The Euro-Par 2017 Workshops proceedings focus on many subjects such as Programming Languages as well as Programming and Operating Systems.

Large-scale applications are ever-increasingly geo-distributed. Maintaining the highest possible data locality is crucial to ensure high performance of such applications. Dynamic replication addresses this problem by dynamically creating replicas of frequently accessed data close to the clients. This data is often stored in decentralized storage systems such as Dynamo or Voldemort, which offer support for mutable data. However, existing approaches to dynamic replication for such mutable data remain centralized, thus incompatible with these systems. In this paper we introduce a write-enabled dynamic replication scheme that leverages the decentralized architecture of such storage systems. We propose an algorithm enabling clients to locate tentatively the closest data replica without prior request to any metadata node. Large-scale experiments on various workloads show a read latency decrease of up to 42% compared to other state-of-the-art, caching-based solutions.

Stream computing is becoming a more and more popular paradigm as it enables the real-time promise of data analytics. Apache Kafka is currently the most popular framework used to ingest the data streams into the processing platforms. However, how to tune Kafka and how much resources to allocate for it remains a challenge for most users, who now rely mainly on empirical approaches to determine the best parameter settings for their deployments. In this poster, we make a through evaluation of several configurations and performance metrics of Kafka in order to allow users avoid bottlenecks, reach its full potential and avoid bottlenecks and eventually leverage some good practice for efficient stream processing.

The increasing gap between computational power and I/O performance in new supercomputers has started to drive a shift from an offline approach to data analysis to an inline approach, termed in situ visualization (ISV). While most visualization software now provides ISV, they typically visualize large dumps of unstructured data, by rendering everything at the highest possible resolution. This often negatively impacts the performance of simulations that support ISV, in particular when ISV is performed interactively, as in situ visualization requires synchronization with the simulation. In this work, we advocate for a smarter method of performing ISV. Our approach is data-driven: it aims to detect potentially interesting regions in the generated dataset in order to feed ISV frameworks with “the interesting” subset of the data produced by the simulation. While this method mitigates the load on ISV frameworks by making them more efficient and more interactive, it also helps scientists focus on the relevant part of their data. We investigate smart ISV in the context of a climate simulation, with a set of generic filters derived from information theory, statistics and image processing, and show the tradeoff between performance and quality of visualization.

This special issue of Concurrency and Computation: Practice and Experience contains revised and extended versions of selected papers presented at the conference Euro-Par 2015. Euro-Par—the European Conference on Parallel Computing—is an annual series of international conferences dedicated to the promotion and advancement of all aspects of parallel and distributed computing. Euro-Par covers a wide spectrum of topics from algorithms and theory to software technology and hardware-related issues, with application areas ranging from scientific to mobile and cloud computing. The major part of the Euro-Par audience consists of researchers in academic institutions, government laboratories and industrial organisations. Euro-Par 2015, the 21st conference in the Euro-Par series, was held in Vienna, Austria. It was organised by the Research Group for Parallel Computing of the Vienna University of Technology (TU Wien). Thirteen broad topics were defined and advertised, covering a large variety of aspects of parallel and distributed computing. The call for papers attracted a total of 190 submissions. The submitted papers were reviewed at least three and, in most cases, four or even more times (four reviews on average). A total of 51 papers were finally accepted for publication. This makes a global acceptance rate of 27 %. The authors of accepted papers came from 21 countries, with the four main contributing countries—the United States, France, Spain and Germany—accounting for a bit more than half of them. Based on the results of the reviews and a majority opinion of the respective topic programme committees, a number of papers were recommended for this special issue. The authors were contacted at the conference and invited to submit revised and extended versions of their papers. These new versions were reviewed independently by three reviewers; two had previously reviewed the conference version, the third had not. Eventually, four papers were accepted for publication. This year, two Euro-Par topics are represented—both covering methods of programming modern computer architectures. Topic 13 on Accelerator Computing is represented with three papers. The paper Performance optimization of sparse matrix-vector multiplication for multi-component PDE-based applications using GPUs, authored by Ahmad Abdelfattah, Hatem Ltaief, David Keyes and Jack Dongarra [1], describes the implementation of a single-GPU and multi-GPU kernel for block-sparse matrix-vector multiplication, a problem that appears in the discretisation of partial differential equations with many dependent variables. The performance of the kernel is measured on a subset of the Florida Sparse Matrix Collection. Especially noted by the reviewers was the uniform interface that applies to a wide range of problem sizes via tunable parameters. This makes it perform efficiently on a wide range of GPU architectures running CUDA. The paper Fast parallel skew and prefix-doubling suffix array construction on the GPU, authored by Leyuan Wang, Sean Baxter and John D. Owens [2], proposes a hybrid GPU implementation of known algorithms for constructing suffix arrays of a string that fits the given GPU architecture best. One highlight pointed out in the reviews is a highly efficient segmented sorting primitive, which is also valuable as independent result. The paper Performance and portability of accelerated lattice Boltzmann applications with OpenACC, authored by Enrico Calore, Jiri Kraus, Sebastiano Fabio Schifano and Raffaele Tripiccione [3], reports on a performance study based on a simple performance model of an OpenACCbased lattice Boltzmann implementation on three different architectures: an NVIDIA GPU, an AMD GPU and a multi-core CPU. The practical relevance of this work was particularly appreciated.

