The rapid scaling of microprocessors has shifted the critical bottleneck of high‐performance computing systems from the computational units to the communication infrastructure. By taking advantage of the parallelism and capacity of dense wavelength‐division‐multiplexed (DWDM) technology, optical interconnects using nanophotonics offer a high‐bandwidth, low‐latency, and energy‐efficient solution within a small footprint, when compared with their electrical counterparts.

关键词：光学互连；自适应；集成硅光子收发器

查看摘要 索取原文[2页]

We are developing an optical line-of-sight signaling technology that enables mobile devices such as smartphones, tablets, and laptops to identify, locate, and receive information from other devices and infrastructure within their vicinity. Using this technology we can determine where particular devices are with respect to each other in a room - the angles between the devices - and we can use our mobile devices to read light-emitting codes similar to Quick Response (QR) codes in the environment. Our technology uses a robust encoding method and can tolerate shaking due to unsteady hands holding the mobile devices. Experimental results using hand-held cameras show our technology can accurately and robustly locate and decode signals from small LED emitters even at indoor distances of at least 5 meters.

关键词：视线；设备定位；设备标识；光信号；移动设备

查看摘要 索取原文[7页]

Computer crashes threaten application data integrity. The threat is particularly acute for expensive industrial equipment such as high-volume printing presses, whose stringent uptime requirements demand both high performance during normal operation and rapid recovery following crashes. We have designed and implemented a novel general-purpose persistent memory buffer that protects application data from crashes and is easy to integrate into existing software. Our solution slid with remarkable ease beneath the mature, complex, highly tuned software that controls HP Indigo printing presses, reducing recovery times from days to minutes while adding negligible overhead to failure-free operation. The new system has been in successful production use at dozens of beta test sites for several months and will eventually ship with all new Indigo presses. Our novel crash resilience strategy is not specific to Indigo presses and is likely applicable in a wide range of HP products, so we have developed a portable implementation that is available upon request.

关键词：容错；靛蓝；存储；持续堆

查看摘要 索取原文[4页]

Expedite the troubleshooting of InfoSphere? BigInsights? applications using the IBMAccelerator for Machine Data Analytics with IBM InfoSphere BigInsights to do Hadoop loganalysis.

关键词：加速器；机器数据分析；故障排除

查看摘要 索取原文[16页]

Given two arbitrary sequences of denoisers for block lengths tending to infinity we ask if it is possible to construct a third sequence of denoisers with an asymptotically vanishing (in block length) excess expected loss relative to the best expected loss of the two given denoisers for all clean channel input sequences. As in the setting of DUDE, which solves this problem when the given denoisers are sliding block denoisers, the construction is allowed to depend on the two given denoisers and the channel transition probabilities. We show that under certain restrictions on the two given denoisers the problem can be solved using a traightforward application of a known loss estimation paradigm. We then show by way of a counter-example that the loss estimation approach fails in the general case. Finally, we show that for the binary symmetric channel, combining the loss estimation with a randomization step leads to a solution to the stated problem under no restrictions on the given denoisers.

关键词：损失评估范式；渠道转型概率；直接应用

查看摘要 索取原文[6页]

Many real-world graphs have complex labels on the nodes and edges. Mining only exact patterns yields limited insights, since it may be hard to find exact matches. However, in many domains it is relatively easy to compute some cost (or distance) between different labels. Using this information, it becomes possible to mine a much richer set of approximate subgraph patterns, which preserve the topol- ogy but allow bounded
label mismatches. We present novel and scalable methods to efficiently solve the approximate isomorphism problem. We show that the mined approximate patterns yield interesting patterns in several real-world
graphs ranging from IT and protein interaction networks to protein structures.

关键词：数据挖掘；图分析；近似技术

查看摘要 索取原文[10页]

Cloud computing is emerging as an alternative to supercomputers for some of the high-performance computing (HPC) applications that do not require a fully dedicated machine. With cloud as an additional deployment option, HPC users are faced with the challenges of dealing with highly heterogeneous resources, where the variability spans across a wide range of processor configurations, interconnections,virtualization environments and pricing rates and models. In this paper, we take a holistic viewpoint to answer the question ? why and who should choose (or not choose) cloud for HPC, for what applications,and how should cloud be used for HPC? To this end, we perform a comprehensive performance evaluation of a set of benchmarks and complex HPC applications on a range of platforms, varying from supercomputers to commodity clusters, both in-house and in the cloud. Having identified the performance bottlenecks in cloud, we demonstrate that alternative lightweight virtualization mechanisms such as thin VMs and OS-level containers, and hypervisor and applicationlevel CPU affinity can greatly lower the overhead and noise of virtualization. Next, we discuss the economic aspects of HPC in the cloud, which we believe is an important area that has not been sufficiently addressed by past research. Finally, we explore the benefits of an intelligent matching of an HPC application to the best deployment platform within a given set, discussing how we can combine a cloud deployment with a supercomputer. Overall results indicate that current public clouds are cost-effective only at small scale ? 4 to 16 cores for the chosen HPC applications, when considered in isolation, but can complement (and co-exist with) supercomputers using models such as cloud burst and application-aware mapping to achieve significant cost benefits.

关键词：高性能计算；云计算；应用程序；超级计算机；绩效评估；经济学

查看摘要 索取原文[13页]

Learn how to successfully capture a running Windows instance to a private catalog in theIBM? SmartCloud? Enterprise environment. Additionally, the authors describe common issuesthat arise during the process and how to troubleshoot them.You have a base IBM SmartCloud Enterprise Windows 2008 instance up and configured exactlythe way you want it and then you decide you need to provision a new instance from this state.Perhaps you want to run multiple instances of your solution, update software to a reference level,or simply want a new golden image to start from. This article guides you through the standardcapture process for a typical SmartCloud Enterprise Windows 2008 instance and points out somepossible stumbling blocks that you might encounter along the way.

关键词：互联网同步时间；捕捉脚本；故障排除

查看摘要 索取原文[6页]

Back in the day, you could bring your personal Blackberry to work in order to access aSoftware as a Service (SaaS) application (if you were in a large building with good cellularreception). Now, you have many device choices for accessing the cloud, from iOs-basedphones and tablets, Android handhelds, to Research In Motion's (RIM) PlayBook. All comewith opportunities for jailbreaking, or gaining root access, by breaking down security to installbanned third-party applications. An infected bring your own device (BYOD) connected to thecorporate network is another security concern. A war walker could steal corporate data byuploading from your BYOD to a personal device. Learn to protect corporate assets with asecurity policy for a cloud-based BYOD environment.

关键词：战争散步；系留设备的安全；圈养的政策决定；信息安全；云计算

查看摘要 索取原文[11页]

Big data is a term that has been used regularly now for almost a decade, and it — alongwith technologies like NoSQL — are seen as the replacements for the long-successfulRDBMS solutions that use SQL. Today, DB2?, Oracle, Microsoft? SQL Server MySQL, andPostgreSQL dominate the SQL space and still make up a considerable proportion of the overallmarket. In this final article of the series, we will look at more automated solutions for migratingdata to and from Hadoop. In the previous articles, we concentrated on methods that takeexports or otherwise formatted and extracted data from your SQL source, load that into Hadoopin some way, then process or parse it. But if you want to analyze big data, you probably don'twant to wait while exporting the data. Here, we're going to look at some methods and tools thatenable a live transfer of data between your SQL and Hadoop environments.

关键词：直接传输；数据交换；实时；大数据

查看摘要 索取原文[13页]