Hi:
The 300 MFLOPS doesn't help me, as I wish to do strange real-time stuff involving interrupt response within 30-100ns, and getting some quick integer/data structure calcs. done and some stuff sent out a GPIO or memory mapped IO port on the XINTF within another few 10s of ns. Basically, I need to get as much done as possible in less than 1us, with as little latency as possible, and with as little latency jitter as possible.
I'm starting to look at the 500-600MHz ADI Blackfins, and thinking they look sweet, except they lack some of the EVM capabilities that I am using, particularly the QEP inputs. I don't know however, what can be done with Blackfin in terms of interrupt latency and determinism. The more complex memory and pipeline architecture is an issue, so that would have to be experimented upon and characterized. Simply getting from 150 MHz --> 300 MHz or so on the C28x family would make me able to have a big smile, to avoid having to think about learning another platform.
Still waiting on more speeeeeeeedd....