AVR-Chat

On Fri, Nov 20, 2009 at 12:58:37PM +0000, Luke Whitmore wrote:
> This is pretty interesting.  I suppose intellectual property law is
> bound to rear it's ugly head when FPGAs are considered.

Why do you sound so negative? Has long been established that a CPU
instruction set, registers, and memory map are all outside of patent law.
One can patent implementations, such as instruction pipelines and
instruction overlapping. Probably the biggest risk of IP violation would
be if one instructed customers to use Atmel documentation for non-Atmel
product.

> I wonder if there's an open-source movement producing designs in the
> commons for FPGA devices?  AFAIK, Atmel have always been very good to
> the open-source community - how can they adapt once FPGA use becomes
> more ubiquitous?

FPGA costs more. It takes a fairly big FPGA to host a CPU.

I suspect Atmel has AVR cores available for licensing, other companies
do the same for their product.

While FPGA's are relatively expensive, for some applications they are
dirt cheap. If you need a rad-hard space qualified CPU about your only
option is an IP core in an FPGA. Unlike a mask-made CPU, each and every
cell of an FPGA can be verified after manufacture. That each cell is a
clone of the others increases confidence. Much NRI can be spent
designing the cell for durability and radiation hardness. It can be
tested with identical cells surrounding it to ensure the others do not
affect it. In a hardwired CPU most every gate has a unique combination
of gates surrounding it.

-- 
David Kelly N4HHE, dkelly@HiWAAY.net
========================================================================
Whom computers would destroy, they must first drive mad.