A high performance parallel vector processor having a modified vector register/element processor configuration

摘要

A parallel vector processor is disclosed. The vector processor comprises a plurality of vector registers, each vector register being subdivided in a plurality of smaller registers. A vector is stored in each vector register, the vector comprising a plurality of elements. The elements of the vectors are assigned for storage in the smaller registers of the vector register. In the parallel vector processor of the present invention, assume that each vector register is sub- divided into M smaller registers. The first successive M elements of an N element vector are assigned for storage in the M smaller registers of the vector register. An element processor is connected to each smaller register. Therefore, the first successive M elements of the N element vector are processed by the element processors 1 through M. The second successive M elements of the N element vector are assigned for storage in the same M smaller registers. The third successive M elements of the N element vector are assigned for storage in the M smaller registers. The second and third successive M elements of the N element vector are each processed by the element processors 1 through M. As a result, if the elements of a vector must be processed sequentially, when a second element, stored in a smaller register, is ready for processing by an element processor, the processing of the second element need not await the completion of the processing of a first element stored in the same vector register.