De Sutter Et Al . : Whole - Program Optimization of Binary Executables 3

|Compiler optimization research has a long history and very complex algorithms were developed to optimize code. Unfortunately these algorithms, however complex they may be, su er from the modular design of software. Compilers compile and optimize one source code module at a time. While they do so, they have no or few knowledge about the context in which the code will be executed. Even if summary information about all involved source code les is collected or if programs consist of one source code le only (which is not trivial for mixed-language programs) all kinds of code libraries will be linked with the programmer's code to produce the nal executable binary. These libraries are often only available in object code and therefor no candidates for global analyses and optimizations by the compiler. A solution to this problem is to relegate part of the global optimizations to the linker, which has -by its naturea global overview of the application. Unfortunately, the linker does not have access to the wealth of (semantic) information about the program that is available in the compiler. Therefor, a linker has to painstakingly reconstruct part of that information by analyzing object les. Nevertheless, it turns out to be possible to do this in a reasonable time, and to produce speedups of 19% for code that was already highly optimized by the compiler. In the paper, we describe Alto, a link time optimizer for the alpha architecture. We describe Alto's framework, the analyses performed, the advantages and disadvantages of optimizing programs at link time, and an analysis of the speedups attained for the Spec benchmarks. We also describe additional program transformations applied by Squeeze, an Alto-derivative aimed at link-time code compression. Squeeze obtains additional link-time code compression of 23% on already compact compiler generated binaries. Keywords|linker, whole-program optimization, code compression,