Read the TechNews Online at: http://technews.acm.org
ACM TechNews
November 29, 2006

MemberNet
The ACM Professional Development Centre
Unsubscribe

Welcome to the November 29, 2006 edition of ACM TechNews, providing timely information for IT professionals three times a week.

Sponsored by Information, Inc.

http://www.infoinc.com/sponsorthenews/contactus.html


HEADLINES AT A GLANCE:

 

E-Voting Measures Sought to Avoid Disputes
San Jose Mercury News (CA) (11/29/06) Davies, Frank

California Senator Diane Feinstein, who will take over the Rules and Administration Committee that oversees how federal elections are run, has made it clear that she will scrutinize the e-voting process. "It's imperative that Congress does everything it can to help ensure that votes cast are recorded accurately," Feinstein said. "Serious questions have arisen about the accuracy and reliability of new electronic voting machines." Even before the previous election, in which Sarasota Country, Fla., confirmed the concerns many had about e-voting, she had been planning legislation, similar to one that failed to pass in the House by two votes, mandating a paper trail for all electronic voting systems in the country. Electiononline.org's Doug Chaplin said, "At first I thought there were lots of fender-benders on Election Day but no major pile-ups. But Sarasota is a pile-up." State officials and voting machine manufacturers are being pointed at to do a better job of testing and auditing equipment before elections. Republicans are pushing for legislation ensuring voter ID and fraud prevention, and Feinstein herself wants to outlaw state election officials from taking part in a federal candidate's campaign committee. Feinstein worries that continued problems, in a district that has greater national ramifications than Sarasota County, or worse, in a presidential election, will lead to a harmful loss of confidence in the nation's ability to conduct elections. Stanford University computer science professor David Dill said lost votes complaints following the 2004 elections were not adequately investigated. He says, "The complaints need to be investigated urgently, or machine problems will lead to more disputed elections in the future." For information about ACM's e-voting activities, visit http://www.acm.org/usacm
Click Here to View Full Article
to the top


The Problem With Programming
Technology Review (11/28/06) Pontin, Jason

Bjarne Stroustrup, who invented C++, explains in this interview that he still stands by the language he built, and thinks that most programming being done now is below par. While there is quality software out there, like Google, he says, "looking at the 'average' piece of code can make me cry. The structure is appalling, and the programmers clearly didn't think deeply about correctness, algorithms, data structures, or maintainability." Rather than being sure of a system's quality and why it works so well, Stroustrup says programmers are "in a constant state of grasping at straws to get our work done. The snag is we often do not know how we did it: a system just 'sort of evolved' into something minimally acceptable." In order to remedy this situation, he thinks that education must be improved, using "more-appropriate design methods, and design for flexibility and for the long haul." However, this fix is difficult to achieve because computer users do not want to be inconvenienced by abrupt changes; only a gradual, wide-ranging effort toward change will be effective. "Software developers have neutralized the astounding performance of modern computer hardware by adding layer upon layer of over-elaborate [software] abstractions," says Stroustrup, whose solution is that more experts should be trained to use C++, as it as fallen out of the mainstream, rather than simply "dumb[ing] down" programming languages. He says the generality built into C++ was the result of his "view that to do higher-level stuff, to build complete applications, you first needed to buy, build, or borrow libraries providing appropriate abstractions." Stroustrup believes that the large amount of criticism that has been aimed at C++ is a testament to how useful it really is.
Click Here to View Full Article
to the top


Design Automation Conference Announces Executive Committee
Business Wire (11/27/06)

