Ok, this is for the advanced class: Dynamic Binary Translation
Ok, assume that we have reached the famed silcon barrier for CPU (no more Moore's Law), and we have to (at runtime) unwind x86 instructions onto multiple CPUs running in a parallel (such as IBM Daisy).
Assuming no multitasking, the optimal number of multiple CPUs running would be relatively small.
However, modern desktop x86 computers run fifty tasks at once, by timeslicing. So what would the optimal number of CPUs running in parallel?
Assuming no multitasking, the optimal number of multiple CPUs running would be relatively small.
However, modern desktop x86 computers run fifty tasks at once, by timeslicing. So what would the optimal number of CPUs running in parallel?