While many parallel visualization tools now provide in situ visualization capabilities, the trend has been to feed such tools with large amounts of unprocessed output data and let them render everything at the highest possible resolution. This leads to an increased run time of simulations that still have to complete within a fixed-length job allocation. In this paper, we tackle the challenge of enabling in situ visualization under performance constraints. Our approach shuffles data across processes according to its content and filters out part of it in order to feed a visualization pipeline with only a reorganized subset of the data produced by the simulation. Our framework leverages fast, generic evaluation procedures to score blocks of data, using information theory, statistics, and linear algebra. It monitors its own performance and adapts dynamically to achieve appropriate visual fidelity within predefined performance constraints. Experiments on the Blue Waters supercomputer with the CM1 simulation show that our approach enables a 5x speedup with respect to the initial visualization pipeline and is able to meet performance constraints.

Hadoop emerged as an important system for large-scale data analysis. Speculative execution is a key feature in Hadoop that is extensively leveraged in clouds: it is used to mask slow tasks (i.e., stragglers) -- resulted from resource contention and heterogeneity in clouds -- by launching speculative task copies on other machines. However, speculative execution is not cost-free and may result in performance degradation and extra resource and energy consumption. While prior literature has been dedicated to improving stragglers detection to cope with the inevitable heterogeneity in clouds, little work is focusing on understanding the implications of speculative execution on the performance and energy consumption in Hadoop cluster. In this paper, we have designed a set of experiments to evaluate the impact of speculative execution on the performance and energy consumption of Hadoop in homo-and heterogeneous environments. Our studies reveal that speculative execution may sometimes reduce, sometimes increase the energy consumption of Hadoop clusters. This strongly depends on the reduction in the execution time of MapReduce applications and on the extra power consumption introduced by speculative execution. Moreover, we show that the extra power consumption varies in-between applications and is contributed to by three main factors: the duration of speculative tasks, the idle time, and the allocation of speculative tasks. To the best of our knowledge, our work provides the first deep look into the energy efficiency of speculative execution in Hadoop.

