Performance analysis of the make and load building algorithms

Abstract

The performance of the make and load building algorithms is analyzed. The average time of compilation of an application is calculated by using the make algorithm. This time depends on the number of changed files, their compilation time, and the number of files affected by the change that have to be compiled. The load building algorithm chooses the files in an application that have to be compiled and places them on the compile list. The files that are not affected by changes are not compiled. This reduces the turnaround time for compilation and load building. The average time of compilation of an application is calculated by using the load building algorithm. This time depends on the number of files in an application, the number of changed files, and the interdependencies between files. It is proven analytically how the system performance can be improved by using the load building algorithm. A relative performance analysis of the make and load building algorithms shows that the load building algorithm allows for better performance if some files in the application do not have to be compiled, and the average time of analysis by the load building algorithm is smaller than the average time of compilation of files that are unnecessarily recompiled by using the make algorithm.<>