Tim Prince wrote:
Do you want the GPU of the future to look so much like the Convex or
Cray vector machines of the past? If you persuade enough people, it
might happen. It doesn't necessarily pay off, if it does nothing but
embed [un]packing in microcode.
Hehe...
Consider the outlook of technology now, x8086 demo for example. There is
a good chance the vector ops will be more of a superset of instructions
rather than a subset, as they are with Opteron, Xeon, Pentium, etc. I'd
imagine the vector ops can easily have a dynamic stride by then with any
available processor on a parallel core to execute the next element until
all elements done. =)
Thanks for your insight
--