Steven P. Levitan, the former chair of the ACM Special Interest Group on Design Automation (SIGDA), will serve as the general chair of the executive committee for the 44th Design Automation Conference (DAC). Dr. Levitan is the John A. Jurenko Professor of Computer Engineering in the Department of Electrical and Computer Engineering at the University of Pittsburgh, and he also has a joint appointment with the Department of Computer Science. He will guide the committee of volunteers from the electronics and electronic design automation (EDA) industry in planning and overseeing management of DAC's operations, including technical programs, exhibitions, new projects, and publicity. Levitan, an expert in design, modeling, simulation, and verification of mixed technology micro-systems, has been involved in DAC executive committees since 1998. The 44th DAC is scheduled for June 4-8, 2007, at the San Diego Convention Center in San Diego, Calif. Carnegie Mellon University's Diana Marculescu will serve as the ACM/SIGDA representative on the executive committee. Other members of the executive committee include Limor Fix of Intel Research, Leon Stok of IBM, Sachin Sapatnekar of the University of Minnesota, Yervant Zorian of Virage Logic, Ellen M. Sentovich of Cadence Berkeley Labs, Narendra Shenoy of Synopsys, Andrew B. Kahng of the University of California at San Diego, Kaushik Roy of Purdue University, Nanette V. Collins of Nanette V. Collins Marketing and PR, Georges Gielen of Katholieke University in Belgium, and Yusuke Matsunaga of Kyushu University Kagus in Japan. "With the combined energy, expertise and experience of this remarkable group of volunteers driving it, we are looking forward to a very strong conference in San Diego next June," says Levitan. For more information on DAC 2007 visit http://www.dac.com/44th/index.html
Click Here to View Full Article
to the top


Vote Disparity Still a Mystery in Fla. Election for Congress
Washington Post (11/29/06) P. A3; Whoriskey, Peter

Florida's 13th Congressional District is still trying to get to the bottom of why there were no votes cast for Congress by 18,000 Sarasota County residents who voted for candidates in other races. Some claim that the touch-screen voting system had a glitch that dropped votes, others that a confusing ballot caused voters to overlook the race, and finally that voters simply decided not to vote in this particular race, a possibility that has received little support. "Our analysis of the results show that something went very wrong," says Kendall Coffey, attorney for challenger Christine Jennings, who is currently being declared the loser of the race, pending further investigation. Coffey dismissed a mock election that showed no signs of machine malfunction, in which clerical workers, not ordinary voters, used the machines to place votes. While 2.5 percent of voters did not cast a vote in every race in other Florida counties, a phenomenon known as "undervoting,"15 percent undervoted in Sarasota County. Two different election experts who had their own troubles with the voting machines support the theory that the machines are to blame, and the Sarasota Herald-Tribune reports over 100 reported problems with the machines. The confusing ballot idea is supported by the CalTech/MIT Voting Technology Project's director, MIT's Ted Selker, who claims that his own tests show 60 percent of voters possibly missing races that are displayed in the way that the race in question was, but Coffey claims that such a high profile race is very unlikely to be simply forgotten or overlooked by so many voters.
Click Here to View Full Article
to the top


Declining Comp. Sci Enrollment Levels Off
Yale Daily News (11/29/06) Balakrishna, Kanya

The number of computer science majors is no longer on the decline at Yale University, says Computer Science Department Chairman Avi Silberschatz, and Stan Eisenstat, director of Undergraduate Studies, believes the number will rise next year, which would mark the first increase in five years. Reflecting a nation-wide trend, Yale has seen its number of computer science majors fall from 71.5 students in the 2001-02 academic year to 24.5 last year. There has not been a comparable decline in enrollment in Yale's M.S. and Ph.D. programs. Nationally, the number of the computer science majors in 2005 was half of the total from 2000, and the Computer Research Association says the number of students pursuing a compute science degree has dropped 70 percent between 2000 and 2005, according to data compiled by the Higher Education Research Institute at the University of California, Los Angeles. Meanwhile, other Ivy League schools have also reported large swings in the number of computer science majors; at Harvard University, computer science majors fell from 174 in 2001 to 67 in 2005, while at Princeton University, students majoring in computer science fell to 14 last year, down from 36 in 2000-2001. Although Eisenstat believes the apparent rebound of dot-coms may have something to do with the leveling off of computer science enrollment, computer science major Nick Piepmeier believes Web 2.0 is too much of a niche market to be such an influence. "I feel like in the aftermath of the bust people are finally realizing that it's still really easy to get jobs in the computer industry, and that there's still money to be made there," says Piepmeier, who serves on the departmental Student Advisory Committee. He also believes the number of computer science majors will pick up in the next few years.
Click Here to View Full Article
to the top


