Wait, what? 128-bit architecture? I'm not sure what the utility of this would be, as 64-bit ints are huge and floating point is already a special case.<p>With 16-byte pointers, your application would be SUV-sized no matter what you do.<p>What kind of things need 128-bits to represent? Is this a case of representing 4-up 32-bit ints or something of the sort, that a 128-bit CPU would be basically a 4-way vector one inherently?