To: elmatador who wrote (5318 ) 9/23/1999 12:05:00 AM From: ftth Respond to of 12823
re:<<Then when the 3G is ready, you upgrade your soft-HW to do 3G stuff. You can do that with FPGA's.>> In theory yes, but the reconfigurable logic portion is a fairly small number of logic cells/gates, so it's not infinitely reconfigurable. A good portion of the processing is still handled by a dedicated signal processor block and a microcontroller. These will in general be faster for many operations since the whole mess is (eventually) fabricated in the same process on a single die. It's awfully hard to size the reconfigurable portion so it has room for growth, while still meeting cost, power, and speed constraints. I'm somewhat skeptical if the FPGA portion offers much benefit at that point, but it could I suppose for offloaded, repetitive tasks that don't require any math operations. FPGA's are not generally cost-effective in razor-thin margin consumer products, despite the statements to the contrary in the original article. You generally will roll the FPGA into an ASIC for the final product, in order to reduce the bill of materials. That article also implied that a universal system that could plug into cable one minute and satellite the next is solved by FPGA's. That's not really true since you need an enitrely different front-end due to the different frequency ranges. That means everyone that buys such a product has to pay for both front ends even if they'll only ever use one. Another bit of misinformation in that article was that FPGA's have only been around for a few years. Actually they've been around for over a decade and were extremely painful to use back then. They've come a long way with the design tools, libraries, and place-and-route, to the point where you can literally slap together a complex design in a couple days that would have taken a month or better in the early days.