Engineers Seek to Equip Operating Room of the Future
JHU Gazette (11/27/06) Vol. 36, No. 12, Sneiderman, Phil

The operating room of the future could be filled with robots, visual displays, and digital workstations, according to engineers and computer scientists at the National Science Foundation Engineering Research Center for Computer-Integrated Surgical Systems and Technology. Researchers at the center on the campus of John Hopkins University believe their work has the potential to improve the safety of surgery, and allow surgeons to proceed with operations that otherwise would have unlikely been pursued. The robotic systems are meant to provide assistance to surgeons, and not replace them, center director and computer science professor Russell H. Taylor cautions. For example, a team of researchers has designed a snakelike robot to allow surgeons to be more precise in making incisions and tying sutures when operating in the throat area. Another team is developing a steady-hand system that is designed to offset uncontrolled hand movements with cooperative manipulation techniques, which should enable surgeons to have greater success in microsurgery. Such robotic assistants, visual displays guiding surgeons through procedures, and digital workstations offering instant access to medical information would all be connected to computers, which could serve as a black box for the operating room, and provide clues into why certain techniques are successful. The researchers believe the technology will one day appear in the operating room, but say several more years of testing and further development are needed.
Click Here to View Full Article
to the top


Backseat Virtual Reality Entertains Passengers
New Scientist (11/24/06) Simonite, Tom

Researchers at the Interactive Institute in Stockholm, Sweden, are testing an in-car gaming system that allows passengers to play an interactive game based on the buildings, forests, and rivers passed along a route while driving. The Backseat Playground uses such landmarks encountered during a trip to build a story, complete with in-game characters and events. The game matches sights for events in an adventure that might involve a murder mystery or a werewolf thriller, and makes use of a GPS receiver to provide geographical data, a handheld computer for player interaction as the story builds, and headphones for players to listen to phone calls and walkie-talkie messages from in-game characters. A laptop in the trunk, which correctly positions the car in the virtual world, connects the GPS receiver, handheld computer, and headphones. "We are trying to suggest spaces and places and events and have the user fill in the gaps to build a narrative," explains John Bichard, who developed the interactive game with colleagues Liselott Brunnberg and Oskar Juhlin. The computer scientists are considering integrating voice recognition into the game, which would allow players to talk directly to the characters. Rob Aspin, with the Center for Virtual Environments at Britain's University of Salford is intrigued by the way in which content is delivered for the game. "It can create a high sense of presence and interaction while hiding most of the technology from the user," says Aspin.
Click Here to View Full Article
to the top


Computers at MSU Take the Lead in High-Speed Studies of Evolution
Lansing State Journal (MI) (11/26/06) Miller, Matthew

The Digital Evolution Lab at Michigan State University is home to computers that simulate the evolution of billions of organisms at rates that would be impossible to observe in the natural world, and shed a great deal of light not only onto evolution, but computer science as well. "If you think of natural organisms, it takes months to years for a generation to go by for a sophisticated organism," says lab director Charles Ofria. "With these digital organisms, we can have a generation go by every second." Computer scientists have begun to use such observations of evolution at work to create strange and exciting innovations. The digital organisms that "live" on the computer's circuitry are simple programmed to self-replicate, but each time they do so, there is the possibility of a mutation occurring. Ofria and California Institute of Technology's Chris Adami created a program called Avida, where they can create habitats in which the organisms must try to survive, in order to stimulate natural selection; when the organisms are able to adapt, they are rewarded with extra computer processing time that allows them to reproduce faster. After many generations, the organisms carry stronger genetic codes, a process of finding solutions that is of great benefit to computer scientists: "In a sense, they're teaching us a shorter way of writing good code," says Ofria. The organisms self-replicate, randomly mutate, and thus adapt, "in such a way it would be extremely difficult for a human programmer to condense these millions of problems into a single, relatively seamless solution," Lenksi says.
Click Here to View Full Article
to the top


