Issues and challenges in compiling for the CBEA

Abstract

It has been demonstrated that expert programmers can develop and hand tune applications to exploit the full performance potential of the CBE architecture. We believe that sophisticated compiler optimization technology can bridge the gap between usability and performance in this arena. To this end we have developed a research prototype compiler targeting the Cell processor. In this talk we describe a variety of compiler techniques to exploit the Cell processor. These techniques are aimed at automatically generating high quality codes for the heterogeneous parallelism available on the Cell processor. In particular we will focus the discussion on managing the small local memories of the SPEs and discuss our approach to presenting the user with a single shared memory image through our compiler controlled software cache. We will also report and discuss the results we have achieved to date, which indicate that significant speedup can be achieved on this processor with a high level of support from the compiler.