Apache Cassandra is an open-source cloud storage system that offers multiple types of operation-level consistency including eventual consistency with multiple levels of guarantees and strong consistency. It is being used by many data-center applications (e.g., Facebook and App Scale). Most existing research efforts have been dedicated to exploring trade-offs such as: consistency vs. Performance, consistency vs. Latency and consistency vs. Monetary cost. In contrast, a little work is focusing on the consistency vs. Energy trade-off. As power bills have become a substantial part of the monetary cost for operating a data-center, this paper aims to provide a clearer understanding of the interplay between consistency and energy consumption. Accordingly, a series of experiments have been conducted to explore the implication of different factors on the energy consumption in Cassandra. Our experiments have revealed a noticeable variation in the energy consumption depending on the consistency level. Furthermore, for a given consistency level, the energy consumption of Cassandra varies with the access pattern and the load exhibited by the application. This further analysis indicates that the uneven distribution of the load amongst different nodes also impacts the energy consumption in Cassandra. Finally, we experimentally compare the impact of four storage configuration and data partitioning policies on the energy consumption in Cassandra: interestingly, we achieve 23% energy saving when assigning 50% of the nodes to the hot pool for the applications with moderate ratio of reads and writes, while applying eventual (quorum) consistency. This study points to opportunities for future research on consistency-energy trade-offs and offers useful insight into designing energy-efficient techniques for cloud storage systems.

With increasingly inexpensive cloud storage and increasingly powerful cloud processing, the cloud has rapidly become the environment to store and analyze data. Most of the large-scale data computations in the cloud heavily rely on the MapReduce paradigm and its Hadoop implementation. Nevertheless, this exponential growth in popularity has significantly impacted power consumption in cloud infrastructures. In this paper, we focus on MapReduce and we investigate the impact of dynamically scaling the frequency of compute nodes on the performance and energy consumption of a Hadoop cluster. To this end, a series of experiments are conducted to explore the implications of Dynamic Voltage Frequency scaling (DVFS) settings on power consumption in Hadoop-clusters. By adapting existing DVFS governors (i.e., performance, powersave, ondemand, conservative and userspace) in the Hadoop cluster, we observe significant variation in performance and power consumption of the cluster with different applications when applying these governors: the different DVFS settings are only sub-optimal for different MapReduce applications. Furthermore, our results reveal that the current CPU governors do not exactly reflect their design goal and may even become ineffective to manage the power consumption in Hadoop clusters. This study aims at providing more clear understanding of the interplay between performance and power management in Hadoop cluster and therefore offers useful insight into designing power-aware techniques for Hadoop systems.

The easily-accessible computation power offered by cloud infrastructures coupled with the revolution of Big Data are expanding the scale and speed at which data analysis is performed. In their quest for finding the Value in the 3 Vs of Big Data, applications process larger data sets, within and across clouds. Enabling fast data transfers across geographically distributed sites becomes particularly important for applications which manage continuous streams of events in real time. Scientific applications (e.g. the Ocean Observatory Initiative or the ATLAS experiment) as well as commercial ones (e.g. Microsoft's Bing and Office 365 large-scale services) operate on tens of data-centers around the globe and follow similar patterns: they aggregate monitoring data, assess the QoS or run global data mining queries based on inter site event stream processing. In this paper, we propose a set of strategies for efficient transfers of events between cloud data-centers and we introduce JetStream: a prototype implementing these strategies as a high performance batch-based streaming middleware. JetStream is able to self-adapt to the streaming conditions by modeling and monitoring a set of context parameters. It further aggregates the available bandwidth by enabling multi-route streaming across cloud sites. The prototype was validated on tens of nodes from US and Europe data-centers of the Windows Azure cloud using synthetic benchmarks and with application code from the context of the Alice experiment at CERN. The results show an increase in transfer rate of 250 times over individual event streaming. Besides, introducing an adaptive transfer strategy brings an additional 25% gain. Finally, the transfer rate can further be tripled thanks to the use of multi-route streaming.

