Exploiting Outer Loops Vectorization in High Level Synthesis

Abstract

Synthesis of DoAll loops is a key aspect of High Level Synthesis since they allow to easily exploit the potential parallelism provided by programmable devices. This type of parallelism can be implemented in several ways: by duplicating the implementation of body loop, by exploiting loop pipelining or by applying vectorization. In this paper a methodology for the synthesis of complex DoAll loops based on outer vectorization is proposed. Vectorization is not limited to the innermost loops: complex constructs such as nested loops, conditional constructs and function calls are supported. Experimental results on parallel benchmarks show up to 7.35x speed-up and up to 40 % reduction of area-delay product.