For the purpose of obtaining empirical data on the possible benefits of our continuous compilation model, we constructed a simulator. This simulator accepts as input information about a particular program and a trace of its execution. It then simulates execution of the program using the continuous compilation model described in Chapter 3 and returns performance data. The simulator could also be used to do just-in-time compilation studies, although they were not done as part of this thesis.
Figure 4.1: Overview of the Simulator
As shown in Figure 4.1, the simulator is designed to accept information about either real or synthetic programs. Information about the program structure is contained in the setup files while the behavior file contains a trace of the program execution. Each of these is described in detail in the next section.
A description of the ProGenitor program can be found in Chapter 5. We discuss how the requisite information about real programs was obtained later in this chapter.