DOD Report to Detail Dangers of Foreign Software
Computerworld (11/27/06) Anthes, Gary

The Defense Science Board (DSB), a military/civilian think tank within the Defense Department, has conducted a study into the security of software developed overseas, and will make recommendations to the DoD based on its finding, but will not advise that all military software be created within the United States. Chairman of the task force Robert Lucky explains that, "The problem is we have a strategy now for net-centric warfare--everything is connected. And if the adversary is inside your network, you are totally vulnerable." The private sector has already experienced changes based on the task forces findings, although many see this attitude as simply xenophobia, stating that all software should be scrutinized equally. Lucky says that users should aim to make trade-offs between the amount of risk and the economics of creating a given piece of software. Protective steps cited by the DSB are: Peer reviews where several programmers review and test code; utilizing scan tools to search for hidden malware; and enforcing industry quality standards; and while each of these remedies is not a perfect fix or prevention, the combination will effectively "raise the bar," as Lucky says, and "eliminate a certain percentage of problems." However, those such as Ira Winkler, author of "Spies Among Us," feels that a single line of foreign-written code contained in U.S. military software is too much a security risk. While such a policy would ideally ensure against foreign malware, there are few, if any, U.S. software companies whose products do not contain any code written overseas, and according to Lucky, "we're talking about complexity that boggles the mind. It's so enormous that no can truly understand a program with millions of lines of source code."
Click Here to View Full Article
to the top


Software Patent Conference Outlines Problems, Possible Solutions
NewsForge (11/27/06) Bisbee-von Kaufmann, Samuel Kotel

Problems with software patents and possible solutions were the focus of the Nov. 17 "Software Patents: A Time for Change?" conference hosted by MIT and Boston University Law School. A temperature reading of the current software patent situation was taken during the first panel discussion, which mentioned companies' acquisition of software patents to defend against litigation from rivals seeking to generate profits from their own portfolios, and the opaque definition of patentable software by the European Patent Office, among other things. Bronwyn Hall of the University of California Berkeley Graduate School and the University of Maastricht observed in the second panel discussion that the growth of software patents does not reflect their value, which is for the most part nonexistent. Participants in the third panel pointed to the lack of consideration the World Wide Web Consortium had for patents initially because of concerns about technology; the creation of monopolies and the impedance of innovation by patent portfolios; and various reasons for the lack of emphasis on patents by entrepreneurs and startup firms. Legal ramifications were covered in the fourth panel, with panelists noting that a thing's patentability is a matter of perspective. For example, it was University of Akron School of Law professor Jay Draftler's opinion that a thing can only be designated an invention if it requires technical risk and thus the risk of failure. The final panel discussed possible reform strategies, and suggestions ranged from greater disclosure in patent applications via the required deposition of source code to increasing understanding of problems within the U.S. Patent Office through the provision of one-page documents to the creation of business incentives.
Click Here to View Full Article
to the top


Bye Swarmbots, Hello Swarmanoid
Wired News (11/28/06) Cole, Emmett

Free University of Brussels in Belgium researchers are developing a swarm of 60 small, autonomous, and specialized swarm bots, known as the "swarmanoid," because the swarm is made of specialized robots. Marco Dorigo, project leader and research director at the university's IRIDIA lab, says the swarm consists of "footbots" that are based on earlier, uniform swarm bots and move objects along the ground, "handbots" that climb walls, and "eyebots" that can attach to the ceiling to use their visual sensors; there are even swarm bots that will fly. By using specialized swarm bots, the same kind that can be seen in the division of labor in ant colonies, the swarmanoid will be able to complete more customized tasks, such as household chores or retrieving an object for a humanoid. Georgia Institute of Technology associate professor in interactive and intelligent computing Tucker Balch explains that, "For robots to really make an impact on the world, we have to get lots of robots into people's hands. The two barriers are cost and utility, but it becomes feasible with the swarm idea, which would allow households to buy several inexpensive robots that could work together. The view of swarms consisting of all the same robots just isn't going to take off." Swarms are also being considered for use at the micro or nano level for procedures inside the human body. Dorigo says he hopes to publish his work towards the end of 2007, with experimental results ready in about two years.
Click Here to View Full Article
to the top


