dynamic-range analog interfaces'' (EDN, June 4, 1998) we inadvertently published Figure A (pg 139) twice. In Figure B, both the shield ground and the attached V_CM connection should be made at the driver side. The correct figure appears here.

In Bill Schweber's "Video switches route analog signals along paths of least resistance" (EDN, June 4, 1998, pg 58), note that Gennum Corp (www.gennum.com) provides analog video switches in addition to their digital-switch family. We apologize for the errors.

Sound off

Send your letters to Signals & Noise Editor, EDN, 275 Washington St, Newton, MA 02458 or e-mail us at mhadro@edn.cahners.com. Our fax is 1-617-558-4470. EDN reserves the right to edit letters for clarity and length.

EDN Access | Feedback | Table of Contents |