Integer and floating point matrix-vector multiplication on the reconfigurable mesh
Jerry L. Trahan, Chun-Ming Lu, Ramachandran Vaidyanathan · 2002
This paper presents the following results for matrix-vector multiplication on the reconfigurable mesh (R-Mesh). Multiplication of an N/spl times/N matrix and an N/spl times/1 vector (each element of which is an integer of w bits) can be performed in O(logw) time on a two-dimensional O(Nw/(logw)log/sup 2/N)/spl times/O(Nw/(logw)log/sup 2/N) R-Mesh or a three-dimensional N/spl times/N/spl times/(w/(logw)log/sup 2/N) R-Mesh; in both cases inputs and outputs are stored as w-bit integers. A natural extension of this problem is for floating point inputs. Floating point numbers have not been handled before on any reconfigurable bus-based model. Matrix-vector multiplication with floating point inputs can be performed in O(max{loglogN,logw}) time on an N/spl times/N/spl times/max {/spl radic/[N/(loglogN)],w/(logw)log/sup 2/N} three-dimensional R-Mesh, with the inputs and outputs stored as w-bit floating point numbers.