Harnessing Grid Computing to Save Women's Lives
IST Results (11/29/06)

The accuracy of breast cancer diagnoses is receiving a valuable boost from grid computing. The rate of misdiagnosis of breast cancer can be as high as a 30 percent due to differences in individuals, equipment, procedures, and problems using the computers that detect changes in breast tissue. MammoGrid, an IST project that ended in August 2005, produced software that provides medical professionals access to digital mammograms stored across Europe. A geographically distributed, grid-based database consisting of 30,000 standardized images from 3,000 corresponding patient data, allows mammograms of current patients to be compared with others and subjected to detection algorithms to identify possible concerns. "The system in its current version allows a user to securely share both resources and patient data which has been treated to ensure anonymity," says Maat Gknowledge's David Manset, who served as leader of the project. Such an innovation brings about a new level of statistical analysis for breast cancer in its many forms, which will hopefully save many lives. The technology is currently being expanded to new hospitals and tested for its ability to meet market demands, before hopefully being expanded across all of Europe.
Click Here to View Full Article
to the top


Canada Experts Find Path Round Internet Firewalls
Reuters (11/28/06) Dabrowski, Wojtek

People living in countries that overly restrict Internet access and block Web sites will be able to circumvent the firewalls of their government using new software developed by computer researchers at the University of Toronto. The program, Psiphon, is designed to turn an Internet user's computer essentially into a server that someone in another country can use to browse the Internet away from the watchful eyes of their government. Psiphon allows anyone living in a country that allows unfettered access to the Internet to set up their account, and then enable someone in a more restrictive country to log on from that computer. The free download, which will be available starting Friday, offers encrypted and secure Internet surfing for users, which will prevent their government from tracing their Web surfing patterns. "The communities that we're helping to connect to each other have a legitimate right to exercise their human rights within this government regime," says Ron Deibert, director of the university's Citizen Lab, who also acknowledges that Psiphon might be unlawful in those countries. "It does conflict with some sovereign states' values, but there are competing legal norms at work."
Click Here to View Full Article
to the top


Super-computer Boss Has a To-Do List for a Better Future; Better Medical Care, Disaster Preparation Are Goals
Triangle Business Journal (11/27/06) Horlbeck, Fred

Dan Reed, the new director of the Renaissance Computing Institute in Chapel Hill (RENCI), has big plans to user supercomputing to bring about weather-response, medical, and economic change in North Carolina. RENCI, a joint venture between the UNC-Chapel Hill, Duke, North Carolina State, and the state of North Carolina, utilizes a state-or-the-art computer network, including the IBM Blue Gene/L supercomputer, the second fastest in North Carolina, and data information technology that allows high levels of collaboration. Reed sees great potential in the "intersection of disciplines," where people and organizations provide knowledge, abilities, and ideas to solve problems more efficiently. RENCI's power, with help from weather-related organizations, will bring about "high-resolution" weather forecasting, according to Reed, that will be able to track a storm's path as well as predict the exact location of flooding. Where medical advancements are concerned, RENCI will alert doctors immediately as to anomalies occurring in patients that have been given a special device to wear, hopefully allowing a crisis to be avoided. Reed predicts that in 10 years, an individual's genome sequence could be profiled, providing doctors with specific vulnerabilities to disease. To aid economic advances, RENCI will be able to take in corporate data and locate areas of growth in order to devise ways to create further growth. Reed says, "This is one of the first attempts to do this in the U.S. We're trying to bring people together from across the state. We're trying to be a catalyst for innovation."
Click Here to View Full Article
to the top


