Abstract
This paper presents a comparison of two models suitable for use in a compiler for high performance computers. The first model, termed `the general model' is based on well-known data dependency techniques and a summary based on Banerjee's presentation is given. The second model, termed the `Unified Model' is a recently proposed model which approaches the compilation issue from a different viewpoint. A summary of the unified model is given. Issues involved in compiling for high performance computers are discussed. These issues form the criteria for the comparison of the models.