11 shows the efficiency of
parallel algorithms proposed in the paper.
In addition, [8] examined the edge detection in both sequential and
parallel algorithms in order to measure the speedup.
For
parallel algorithms, we adaptively select the parallelism p according to the number of samples n and set p = ceil(n), where ceil(v) means rounding up x to an integer.
To achieve the maximum performance on multi-core processors,
parallel algorithms have become inevitable to make each core execute different instructions simultaneously.
However, fast and efficient parallel computations require application of the appropriate
parallel algorithms. The permanent development of the parallel computing technologies with all their diversity requires a constant attention, which must be paid to a development and selection of proper
parallel algorithms for the solution of various problems.
The orthogonalization of [AB.sub.j] can be done by using various
parallel algorithms [37] such as RODDEC [27] or CA-QR [12].
Fast and
parallel algorithms such as the fast multipole method (FMM) [15-18], hybrid FMM and fast Fourier transform (FFT) [19], and parallel adaptive integral method [20,21] have been developed to accelerate the dense matrix-vector product (MVP).