MER Center

13 posts

Data Center Building-Blocks

Posted by BrandiX May 8, 2012

Due to the varied utilization of servers in a data center at any given moment, energy consumption typically runs higher that strictly necessary to keep the servers at their most useful. To try improving energy efficiency, three researchers in California tested paticular clusters for their effectiveness, with positive results.


Read the paper on SpringerLink: The Search for Energy-Efficient Building Blocks for the Data Center

The wireless sensors used in hard-to-reach places are often deployed in somewhat haphazard networks that can hurt their usefulness. In a paper from the International Journal of Computer Applications, three individuals in computer science and engineering from India detail a more carefully set grid for sensor clusters to minimize energy use and maximize battery life.


Access the PDF on An Energy-efficient Grid based Clustering Topology for a Wireless Sensor Network


Improving Mobile Memory I/O

Posted by BrandiX Apr 20, 2012

As phones become able to do more and are used for more intense operations, their batteries suffer and processor limitations get tight, so operations take longer while the charge lasts for a shorter time. To combat the issue, several engineers present a possible I/O interface with bi-directional dual-band signaling, to improve energy efficiency and increase bandwidth.


Read the full paper at An Energy-Efficient and High-Speed Mobile Memory I/O Interface Using Simultaneous Bi-Directional Dual (Base+RF)-Band Signaling


Looking to Nature

Posted by BrandiX Apr 10, 2012

Scientists often use math and computing to better understand nature and biology, but Yehuda Afek, Noga Alon, Omer Barad, Eran Hornstein, Naama Barkai, and Ziv Bar-Joseph believe that norm is worth turning around. They beleive that a study of biological systems can help better define structures for distributed computing.


Read the full paper on A Biological Solution to a Fundamental Distributed Computing Problem

Using maximal information coefficient and maximal information-based nonparametric exploration statistics, nine researchers discuss how to discover previously unrecognized relationships in any variety of big sets of data. Finding these new relationships can lead to new ways of understanding a given set of data.


Read more on Detecting Novel Associations in Large Data Sets

To reduce the energy used by sorting algorithms, Cheng-Jen Tang, Hui-Chin He, Miau-Ru Dai, and Chi-Cheng Chuang have experimented with evaluating the memory required for using different sorting algorithms on different ammounts of data to sort. They also discuss the idea memory useage by a sorting algorithm.


Download the PDF from Evaluating Memory Size for Energy Efficient Sorting


Real-Time Scheduling Algorithm

Posted by BrandiX Mar 20, 2012

To better deal with and reduce the energy consumed for real-time and sporadic tasks in multi-core systems, researchers in China and Austrialia developed an algorithm that focuses on Time Local - Dynamic Voltage Frequency Scaling. Simulations of the algorithm have all shown promise.


Access the full paper in PDF in the Association for Computing Machinery's Digital Library (available via the Intel Library): TL-plane-based multi-core energy-efficient real-time scheduling algorithm for sporadic tasks

A collection of scientists and researchers from the University of California's Departments of Physics and of Electrical Engineering, the California NanoSystems Institute, and NEC Corporation's Green Innovation Research Laboratories in Japan demonstrate and discuss quantum central processing in a "classical computer" using the von Neumann architecture. Their results show promise for factoring and quantum error correction.


Read the paper at Implementing the Quantum von Neumann Architecture with Superconducting Circuits

To improve the energy effiency of clustered wireless sensor networks, Jayanthi K. Murthy, Suthikshn Kumar, and A. Srinivas present a scheduling algorithm focused on optimizing "Time Division Multiple Access." Using it, they have gotten results that show reduced delay and energy used per-bit.


Read or download the paper from the International Journal of Computer Science and Communication: Energy Efficent Scheduling in Cross Layer Optimized Clustered Wireless Sensor Networks

Modular functions can be calculated by threshold circuits that save energy by mimicking the brain's nuerons, according to Akira Suzuki, Kei Uchizawa, and, Xiao Zhou.


Read their proof in The Australian Computer Society'sTheory of Computing 2011: Energy-Efficient Threshold Circuits Computing Mod Functions


Jungle Computing

Posted by BrandiX Feb 22, 2012

Even as systems are being refined and simplified to improve efficiency, others are looking deeper into more distributed systems that use multiple platforms at once. Current issues and desires with scalability, data distribution, and more seem to be pushing things further toward "jungle computing."


Read or download the chapter from Grids, Clouds, and Virtualization at SpringerLink: Jungle Computing: Distributed Supercomputing Beyond Clusters, Grids, and Clouds


Exascale Supercomputers

Posted by BrandiX Feb 15, 2012

Researchers already have uses for supercomputers capable of exaflops (1018 floating point operations per second) - such as accurate global climate simulation - but existing technology cannot yet support it and the current trend won't make it practical. To reach achieve the goal, systems will have to be far more efficient with energy, memory, error-handling, and more.


Read the article by Robert F. Service at What It'll Take to Go Exascale

Researchers from IBM*, the Max Planck Research Group, and the University of Basel in Switzerland explore the magnetic control of antiferromagnetic nanostructures to make data storage devices even smaller. Paper by Sebastian Loth, Susanne Baumann, Christopher P. Lutz, D. M. Eigler, and Andreas J. Heinrich.


See the paper at Bistability in Atomic-Scale Antiferromagnets.