DMA for a proposed new design would be useful, if only for the "configure transfer and get on with core duties, I'll check when it's finished" aspect that reduces load on the CPU/OS *greatly* even if it doesn't hugely increase the speed of the transfer. As many tasks are CPU-bound, they can continue relatively unharmed.
DMA systems, even if not significantly faster by design, don't exhibit many of the 'locking' behaviors of non-DMA systems.
One of the big snags of the option of improving the Q68 is if the original supporters who have already bought a unit will get upgrades, and if those upgrades will carry the same heavy mark-ups of the Q68 itself. It would be good to see a self-administerable option to upgrade a system that doesn't require sending the unit internationally to Derek, for many of us. A JTAG programming cable and download should do it. The only reason not to do something like this would be to keep the ability to program ICs within Peter and Derek's personal control.
This is why I voted for new 68060 hardware of discrete devices that can be understood and upgraded - and that represent an open system - which will always beat out a closed and secretive black box system, in my mind.
I have an embedded application with customers that would ideally suit the Q68 system, but it needs to be tightly integrated. Even thinking of trying to license it to use in this custom system gives me a headache. The level of support it would receive would be unworkably low.
This is why I decided anything I personally produce along those lines will be Open Source.