Today's continuously growing cloud infrastructures provide support for processing ever increasing amounts of scientific data. Cloud resources for computation and storage are spread among globally distributed datacenters. Thus, to leverage the full computation power of the clouds, global data processing across multiple sites has to be fully enabled. However, managing data across geographically distributed data enters is not trivial as it involves high and variable latencies among sites which come at a high monetary cost. In this work, we propose a uniform data management system for scientific applications running across geographically distributed sites. Our solution is environment-aware, as it monitors and models the global cloud infrastructure, and offers predictable data handling performance for transfer cost and time. In terms of efficiency, it provides the applications with the possibility to set a trade off between money and time and optimizes the transfer strategy accordingly. The system was validated on Microsoft's Azure cloud across the 6 EU and US data enters. The experiments were conducted on hundreds of nodes using both synthetic benchmarks and the real life A-Brain application. The results show that our system is able to model and predict well the cloud performance and to leverage this into efficient data dissemination. Our approach reduces the monetary costs and transfer time by up to 3 times.

The increasing gap between computational power and I/O performance in new supercomputers drives a shift from an offline approach of data analysis to an inline approach, termed in situ visualization (ISV). While many parallel visualization tools now provide ISV, the trend has been to feed such software with what previously was large dumps of raw data, and let them render everything at the highest possible resolution. This leads to a potentially large performance impact in simulations that support ISV, in particular when ISV is performed interactively. In this paper, we present a smarter method of performing ISV. Our approach aims to detect potentially interesting regions in the generated dataset in order to feed ISV frameworks with only a subset of the data produced by the simulation. While this method mitigates the load on ISV frameworks, making them more efficient and more interactive, it also helps scientists focus on the relevant part of their data. We investigate Smart ISV in the context of a climate simulation, with a set of generic filters derived from information theory, statistics and image processing, and discuss possible trade-offs between performance and quality of data.

The increasing scale at which data processing is being performed nowadays calls for data management systems that enable high-performance data exchanges among geographically remote instances of large web services. In this demonstration we show how JetStream can increase the transfer rate of events which are streamed between geographically remote cloud data centers. The demonstration setup focuses on presenting how the binding can be done between JetStream and the event source on one hand and with the StreamInsight processing engine on the other hand. By considering a data source with an event generation rate that is variable in time, we demonstrate the importance of adapting the transfer scheme to the streaming context.

The continuous growth of sensor networks, stock exchanges, climate monitoring or scientific applications produces new streaming data at increasing rates. Managing and processing such data, sometimes generated from multiple geographical locations, raises important challenges as it requires real-time processing or data aggregation. Conventional solutions like DBMS, MapReduce or dedicated solutions adopting single-located environments fail to meet the demands required for processing the Geo-distributed streaming data. Public clouds like Azure, with data centers spread around the globe, offer the infrastructure which can handle such a processing. Our approach, proposes a service-oriented cloud architecture for performing the stream analysis, by composing services which are distributed among multiple cloud data centers. Hence, the computation is moved towards the multiple data sources exploiting the geographical data locality. The initial results showed good scalability of the approach, reaching 1000 cores in the Azure cloud, and performance improvements compared to single location processing of a factor of 3.3.

evolution towards an in reasing integration of large-s ale distributed platforms ( louds, loud federations, enterprise desktop grids, et .). We propose an approa h whi h aims to over ome the urrent limitations of existing Map-Redu e frameworks, in order to a hieve s alable, on urren y-optimized, fault-tolerant Map-Redu e data pro essing on hybrid infrastru tures. This approa h will be evaluated with real-life bio-informati s appli ations on existing Nimbus-powered loud testbeds inter onne ted with desktop grids.

