Seeds for a Heterogeneous Interconnect

Current trends in HPC (high performance computing) suggest that clusters will soon consist with hundreds, if not thousands, processors and the size of current scientific problems becomes much larger than before. Many researchers have predicted that the communication among these processors has dominated the execution time of the scientific parallel applications. Users will need well understanding on communication patterns among scientific parallel applications and their similarities so that users benefit not only from cost saving on constructing the running environment for these applications but also from obtaining better performance. In this paper, we address the communication pattern matching, and focus on point-to-point communication, which is primarily utilized (over 90% all MPI (message passing interface) calls) in most MPI codes and has much more impact on the communication performance than collective communication does. In this work, our contribution is that we propose a new approach MPACP (matching of parallel application communication patterns) to automate the analysis of the similarity between two parallel applications and provide a reliable report which will help users or developers understand the similarity among communication patterns of parallel applications. Furthermore, experimental results demonstrate the effective performance of our scheme in terms of the automatic matching of parallel application communication patterns.

PurposeThis paper explores the changes in communication patterns when companies implement lean, and how those changes relate to implementation success.Design/methodology/approachThis is a multiple-site case study involving four business units of a manufacturing company in South America, including two repeated measurement instances separating 24 months for approximately 600 direct workers and 65 supervisors. The analytical models include social network analysis measures and Ordinary Least Squares regression.FindingsWhen companies implement lean, (1) teams have a higher frequency of communication among members; (2) teams become more decentralized; (3) teams communicate more with supervisors and (4) supervisors communicate more amongst themselves and collaborate more. Also, (5) better performing teams change more pronouncedly.Research limitations/implicationsThe study contains data for four business units but within only one company, limiting the external validity of the conclusions. The sample was predominantly male. Participant attrition and other potential covariates not included in the study can be additional limitations.Practical implicationsLean implementations could be practically helped by managers by embracing and supporting the more intense communication patterns associated with lean success, and alternatively, they could proactively detect barriers in communication by measuring how these patterns change or fail to change and try to unlock communication by working on those barriers and supply communications infrastructure and opportunities for collaboration to try to boost the chances of success.Originality/valueThis is to our knowledge the first study measuring communication networks from the point of view of team members and low-level supervisors in lean implementations. This is also the first study showing that communication patterns change more rapidly in more successful teams, and also that communication pattern changes when implementing lean can be an indicator of success.

Chordal rings (or circulant graphs) are a popular class of fault-tolerant network topologies which include rings and complete graphs. For this class, the fundamental problem of leader election has been extensively studied assuming either a fault-free system or an upper-bound on the number of link failures. We consider chordal rings where an arbitrary number of lines has failed and a processor can only detect the status of its incident links. We shows that a leader election protocol in a faulty chordal ring requires only O(n log n) messages in the worst-case, where n is the number of processors. Moreover, we show that this is optimal. If the network is not partitioned, the algorithm will detect it and will elect a leader. In case the failures have partitioned the network a distinctive element will be determined in each active component and will detect that a partition has occurred; depending on the application, these distinctive elements can thus take the appropriate actions. >

In high performance computing (HPC), applications usually have many parallel tasks running on multiple machine nodes. As these tasks intensively communicate with each other, the communication overhead has a significant impact on an application's execution time. This overhead is determined by the application's communication pattern as well as the network distances between communicating tasks. By mapping the tasks to the available machine nodes in a communication-aware manner, the network distances and the execution times can be significantly reduced.

Workstation-based parallel systems are attractive due to their low cost and competitive uniprocessor performance. However, supporting a cache-coherent global address space on these systems involves significant overheads. We examine two approaches to coping with these overheads. First, DSM-specific hardware can be added to the off the-shelf component base to reduce overheads. Second, application-specific coherence protocols can avoid some overheads by exploiting programmer (or compiler) knowledge of an application's communication patterns. To explore the interaction between these approaches, we simulated four designs that add DSM acceleration hardware to a collection of off-the-shelf workstation nodes. Three of the designs support user-level software coherence protocols, enabling application-specific protocol optimizations. To verify the feasibility of our hardware approach, we constructed a prototype of the simplest design. Measured speedups from the prototype match simulation results closely. We find that, even with aggressive DSM hardware support, custom protocols can provide significant speedups for some applications. In addition, the custom protocols are generally effective at reducing the impact of other overheads, including those due to less aggressive hardware support and larger network latencies. However, for three of our benchmarks, the additional hardware acceleration provided by our most aggressive design avoid the need to develop more efficient custom protocols.

