The Community for Technology Leaders

Very Large-Scale Systems and Some History

Erik R. ,

Pages: pp. 2-3

Abstract—This column briefly introduces the issue's selections and discusses the ACM SIGMICRO's Oral History project.

Keywords—Very large-scale computing, ACM SIGMICRO Oral History project, distributed computing, data center, accelerator, simulation, power, energy

This issue marks the first special issue developed under my editorship. Guest editors Natalie Enright Jerger and Mikko Lipasti have done a fine job recruiting five excellent articles on the theme of systems for very large-scale computing. We are seeing a proliferation of styles for very large-scale computing—from data centers for commercial computing to highly distributed efforts like SETI@home or ProteinFolding@home, and from systems composed of high-performance processors like Blue Waters to systems relying on GPUs for most of their performance, like the Tianhe-1A system (ranked #1 in the Top500 as I write this), to specialized systems like D.E. Shaw's Anton system (see page 8). With so much variation and flux, it is important to study such systems so we can understand how best to proceed with future systems.

In this issue

In addition to the Anton article, "Overcoming Communication Latency Barriers in Massively Parallel Scientific Computation" by Ron O. Dror et al., this issue has several other articles that help us understand the design of very large systems.

Two of these articles also address specialized systems. "CogniServe: Heterogeneous Server Architecture for Large-Scale Recognition" by Ravi Iyer et al. continues in the special-purpose vein by describing a system designed to efficiently perform many types of recognition, from images to speech to text.

In "Simulating Whole Supercomputer Applications," Juan Gonzalez et al. step back and look broadly at how the performance of such systems can be modeled effectively. Similarly, "Automated Full-System Power Characterization" by Stijn Polfliet et al. provides a framework for estimating such systems' power. "Energy-Aware Accounting and Billing in Large-Scale Computing Facilities" by Víctor Jiménez et al. works in a related vein to provide users and owners an accurate accounting of the energy requirements in such large-scale systems.

As with previous issues, you can rate the articles and overall quality of this issue at

Alas, we have had limited response to the surveys for the first two issues of 2011, and not enough to provide a statistically significant view of favorite articles. However, one anonymous respondent asked, "Can we broaden the base of Micro to make it attractive for other practitioners in computing? Micro is a great magazine in general, but its impact seems limited due to a limited number of subscribers. Computer is very well known in the field, for example, but Micro is not." Please let me know your views on this suggestion, either by further anonymous comments via surveys or by e-mailing me at

ACM Oral History project

Finally—and despite this being an IEEE publication—I would like to draw attention to ACM SIGMICRO's Oral History project, which is being led by Yan Solihin. Yan engaged historian Paul Edwards to interview two luminaries in our field, Bob Colwell and Edward Davidson. Transcripts and recordings of those interviews are now available online. I think they will be of interest to many IEEE Micro readers, both in personal and technical terms (to learn, for example, that in the early 1990s, Intel was worried that every generation of x86 would be the last). You can read and hear more at


Erik R. Altman

Editor in Chief

IEEE Micro

66 ms
(Ver 3.x)