>
> Re Ethernet interfaces: I suggest that the first version be built
> with 100mbit FDX Ethernet, and that the follow-on design upgrade
> to gigabit FDX Ethernet. This will be a pretty straightforward
> drop-in replacement at 10x the data rate, with the same software
> interface, same packet formats, etc.
I don't think computers of the next couple of years would be able to consume
and process more than about 100 mbits/sec of data, so 100mbit is fine, I
think. Besides, I'd like the ethernet part to be the easy part of the
design.
>
> You could do the design for gigabit today. PCI cards for it are down
> to hundreds of dollars. They can't do the full thruput because it's
> faster than the PCI bus can handle, but they're much faster than
> 100mbit. I suspect that the catch is that you'll be able to find lots
> of cheap and available IP cores and/or chips for 100mbit, but few for
> gigabit today.
Well, I can't find any free 100mbit cores except from opencores, and I'm not
sure of the state of that one. Cheap isn't good enough for free hardware...
> PS: I also think you should only put in the work to design hardware
> for SDR's if you think it's fun. Trying to sell niche computer
> hardware is a great way to do lots of work and lose money. Trying to
> design mass market computer hardware is a great way to do lots of work
> and lose even more money, unless you know how to market and sell mass
> market computer hardware.
I wasn't thinking of marketing it. I'd just like to design it, and if more
people like it, some organization could kit and sell it.