This special issue of Concurrency and Computation: Practice and Experience contains revised and extended versions of selected papers presented at the conference Euro-Par 2012. Euro-Par—the European Conference on Parallel Computing—is an annual series of international conferences dedicated to the promotion and advancement of all aspects of parallel and distributed computing. Euro-Par covers a wide spectrum of topics from algorithms and theory to software technology and hardware-related issues, with application areas ranging from scientific to mobile and cloud computing. The major part of the Euro-Par audience consists of researchers in academic institutions, government laboratories, and industrial organizations. Euro-Par 2012, the 18th conference in the Euro-Par series, was organized by the Computer Technology Institute & Press ‘Diophantus’ (CTI) in Patras, and held on Rhodes Island, Greece. Sixteen broad topics were defined and advertised, covering a large variety of aspects of parallel and distributed computing. The call for papers attracted a total of 228 submissions. The submitted papers were reviewed at least three and, in many cases, four times (3.83 on average). A total of 75 papers were finally accepted for publication. This makes a global acceptance rate of 32.9%. The authors of accepted papers come from 29 countries, with the four main contributing countries—the USA, Germany, Spain and France—accounting for about 52% of them. The distribution of papers follows the pattern typical for a Euro-Par conference: 65% are authored by academic researchers, 24% by students, and 11% by other authors (industry, NGO and government). Based on the results of the reviews and a majority opinion of the respective topic program committees, several papers were recommended for a special journal issue. The authors were contacted at the conference and invited to submit revised and extended versions of their papers. These new versions were reviewed independently by three reviewers; two had also reviewed the conference version, the third had not. Eventually, five papers were accepted for publication. Topic 2 on Performance Prediction and Evaluation is represented by the paper Adaptive Sampling for Performance Characterization of Application Kernels authored by Pablo de Oliveira Castro, Eric Petit, Asma Farjallah, and William Jalby 1. Its subject is the open-source Adaptive Sampling Kit (ASK) that facilitates, with relatively few samples, a characterization of the performance trade-off in large design spaces. To this end, the kit is equipped with a number of adaptive sampling strategies. The authors add a new strategy, hierarchical variance sampling, and present the performance characterization of three problems: memory stride accesses, Jacobian (2D) stencil codes and industrial seismic modeling with a 3D stencil. Topic 8 on Distributed Systems and Algorithms is represented by the paper Applying the Dynamics of Evolution to Achieve Reliability in Master-Worker Computing authored by Evgenia Christoforou, Antonio Fernandez Anta, Chryssis Georgiou, Miguel A. Mosteiro, and Angel Sánchez 2. This is an original and interesting approach to unreliable Internet-based master-worker computing. The assumption is that the workers cannot be trusted to return correct results and that their interests may even change during the course of the computation. The authors employ the dynamics of evolution to study the conditions under which the master can obtain reliable results. Via the technique of reinforcement learning, the workers receive incentives to become truthful within a bounded time. Topic 9 on Parallel and Distributed Programming is represented by the paper Extending the Scope of the Checkpoint-on-Failure Protocol for Forward Recovery in Standard MPI authored by Wesley Bland, Peng Du, Aurelien Bouteiller, Thomas Herault, George Bosilca, and Jack J. Dongarra 3. Checkpoint-on-Failure (CoF) is a new protocol for fault-tolerant MPI applications that avoids the need for periodic checkpointing and that, unlike other approaches, only requires standard MPI. At the occurrence of a failure, the application makes a checkpoint as last action before abortion. This checkpoint is reloaded in a new MPI application, which recovers from the failure. The validity and performance of this approach are demonstrated on the example of QR factorization. Topic 11 on Multicore and Manycore Programming is represented by the paper Efficient Support for In-Place Metadata in Java Software Transactional Memory authored by Ricardo J. Dias, Tiago M. Vale, and João M. Lourenço 4. The accesses in software transactional memories are governed by metadata. There are two strategies for associating these metadata with its memory location: in the out-place strategy, a separate table is constructed; in the in-place strategy, the metadata are stored adjacant to the memory cell. A fair comparison of both strategies is difficult because implementations usually have a strong bias toward one of them. The authors offer an in-place strategy that facilitates an unbiased implementation and offer one such implementation in the system Deuce. This enables them to conduct a fair comparison of the two metadata management strategies. Topic 13 on High-Performance Network and Communication is represented by the paper Tailoring the Network to the Problem: Topology Configuration in Hybrid EPS/OCS Interconnects authored by Kostas Christodoulopoulos, Kostas Katrinis, Marco Ruffini, and Donal O'Mahony 5. The authors consider a hybrid electronic packet switching (EPS) and reconfigurable optical circuit switched (OCS) network for future high-performance computing and data center systems. Their objective is to map the task communication graph of an application to the compute resources and, thereby, find an optimal configuration of the optical circuit. They prove the NP-completeness of this problem and offer two algorithms: one for the small and one for the large scale. The small-scale algorithm is based on integer linear programming and finds an optimal solution, the large-scale algorithm is based on a simulated annealing heuristics and trades off performance for responsiveness. The reviewers were particularly pleased by the thoroughness of this treatise. Concluding this preface, we would like to thank Prof. Geoffrey Fox and Prof. Luc Moreau, editors of this journal, for their support of this special issue. We would also like to thank our peers who assisted us in reviewing the papers and helped strengthen the final versions. Last, but not least, we also appreciate the support of Springer, who agreed to the publication of the extended versions of the articles that appeared originally in the series Lecture Notes in Computer Science.