Smart Spaces: If These Walls Could Talk
Computerworld (11/27/06) Anthes, Gary

The concept of "smart spaces" has been around for quite some time, and while the technology required for the individual components exists today, interoperability, accuracy, and reliability prove to be stumbling blocks. Different types of sensors, large touch-screen displays, cameras, microphones, and other devices were incorporated into a prototypical "interactive room," or iRoom, by Stanford researchers, which utilizes the Interactive Room Operating System (IROS), a metaoperating system that they describe as having "taken the operating system idea to the space level, so people can coordinate their work in an environment with multiple devices," says Stanford computer science professor Terry Winograd. The goal in such a project, as Winograd explains, is to maximize seamlessness and transparency, because, "Whenever you have to stop focusing on what you care about to focus on how the machine is doing, you lose fluency." IBM Research senior manager for responsive enterprise solutions Stefan Hild, who worked on an IBM prototypical interactive office, explains that, "The investment of taking an office building and enabling it that way is fairly high. But you can get 80 percent benefit with 20 percent of the cost." While such an investment could pay off, technology needs to make some progress first. Hild recognizes that turning an office building into a completely interoperable and interactive, real-time environment would require drastically scaling up networks and processors.
Click Here to View Full Article
to the top


It's a Woman's World Wide Web
New Scientist (11/25/06) Vol. 192, No. 2579, P. 58; Biever, Celeste

Wendy Hall claims, "There is nothing traditional or geeky about me," and this is only the beginning of the way in which she shatters stereotypes in an IT world dominated by men: She is currently the head of the University of Southampton's world-class electronics and computer science department, a senior VP of the Royal Academy of Engineering, VP of the ACM, and sits on the Council for Science and Technology, which advises the prime minister; all this despite being told she didn't get her first job because of her sex, and having many of her later ideas ignored by male counterparts. After teaching herself to program in the 1980s, Hall was attracted to the ability that computers had to improve people's lives: "I could see what could be possible once the technology developed." She launched a program called Microcosm in 1989: A database of electronic photos, documents, and recordings that could be linked to each other in different ways, depending on the user. Links were created in real time while the document was read by the user by comparing the contents of a given document and related contents of the hard drive, so the links could be shown dynamically based on the user's browsing habits. However, Tim Berners-Lee's invention of the World Wide Web took off instead of Microcosm, because its links were embedded and it worked on a global network that could be accessed by anyone with an Internet connection, while Microcosm could be used only in standalone hard drives. Recently, Hall has been involved with the creation of the Web science research initiative, which will focus on the relationship between computer science and social science. She says merging the two disciplines could attract more women to computing, a cause she champions because computer science is a field she loves. Hall says, "All the wonderful things I am doing are because I am a computer scientist. IT and computing are the basis of everything."
Click Here to View Full Article
to the top


Learning Through Multimedia: Automatic Speech Recognition Enhancing Accessibility and Interaction
University of Southampton (ECS) (11/26/06) Wald, Mike

Researcher Mike Wald demonstrates the enhancement of learning and teaching quality via automatic speech recognition (ASR) to access, manage, and leverage online multimedia content. His presentation shows that ASR technology can help guarantee that both in-person learning and online learning is universally accessible via the cost-effective generation of synchronized and captioned multimedia. According to Wald, this strategy accommodates preferred learning/teaching approaches, and can help those who have problems taking notes because of cognitive, sensory, or physical difficulties. In addition, the approach can aid learners with the management and mining of online digital multimedia resources, as well as offer automatic speech captioning to hearing-impaired learners or any others to whom speech is unavailable, unsuitable, or inaudible. Users with blindness or other visual impairments can also benefit from the method, which helps them read and search learning material through the enhancement of synthetic speech with natural recorded real speech. Furthermore, teachers as well as learners can improve their spoken communication skills through reflection afforded by ASR. "Although it can be expected that developments in ASR will continue to improve accuracy rates, the use of a human intermediary to improve accuracy through correcting mistakes in real time as they are made by the ASR software could, where necessary, help compensate for some of ASR's current limitations," Wald writes. The projection of text onto a large screen has had some success in classroom situations, but many circumstances call for the provision of an individual personalized and customizable display. Wald concludes that the ideal system for digitally recording and replaying multimedia content would automatically produce a mistake-proof transcript of spoken language that is synchronized with audio, video, and any graphical elements, which would be displayed in the most suitable manner on diverse instruments and with adjustable replay speed; annotation would be provided via pen or keyboard and mouse, and have synchronicity with the multimedia content.
Click Here to View Full Article
to the top


The Ultimate White Light
Scientific American (12/06) Vol. 295, No. 6, P. 86; Alfano, Robert R.

