Tue Mar 9 11:22:10 CET 2010

Dependency analysis

Modeling a CPU.  Let's stick to the 2-phase model:

  - static: dependencies (connectivity: input influence & output spill)
  - dynamic: boolean functions

The idea is to express composition:

  - static: this is type inference: a sequence of instructions has
    certain static information associated that can be composed at
    compile time as concrete info (i.e. register usage).

  - dynamic info can be functionally modeled.

It might be better to do this in a typed language then?