This special issue of Concurrency and Computation: Practice and Experience contains revised and extended versions of selected papers presented at the conference Euro-Par 2011. Euro-Par—the European Conference on Parallel Computing—is an annual series of international conferences dedicated to the promotion and advancement of all aspects of parallel and distributed computing. Euro-Par covers a wide spectrum of topics from algorithms and theory to software technology and hardware-related issues, with application areas ranging from scientific to mobile and cloud computing. The major part of the Euro-Par audience consists of researchers in academic institutions, government laboratories and industrial organizations. Euro-Par 2011, the 17th conference in the Euro-Par series, was organized by the National Institute for Research in Computer Science and Control (INRIA) and the University of Bordeaux, and held at the University of Bordeaux in the center of Bordeaux, France. Sixteen broad topics were defined and advertised, covering a large variety of aspects of parallel and distributed computing. The call for papers attracted a total of 271 submissions. The submitted papers were reviewed at least three and, in many cases, four times (3.58 on average). A total of 81 papers were finally accepted for publication. This makes a global acceptance rate of 29.9%. The authors of accepted papers come from 26 countries, with the four main contributing countries—the USA, France, Spain and Germany—accounting for about 64% of them. The distribution of papers follows the pattern typical for a Euro-Par conference: 58% are authored by academic researchers, 26% by students, and 12% by other authors (industry, non-governmental organizations, and government). With the results of the reviews and a majority opinion of the respective topic program committees, several papers were recommended for a special journal issue. The authors were contacted at the conference and invited to submit revised and extended versions of their papers. These new versions were reviewed independently by three reviewers; two had also reviewed the conference version, the third had not. Eventually, six papers were accepted for publication. Topic 2 on Performance Prediction and Evaluation is represented by the paper Backfilling with guarantees made as jobs arrive, authored by Alexander Lindsay, Maxwell Galloway-Carson, Christopher Johnson, David Bunde, and Vitus Leung 1. In job scheduling, the term backfilling refers to the approach of assigning a job an earlier time than the predicted starting time if holes appear in the scheduling plan. Their new contribution is to give preference to jobs with higher priority, as defined by a job selection function. They explore two different such functions and validate their benefits with simulation experiments. Topic 5 on Parallel and Distributed Data Management is represented by the paper ISABELA for effective in situ compression of scientific data, authored by Sriram Lakshminarasimhan, Neil Shah, Stephane Ethier, Seung-Hoe Ku, C. S. Chang, Scott Klasky, Robert Latham, Robert Ross, and Nagiza F. Samatova 2. The authors address the fundamental problem of compressing the terabytes of numerical data produced by modern large-scale scientific simulations on high performance computing (HPC) systems. Classical lossless compression schemes are hardly suitable in this case, as such data are highly entropic by nature. On the other hand, traditional lossy schemes cannot be used, as they do not respect the numerical semantics of such data. The authors propose a surprisingly elegant preconditioning idea to overcome this problem: first, sort the numbers, saving the correspondence table for later restoration, and then approximate the resulting increasing curve by some spline, with a controlled error. Their experiments on actual scientific data sets show exceptionally good results, typically achieving an impressive 80% compression rate. Topic 6 on Grid, Cluster and Cloud Computing is represented by the paper Shrinker: efficient