Optical data transmission could achieve unprecedented speeds with the advent of "supercontinuum" (SC) laser light, which melds useful properties of laser light with the broad bandwidth spectrum of white light, writes City College of the City University of New York professor Robert Alfano. Alfano pioneered SC light with Stanley Shapiro at General Telephone and Electronics Laboratories (since renamed Verizon) in 1969. SC light is primarily generated today by transmitting high-intensity pulses of laser light through specially designed microstructure fibers. The light and the fiber material interact through a series of nonlinear optical processes that extend the light's bandwidth. One such process is self-phase modulation. SC light can be applied to provide extremely accurate frequency measurements and clocks, detection of airborne chemicals such as pollutants and aerosols, and high-resolution medical imaging via optical coherence tomography. High throughput telecommunications with data transmission rates that beat current systems by a factor of 1,000 is another application of SC light, one with more immediate commercial ramifications.
Click Here to View Full Article - Web Link to Publication Homepage
to the top


The Future of Simulation: A Field of Dreams?
Computer (11/06) Vol. 39, No. 11, P. 22; Yi, Joshua J.; Eeckhout, Lieven; Lilja, David J.

There is a growing reliance on simulators among computer architecture researchers because simulation can strike a balance between cost, flexibility, and timeliness, and the diversity of benchmarks, methodology, and data sets raises numerous questions about whether simulators suitably model processor or system behavior, what the essential elements of future simulators are, ways to design benchmark suites with representative benchmarks without excessive redundancy, etc., write Freescale Semiconductor's Joshua Yi et al. Addressing these questions was the impetus behind a panel discussion on simulation infrastructure, benchmarks, and simulation methodology at the International Symposium on Performance Analysis of Systems and Software in March 2004. The efficient traversal and characterization of the design space is problematic, and among the alternatives the authors recommend are analytical models, statistical simulation, and specialized trace-driven simulation. Analytical modeling and statistical simulation offer faster speed but less accuracy than cycle-accurate simulation, but the speed advantage is more critical because relative accuracy is usually enough to track substantial shifts in processor performance, while deployment time is fractionally shorter. A major benchmarking problem is the lack of certainty in the representativeness of average benchmark suites, and solving this problem entails the computer architecture community's introduction of additional benchmark characterization and classification techniques, specifically those that offer greater accuracy or efficiency than current approaches. Benchmark length and the simulation time this translates into is another problem, and the use of sampling-based techniques such as SMARTS and SimPoint is recommended by the authors as a mitigation strategy. Notable problems with current simulation methodology include ad hoc simulation, which can be addressed via comprehensive documentation and justification of the methodology and the addition of more statistical rigor. Solving reproducibility and comparability problems stemming from widely variable simulation workloads involves agreement among the research community on well-balanced processor and memory hierarchy configurations, common benchmark subsets, and common data sets.
Click Here to View Full Article - Web Link to Publication Homepage
to the top


To submit feedback about ACM TechNews, contact: technews@hq.acm.org

To unsubscribe from the ACM TechNews Early Alert Service: Please send a separate email to listserv@listserv.acm.org with the line

signoff technews

in the body of your message.

Please note that replying directly to this message does not automatically unsubscribe you from the TechNews list.

ACM may have a different email address on file for you, so if you're unable to "unsubscribe" yourself, please direct your request to: technews-request@ acm.org

We will remove your name from the TechNews list on your behalf.

For help with technical problems, including problems with leaving the list, please write to: technews-request@acm.org

to the top

News Abstracts © 2006 Information, Inc.


© 2006 ACM, Inc. All rights reserved. ACM Privacy Policy.

About ACM | Contact us | Boards & Committees | Press Room | Membership | Privacy Policy | Code of Ethics | System Availability | Copyright © 2014, ACM, Inc.