A Functional Implementation of the Jacobi Eigen - Solver A

In this paper, we describe the systematic development o f t w o implementations of the Ja-cobi eigen-solver and give their performance results for the MITTMotorola Monsoon dataaow machine. Our study is carried out using MINT, the MIT Monsoon simulator. The design of these implementations follows from the mathematics of the Jacobi method, and not from a translation of an existing sequential code. The functional semantics with respect to array updates, which cause excessive array copying, has lead us to a new implementation o f a parallel group-rotations" algorithm rst described by Sameh. Our version of this algorithm requires On 3 operations, whereas Sameh's original version requires On 4 operations. The implementations are programmed in the language Id, and although Id has non-functional features , we h a v e restricted the development of our eigen-solvers to the functional subset of the language.