> Wouldn't we be able to process 16 of them at the same time?
Yes you would if you had 4 execution ports available and no data dependencies. Of course, those execution ports could also be processing 256 bit wide SIMD registers instead of just 32 bits. So it's a bad idea.
Instruction count is also higher, which is never a good thing.
> Actually, isn't that kind of what GPUs already do?
Yes you would if you had 4 execution ports available and no data dependencies. Of course, those execution ports could also be processing 256 bit wide SIMD registers instead of just 32 bits. So it's a bad idea.
Instruction count is also higher, which is never a good thing.
> Actually, isn't that kind of what GPUs already do?
No.