MARS: A Distributed Memory Approach to Shared Memory Compilation
LCR '98 Selected Papers from the 4th International Workshop on Languages, Compilers, and Run-Time Systems for Scalable Computers
| Hi-index | 0.00 |
This paper describes the integration of a formal loop generation technique into an auto-parallelizing compiler, MARS. A brief survey of loop generation techniques is given and is followed by the description of the loop generation strategy employed in our implementation. We describe the necessary input and output representations required for formal loop generation and describe how such a transformation fits into a complete compiler strategy. Given MARS' extended linear algebraic program representation and the constraints of a global compiler strategy, we have successfully integrated a formal tool into a FORTRAN compiler and have shown that combined they can outperform an existing commercial compiler.