In April 2008, we organised the 7th International workshop on Performance Modelling, Evaluation and Optimization of Ubiquitous Computing and Networked Systems (PMEO-UCNS'2008) held in conjunction with the IEEE-IPDPS, 14–18 April 2008, Miami, Florida, USA. We selected seven high quality papers (out of 17 papers, which were presented at the workshop) and invited the authors of the selected papers to extend them and submit them for a complete new peer-review for consideration in this special issue. The final decision for the inclusion in the special issues has been strictly based on the outcome of the review process. International Journal of Parallel, Emergent and Distributed Systems (IJPEDS) is one of the well-established and reputable journals with a wide international readership. The main objective of the Special Issue is to make available the latest results in the field to the research community. The performance modelling, evaluation and optimisation of ubiquitous computing and networked systems have been an important research topic over the past years and poses challenging problems that require new tools and methods to keep up with the rapid evolution and increasing complexity of such systems. The purpose of this Special Issue is to report state-of-the-art and in-progress research on all aspects of performance modelling, evaluation, and optimisation of these systems. The selected papers span a broad range on the performance modelling and evaluation of such systems. The contributions of these papers are outlined below.
The performance of collective communication operations is known to have a significant impact on the scalability of some applications. Indeed, the global, synchronous nature of some collective operations directly implies that they will become the bottleneck when scaling to hundreds of thousands of nodes. This has led many researchers to try to improve the efficiency of collective operations. Although different measurement schemes for blocking collective operations are implemented in well-known benchmarks, many of these schemes introduce different systematic errors in their measurements. Hoefler, Schneider and Lumsdaine characterise these errors and select a window-based approach as the most accurate method. However, this approach complicates measurements significantly and introduces a clock synchronisation as a new source of systematic errors. They analyse approaches to avoid or correct those errors and develop a scalable synchronisation scheme to conduct benchmarks on massively parallel systems. Their results are compared to the window-based scheme implemented in the SKaMPI benchmarks and show a reduction of the synchronisation overhead by a factor of 16 on 128 processes. Within the collective communication aspects, Nasri investigates the performance of collective communications in heterogeneous and hierarchical environments. Precisely, an analytical model able to accurately predict the performance of communication operations in execution environments characterised by both heterogeneity and hierarchical structure has been proposed.
Routing is one of the most important issues in networked and distributed environments. Keqin Li evaluates the average-case performance of eight offline heuristic algorithms to solve the routing and wavelength assignment problem and the related throughput maximisation problem in wavelength division multiplexing optical networks. On the other hand, Migas investigates the QoS within the routing level in ad hoc networks. In particular, Migas tackles this by identifying the best metrics that can describe any route within a graph, in terms of overall throughput, reliability and minimum energy consumption. Simulation experiments have been conducted by varying critical factors of mobile devices such as battery reserves, memory and CPU utilisation, and results recorded the effect that this has on the device's overall routing metric.
The system architecture benchmarks provide insight into the architectural features of the proposed system. In this context, Fatoohi investigates the performance of six scientific application codes on four high-speed machines: three 512-processor SGI Altix machines (models 3700, 3700/BX2 and 4700) and a 128-processor Cray Opteron cluster. These applications are quite important since their performance will be a key factor in selecting the next-generation high-speed computing environment by NSF. These codes achieved different rates of performance on the SGI Altix machines ranging between less than 7% to over 67% of the peak performance of a single processor.
Network-on-Chip (NoC) is a promising communication paradigm for multiprocessor system-on-chips. This communication paradigm has been inspired from the packet-based communication networks and aims to overcome the performance and scalability problems of shared buses and large area overhead of point-to-point dedicated links in multi-core System on Chips. At the topology level, Sabbaghi-Nadooshan, Modarressi and Sarbazi-Azad propose a 2D shuffle-exchange based mesh topology, or 2D shuffle-exchange mesh for short for NoCs. The proposed two-dimensional topology applies the conventional well-known shuffle-exchange structure in each row and each column of the network. Compared to an equal sized mesh which is the most common topology in on-chip networks, the proposed shuffle-exchange based mesh network has smaller diameter but for an equal cost. Finally for better performance cross-shuffle is proposed. On the other hand, at the performance level, Talebi, Jafari and Khonsari address the problem of flow control for Best Effort traffic in NoC systems. Flow control was modelled as the solution to a maximisation problem whose objective was the sum of weighted logarithmic functions. They solved the problem indirectly through its dual using gradient projection method, which was led to a flow control algorithm that can be used to determine optimal BE source rates.
Finally, as guest –co-editors of this Special Issue, we would like to express our deepest thanks to the Editor-in-Chief, Professor Ivan Stojmenovic for hosting this Issue in the IJPEDS and for his continued support and helpful guidance throughout all the stages of preparing this Special Issue. Our sincere thanks also go to the Editorial-office staff of the journal for their excellent job during the course of preparing this special issue. We also thank the authors for their contributions, including those whose papers were not included. We thank and greatly appreciate the thoughtful work of many reviewers who provided invaluable evaluations and recommendations.