Array processor with multiple broadcasting
Journal of Parallel and Distributed Computing
Mesh Computer Algorithms for Computational Geometry
IEEE Transactions on Computers
Connection autonomy in SIMD computers: a VLSI implementation
Journal of Parallel and Distributed Computing
Image Computations on Meshes with Multiple Broadcast
IEEE Transactions on Pattern Analysis and Machine Intelligence
Solving linear recurrence systems on mesh-connected computers with multiple global buses
Journal of Parallel and Distributed Computing
Square Meshes are Not Always Optimal
IEEE Transactions on Computers
Journal of Parallel and Distributed Computing
A time-optimal multiple search algorithm on enhanced meshes, with applications
Journal of Parallel and Distributed Computing
Prefix Computations on a Generalized Mesh-Connected Computer with Multiple Buses
IEEE Transactions on Parallel and Distributed Systems
Sorting on a mesh-connected parallel computer
Communications of the ACM
Introduction to VLSI Systems
A Fast Selection Algorithm for Meshes with Multiple Broadcasting
IEEE Transactions on Parallel and Distributed Systems
Hi-index | 0.00 |
Consider a two-dimensional mesh-connected computer with segmented buses (2-MCCSB). A k1n1 × k1n2 2-MCCSB is constructed from a k1n1 × k1n2 mesh organization by enhancing the power of each disjoint n1 × n2 submesh with multiple buses (sub-2-MCCMB). Given a set of n elements, this paper presents a parallel algorithm for finding the median in O(n1/8 log n) time on an n1/2 × n1/2 square 2 MCCSB, where each disjoint sub-2-MCCMB is of dimension n3/8 × n3/8. This result is competitive with the previous result with time bound of O(n1/6(log n)2/3) for finding the median on an n1/2 × n1/2 square 2-MCCMB and our time bound is equal to the previous time bound of O(n1/8 log n) on an n5/8 × n3/8 rectangular 2-MCCMB. Furthermore, the time bound of our parallel algorithm can be reduced to O(n1/10 log n) time on an n3/5 × n2/5 rectangular 2-MCCSB, where each disjoint sub-2-MCCMB is of dimension n1/2 × n3/10. We also show that the time bound can be further reduced to O(n1/11 log n) if O(n10/11) processors are used. Our algorithm can be modified to solve the selection problem.