live migration of virtual clusters over WANs, authored by Pierre Riteau, Christine Morin and Thierry Priol 3. Shrinker is a system, which detects common portions of virtual machines on separate nodes in a wide-area network, and prevents them from taking part in the migration of a virtual machine from one node to another. Although the concepts are not new, one reviewer appreciates explicitly the good use of the data deduplication technique. He also praises the authors’ nice solution that would actually make sense in the ‘real world’, in contrast to most contributions to live migration that are pure prototypes or proofs of concept. Topic 7 on Peer-to-Peer Computing is represented by the paper Gossip learning with linear models on fully distributed data, authored by Róbert Ormándi, István Hegedü, and Márk Jelasity 4. Their problem is to extract knowledge from a set of data that is widely distributed across a large peer-to-peer network. Their constraint is that the data must stay at the node on which they reside, due to security or privacy requirements. The authors call their approach gossip learning: multiple models take random walks across the network to improve themselves. Their new idea is to combine a large number of models repeatedly, leading to a kind of voting mechanism. They prove convergence and provide extensive benchmark results. One reviewer praises the paper as particularly original in a field that has already been well studied. Topic 9 on Parallel and Distributed Programming is represented by the paper Correlated set coordination in fault tolerant message logging protocols for many-core clusters, authored by Aurelien Bouteiller, Thomas Herault, George Bosilca, and Jack Dongarra 5. Next-generation exascale HPC systems will inherently have to cope with very frequent failures. Some failure-recovery mechanism is mandatory but, unfortunately, the classical uncoordinated checkpoint approach is not scalable. The authors point out that the failures in such systems are not distributed uniformly: two processes running on cores inside the same node are very likely to fail together. Therefore, they propose a hybrid approach: coordinated checkpointing between failure-correlated processes, typically on the same node, uncoordinated checkpointing between failure-independent ones, typically on distinct nodes. Combined with the fact that communication patterns are likely to be consistent with the hardware topology, this simple and elegant idea leads to a drastic reduction of log volume. Topic 16 on GPU and Accelerators Computing is represented by the paper Iterative sparse matrix–vector multiplication for accelerating the block Wiedemann algorithm over GF(2) on multi-GPU systems, authored by Bertil Schmidt, Hans Aribowo and Hoang Vu Dang 6. Integer factorization constitutes the core of RSA cryptographic methods. The number field sieve (NFS), the current state-of-the-art method, requires solving a large sparse linear system over a finite field GF(2) with just two values: 0 and 1. Although this may look trivial, it is a very complex challenge: for large integer numbers with, for instance, more than 750 decimal digits, running times are on the order of months. The key time-consuming operation is the iterative multiplication of a sparse matrix with a vector. The authors propose an efficient CUDA implementation of it using a newly designed hybrid sparse matrix format. They also demonstrate the use of not only one GPU but a cluster of GPUs, achieving considerable speed-up over traditional CPU cluster and grid implementations. Concluding this preface, we would like to thank Prof. Geoffrey Fox and Prof. Luc Moreau, editors of this journal, for their support of this special issue. We would also like to thank our peers who assisted us in reviewing the papers and helped strengthen the final versions. Last, but not least, we also appreciate the support of Springer, who agreed to the publication of the extended versions of the articles that appeared originally in the series Lecture Notes in Computer Science.