Parallel Computing on Heterogeneous Networks

Abstract

In the paper, we analyse challenges associated with parallel programming for common networks of computers (NoCs) that are, unlike dedicated parallel computer systems, inherently heterogeneous and unreliable. This analysis results in description of main features of an ideal parallel program for NoCs. We also outline some recent parallel programming tools, which try and respond to some of the challenges.

Improving the Performance of Hyperspectral Image and Signal Processing Algorithms Using Parallel, Distributed and Specialized Hardware-Based Systems

“…Before introducing our implementation of the PPI algorithm for heterogeneous systems, we must first formulate a general optimization problem in the context of fully heterogeneous systems (composed of different-speed processors that communicate through links at different capacities) [20]. Such a computing platform can be modeled as a complete graph where each node models a computing resource p i weighted by its relative cycle-time w i .…”

“…Although most parallel techniques and systems for image information processing employed by NASA and other institutions during the last decade have chiefly been homogeneous in nature (i.e., they are made up of identical processing units, thus simplifying the design of parallel solutions adapted to those systems), a recent trend in the design of HPC systems for data-intensive problems is to utilize highly heterogeneous computing resources [20]. This heterogeneity is seldom planned, arising mainly as a result of technology evolution over time and computer market sales and trends.…”

Design and Implementation of a Parallel Heterogeneous Algorithm for Hyperspectral Image Analysis Using HeteroMPI

“…Although most dedicated parallel machines for remote sensing data analysis employed by NASA and other institutions during the last decade have been chiefly homogeneous in nature [3], computing on heterogeneous networks of computers (HNOCs) has soon become a viable alternative to expensive parallel computing systems [4]. These networks enable the use of existing resources and provide incremental scalability of hardware components with performance isolation.…”

Event Logging: Portable and Efficient Checkpointing in Heterogeneous Environments with Non-FIFO Communication Platforms

“…In the near future, an important improvement is to implement Selective Checkpoint [15] based on Event Logging. Furthermore, we plan to optimize the checkpoint by using compiler technologies to minimize the data that need to be saved [18,19].…”

“…Furthermore, we plan to optimize the checkpoint by using compiler technologies to minimize the data that need to be saved [18,19]. Integration of our protocol with mpC language [15] is also in our schedule. …”