The finite element method is well reputed for its great flexibility in
solving problems with complex geometries and heterogeneous structures
! but, in its classical form, it has fairly low accuracy and poor comp
utational efficiency. This makes an adverse impact on a large-scale nu
merical simulation of acoustic wavefield propagation. It has been show
n by the author and his co-workers that the spectral approach, based o
n the use of high-order orthogonal interpolating functions (the Spectr
al Element Method), yields high accuracy with almost no numerical arti
facts; it also significantly reduces the simulation computing costs. I
n this paper, the method is presented in conjunction with an iterative
solution technique in such a way that it fully exploits the currently
available parallel computers. The underlying algorithm is based on th
e observations that the assembly of mass and stiffness matrices is not
really needed, and that the matrix-vector product required for the it
erations can be done concurrently via an element-by-element approach.
Moreover, by using a tenser-product sum-factorization scheme, computat
ional and storage requirements can be further reduced as neither eleme
nt nor global matrices are ever formed. The method is tested on the Gr
ay T3D MPP computer, and its parallel efficiency and speed performance
are discussed.