PurposeThe paper examines the impact of COVID-19 on the network structure of the Ghana Stock Exchange (GSE) market.Design/methodology/approachThe study utilises the minimum spanning tree (MST) and network measures, including normalised tree length, average shortest path length, node degree, and node strength, to analyse the GSE structure across three periods: pre-COVID-19 (January–December 2019), COVID-19 (January–December 2020) and post-COVID-19 (January–December 2021).FindingsThe study reveals a contraction of the MST, reduced normalised tree length and average shortest path length during COVID-19 compared to pre-COVID-19. The GSE market network structure underwent significant changes during COVID-19, with the MST expanding and normalised tree length increasing post-COVID-19. Node degree and strength analyses identified ICT stock as the central node pre-COVID-19, while financial stock assumed this role during and after COVID-19. These findings underscore the profound impact of the COVID-19 pandemic on the structure and dynamics of the GSE market network.Research limitations/implicationsThe study used correlation to measure the linear relationships between stocks without accounting for nonlinear associations. It focused exclusively on the GSE market. Despite this limitation, the empirical findings and robustness tests reveal the significant impact of COVID-19 on the GSE market’s network structure. These results highlight the dynamic nature of the GSE market’s network topology.Practical implicationsThis paper highlights the dynamic nature of the GSE market network structure, providing valuable insights for investors and market participants to navigate similar crises in the future. Additionally, GSE regulators can leverage these findings to identify stocks that may pose risks during extreme market events.Originality/valueThis study is the first in Ghana to examine the effects of extreme events, such as COVID-19, on the network structure of the GSE market.

This study investigates the impact of different field surfaces on passing networks and tactical performance of youth football teams. Using observational analysis, tactical positioning data were collected, and passing networks were constructed. The results suggest differences in network metrics based on the surface type: average shortest path length F(2.052)=6.099; p<0.006, n2=0.289; betweenness centrality F(2.001)=7.294; P<0.003, n2=0.327; closeness centrality F(2.025)=5.207; p<0.011, n2=0.258; clustering coefficient F(2.032)=23.679; p<0.001, n2=0.612; and radially F(2.001)=6.099; p<0.006, n2=0.289. Closeness centrality varied significantly between tactical positions F(10.009)=1.918, p<0.05, n2=0.466. Passing relationships based on field zones also showed significant differences: average shortest path length F(23.193)=6.057; P<0.001, n2=0.744; betweenness centrality F(23.002)=5.103; p<0.001, n2=0.710; closeness centrality F(23.015)=6.835; p<0.001, n2=0.766; degree F(23.592)=5.298; p<0.001, n2=0.717; radiality F(23.001)=8.366; p<0.001, n2=0.800; and stress F(23.773)=5.302; p<0.001, n2=0.718. This study provides valuable insights for coaches and analysts on optimizing youth soccer performance, highlighting the importance of considering field surface in tactical planning and training strategies.

Gene Networks with Transcriptional Bursting Recapitulate Rare Transient Coordinated High Expression States in Cancer.

All-to-all personalized communication (AAPC) is one of the most commonly used communication patterns in parallel applications. Developing an efficient AAPC routine is difficult since many system parameters can affect the performance of an AAPC algorithm. In this paper, we investigate an empirical approach for automatically generating efficient AAPC routines for Ethernet switched clusters. This approach applies when the application execution environment is decided, and it allows efficient customized AAPC routines to be created. Experimental results show that the empirical approach generates routines that consistently achieve high performance on clusters with different network topologies. In many cases, the automatically generated routines out-perform conventional AAPC implementations to a large degree.

A trend that has materialized, and has given rise to much attention, is of the increasingly heterogeneous computing platforms. Recently, it has become very common for a desktop or a notebook computer to be equipped with both a multi-core CPU and a GPU. Application development for exploiting the aggregate computing power of such an environment is a major challenge today. Particularly, we need dynamic work distribution schemes that are adaptable to different computation and communication patterns in applications, and to various heterogeneous configurations. This paper describes a general dynamic scheduling framework for mapping applications with different communication patterns to heterogeneous architectures. We first make key observations about the architectural tradeoffs among heterogeneous resources and the communication pattern of an application, and then infer constraints for the dynamic scheduler. We then present a novel cost model for choosing the optimal chunk size in a heterogeneous configuration. Finally, based on general framework and cost model we provide optimized work distribution schemes to further improve the performance.

