Conference
High Performance Computing Systems and Applications
About: High Performance Computing Systems and Applications is an academic conference. The conference publishes majorly in the area(s): Parallel algorithm & Scalability. Over the lifetime, 369 publications have been published by the conference receiving 3011 citations.
Papers published on a yearly basis
Papers
21 Jul 2014
TL;DR: The UL HPC facility and the derived deployed services is a complex computing system to manage by its scale and the aspects in relation to the management of such a complex infrastructure, whether technical or administrative are covered.
Abstract: The intensive growth of processing power, data storage and transmission capabilities has revolutionized many aspects of science. These resources are essential to achieve high-quality results in many application areas. In this context, the University of Luxembourg (UL) operates since 2007 an High Performance Computing (HPC) facility and the related storage by a very small team. The aspect of bridging computing and storage is a requirement of UL service - the reasons are both legal (certain data may not move) and performance related. Nowadays, people from the three faculties and/or the two Interdisciplinary centers within the UL, are users of this facility. More specifically, key research priorities such as Systems Bio-medicine (by LCSB) and Security, Reliability & Trust (by SnT) require access to such HPC facilities in order to function in an adequate environment. The management of HPC solutions is a complex enterprise and a constant area for discussion and improvement. The UL HPC facility and the derived deployed services is a complex computing system to manage by its scale: at the moment of writing, it consists of 150 servers, 368 nodes (3880 computing cores) and 1996 TB of shared storage which are all configured, monitored and operated by only three persons using advanced IT automation solutions based on Puppet [1], FAI [2] and Capistrano [3]. This paper covers all the aspects in relation to the management of such a complex infrastructure, whether technical or administrative. Most design choices or implemented approaches have been motivated by several years of experience in addressing research needs, mainly in the HPC area but also in complementary services (typically Web-based). In this context, we tried to answer in a flexible and convenient way many technological issues. This experience report may be of interest for other research centers and universities belonging either to the public or the private sector looking for good if not best practices in cluster architecture and management.
352 citations
23 Jun 1997
TL;DR: This paper shows the design of a Reliable Multicast data Distribution Protocol (RMDP) that is built using FEC techniques, and Experimental results show that, albeit somewhat expensive, doing encoding/decoding in software is affordable for speeds in the Mbit/s range even on low-end PCs.
Abstract: Applications requiring the reliable distribution of data to groups of clients would be supported perfectly by reliable multicast protocols. In many cases, the problem of congestion control (a major research issue otherwise) does not exist because downlink bandwidth is "owned" or can be preallocated to a particular server by independent means, but the problems of insuring reliable data delivery to larcre groups, and adaptability to heterogeneous clients, still remain. These problems can be solved at once with the use of FEC techniques. In this paper we show the design of a Reliable Multicast data Distribution Protocol (RMDP) that we have built using these techniques, and discuss the implementation tradeoffs. Experimental results show that, albeit somewhat expensive, doing encoding/decoding in software is affordable for speeds in the Mbit/s range even on low-end PCs. Slower machines can still receive at high speed, thus optimizing network usage, by taking advantage of the fact that decoding needs not to be done in real time. Finally, our RMDP can work even without any feedback from the receivers, thus making it well suited to mobile/wireless systems.
148 citations
15 May 2005
TL;DR: This article presents the available programs that can provide, with different technology, a virtualization of Linux computers and defines some usage criteria that allows the reader to chose the relevant virtualization technology according to its specific needs.
Abstract: This article presents the available programs that can provide, with different technology, a virtualization of Linux computers. We define some usage criteria that allows the reader to chose the relevant virtualization technology according to its specific needs. We focus on the Linux-VServer technology, which is a very lightweight and effective technology for the regular Linux user not interested in Kernel hacking. The Linux-VServer project also supports additional security options and resources limitation that can be very useful. This is also the most mature open-source technology and several users have production servers using this technology for years now.
71 citations
21 Jul 2014
TL;DR: It is believed that the new energy formulation and the two new heuristics contribute significantly towards achieving green cloud computing.
Abstract: Cloud data centers consume an enormous amount of energy. Virtual Machine (VM) migration technology can be applied to reduce energy consumption by consolidating VMs onto the minimal number of servers and turn idle servers into power-saving modes. While most existing energy models consider mainly computing energy, an enhanced energy consumption model is formulated, which includes energy consumption for computation, for servers to switch from standby to active modes, and for communication during VM migrations. Next, two new dynamic VM migration algorithms are proposed. They apply a local regression method to predict potentially over-utilized servers, and the 0-1 knapsack dynamic programming to find the best-fit combination of VMs for migration. The time complexity of these algorithms is analyzed, which indicates that they are highly scalable. Performance is evaluated and compared with existing algorithms. The two new heuristics have significantly reduced the number of VM migration, the number of rebooted servers, and energy consumption. Furthermore, one of them has achieved the least overall SLA violations. We believe that the new energy formulation and the two new heuristics contribute significantly towards achieving green cloud computing.
62 citations
13 May 2007
TL;DR: It is shown that it is possible to get speed increases of several hundred times over a typical CPU implementation, catapulting GPU processing for genetic programming approaches into the realm of HPC.
Abstract: In this paper we demonstrate the use of the graphics processing unit (GPU) to accelerate evolutionary computation applications, in particular genetic programming approaches. We show that it is possible to get speed increases of several hundred times over a typical CPU implementation, catapulting GPU processing for these applications into the realm of HPC This increase in performance also extends to artificial developmental systems, where evolved programs are used to construct cellular systems. Feasibility of this approach to efficiently evaluate artificial developmental systems based on cellular automata is demonstrated.
62 citations
Performance Metrics
| Year | Papers |
|---|---|
| 2021 | 1 |
| 2020 | 17 |
| 2019 | 7 |
| 2018 | 4 |
| 2017 | 13 |
| 2016 | 8 |