Transparent Parallelization of Binary Code

This paper describes a system that applies automatic parallelization techniques to binary code. The system works by raising raw executable code to an intermediate representation that exhibits all memory accesses and relevant register definitions, but outlines detailed computations that are not relevant for parallelization. It then uses an off-the-shelf polyhedral parallelizer, first applying appropriate enabling transformations if necessary. The last phase lowers the internal representation into a new executable fragment, re-injecting low-level instructions into the transformed code. The system is shown to leverage the power of polyhedral parallelization techniques in the absence of source code, with performance approaching those of source-to-source tools.