People-to-people interactions in the real world and in virtual environments (e.g., Facebook) can be represented through complex networks. Changes of the structural properties of these complex networks are caused through a variety of dynamic processes. While accepting the fact that variability in individual patterns of behavior (i.e., establishment of random or FOAF-type potential links) in social environments might lead to an increase or decrease in the structural properties of a complex network, in this paper, we focus on another factor that may contribute to such changes, namely the size of personal networks. Any personal network comes with the cost of maintaining individual connections. Despite the fact that technology has shrunk our world, there is also a limit to how many close friends one can keep and count on. It is a relatively small number. In this paper, we develop a multi-agent based model to capture, compare, and explain the structural changes within a growing social network (e.g., expanding the social relations beyond one's social circles). We aim to show that, in addition to various dynamic processes of human interactions, limitations on the size of personal networks can also lead to changes in the structural properties of networks (i.e., the average shortest-path length). Our simulation result shows that the famous small world theory of interconnectivity holds true or even can be shrunk, if people manage to utilize all their existing connections to reach other parties. In addition to this, it can clearly be observed that the network's average path length has a significantly smaller value, if the size of personal networks is set to larger values in our network growth model. Therefore, limitations on the size of personal networks in network growth models lead to an increase in the network's average path length.


 
 
 
 This study examines the communication patterns made by agricultural extension agents for the Special Region of Yogyakarta (DIY) with millennial farmers after the Covid-19 pandemic. The Covid-19 outbreak has changed the way we communicate, especially the use of communication media. A very real change can be seen by keeping a distance when carrying out communications and activities. Everyone is forced by circumstances to be able to use communication media, namely gadgets as a communication tool. The words of millennial farmers started to be heard at the start of the pandemic, where these members of the millennial farmers were different from previous farmers, namely farmers who were media literate and able to adapt quickly to circumstances. Many social media applications are used to communicate during the pandemic, and these millennial farmers choose applications that easily facilitate their communication. Collecting data in this study using interviews, observation, documentation, and focus group discussion (FGD). Data analysis techniques used with qualitative descriptions. Research informants consisted of extension workers and millennial farmers from four districts and one city, namely Kulonprogo, Bantul, Gunung Kidul, Sleman, and Yogyakarta City. The results of the study indicate that there has been a change in communication patterns carried out by extension agents with millennial farmers after the pandemic, due to the use of technology. After the Covid-19 pandemic, information was digitized for farmer groups, they used information technology to communicate. The pattern of farmer communication during the post-pandemic period has also changed, where millennial farmers have taken a role in coordinating information
 
 
 

Social media networks are increasingly important tools for crisis communication and the Novel Coronavirus 2019 (COVID-19) provides a unique opportunity to explore the role of social media in crisis communication. The goal of this research is to better understand how from political leaders adapt their communication strategy to on-the-ground changes during a crisis and how the communication of crisis-relevant information on social media affects the public’s perception of the response effort. We achieve this through an analysis of pandemic data from The Covid Tracking Project at The Atlantic, survey data from the Covid States Project, and pandemic-relevant Twitter statuses posted by US state governors. Our results show that governors alter their communication patterns and tone after developments in the pandemic and that such changes in communication patterns are accompanied by subsequent changes in public perception of the governor’s response to the pandemic. Implications for theory and practice are discussed.

Large graphs with millions and even billions of vertices are found in many real-life network analysis, the processing of which is challenging. One of the toughest tasks is computing the average shortest path length in a large network, which requires a lot of memory and processing time while calculating many independent paths. Hence, this task becomes a good candidate for parallelizing. The idea of using graphics processing units (GPUs) for general purpose computing is not new, and with recent increases in performances and memory capacity, they make a perfect candidate for working with graphs. We will explore how the Dijkstra's algorithm can be used to calculate the average shortest-path length, and how it can be used in a massively parallelized system and compare the performance gains and drawbacks.

We propose and study an evolving network model for instant messaging system (IMS) with local-world preferential connections and acquaintance introduction mechanisms, which exist in many physical complex networks especially in IMS evolving network. We put forward the algorithms to generate IMS evolving network model and through simulation results we talk about the topological properties of the IMS evolving network. First of all, degree distribution, clustering coefficient along with average path length are simulated and then the community structure of the model is analyzed. We can draw the conclusion that the degree distribution of IMS evolving network doesnpsilat obey power-law distribution, and this network has a short average path length and a high clustering coefficient. In a word, the simulation result shows that the IMS evolving network is a complex network possessing properties of small-world and clear community structure, but doesnpsilat have scale-free property.