Split Compilation: an Application to Just-in-Time Vectorization

In a world of ubiquitous, heterogeneous parallelism, achieving portable performance is a challenge. It requires finely tuned coor- dination, from the programming language to the hardware, through the compiler and multiple layers of the run-time system. This doc- ument presents our work in split compilation and parallelization. Split compilation relies on automatically generated semantical an- notations to enrich the intermediate format, decoupling costly of- fline analyzes from lighter, online or just-in-time program transfor- mations. Our work focuses on automatic vectorization, a key optimiza- tion playing an increasing role in modern, power-efficient architec- tures. Our research platform uses GCC's support for the Common Language Infrastructure (CLI ECMA-335 (8) and ISO 23271:2006 (10)); this choice is motivated by the unique combination of opti- mizations and portability of GCC, through a semantically rich and performance-friendly intermediate format. Implementation is still in progress.