Visualizzazione post con etichetta Distributed Computing. Mostra tutti i post
Visualizzazione post con etichetta Distributed Computing. Mostra tutti i post
domenica 24 maggio 2009
venerdì 19 ottobre 2007
Computers With 'Common Sense'
Source:
ScienceDaily (Oct. 18, 2007) — Using a little-known Google Labs widget, computer scientists from UC San Diego and UCLA have brought common sense to an automated image labeling system. This common sense is the ability to use context to help identify objects in photographs.
For example, if a conventional automated object identifier has labeled a person, a tennis racket, a tennis court and a lemon in a photo, the new post-processing context check will re-label the lemon as a tennis ball.
“We think our paper is the first to bring external semantic context to the problem of object recognition,” said computer science professor Serge Belongie from UC San Diego.
The researchers show that the Google Labs tool called Google Sets can be used to provide external contextual information to automated object identifiers.
Google Sets generates lists of related items or objects from just a few examples. If you type in John, Paul and George, it will return the words Ringo, Beatles and John Lennon. If you type “neon” and “argon” it will give you the rest of the noble gasses.
“In some ways, Google Sets is a proxy for common sense. In our paper, we showed that you can use this common sense to provide contextual information that improves the accuracy of automated image labeling systems,” said Belongie.
The image labeling system is a three step process. First, an automated system splits the image up into different regions through the process of image segmentation. In the photo above, image segmentation separates the person, the court, the racket and the yellow sphere.
Next, an automated system provides a ranked list of probable labels for each of these image regions.
Finally, the system adds a dose of context by processing all the different possible combinations of labels within the image and maximizing the contextual agreement among the labeled objects within each picture.
It is during this step that Google Sets can be used as a source of context that helps the system turn a lemon into a tennis ball. In this case, these “semantic context constraints” helped the system disambiguate between visually similar objects.
In another example, the researchers show that an object originally labeled as a cow is (correctly) re-labeled as a boat when the other objects in the image – sky, tree, building and water – are considered during the post-processing context step. In this case, the semantic context constraints helped to correct an entirely wrong image label. The context information came from the co-occurrence of object labels in the training sets rather than from Google Sets.
The computer scientists also highlight other advances they bring to automated object identification. First, instead of doing just one image segmentation, the researchers generated a collection of image segmentations and put together a shortlist of stable image segmentations. This increases the accuracy of the segmentation process and provides an implicit shape description for each of the image regions.
Second, the researchers ran their object categorization model on each of the segmentations, rather than on individual pixels. This dramatically reduced the computational demands on the object categorization model.
In the two sets of images that the researchers tested, the categorization results improved considerably with inclusion of context. For one image dataset, the average categorization accuracy increased more than 10 percent using the semantic context provided by Google Sets. In a second dataset, the average categorization accuracy improved by about 2 percent using the semantic context provided by Google Sets. The improvements were higher when the researchers gleaned context information from data on co-occurrence of object labels in the training data set for the object identifier.
Right now, the researchers are exploring ways to extend context beyond the presence of objects in the same image. For example, they want to make explicit use of absolute and relative geometric relationships between objects in an image – such as “above” or “inside” relationships. This would mean that if a person were sitting on top of an animal, the system would consider the animal to be more likely a horse than a dog.
Reference: “Objects in Context,” by Andrew Rabinovich, Carolina Galleguillos, Eric Wiewiora and Serge Belongie from the Department of Computer Science and Engineering at the UCSD Jacobs School of Engineering. Andrea Vedaldi from the Department of Computer Science, UCLA.
The paper will be presented on Thursday 18 October 2007 at ICCV 2007 – the 11th IEEE International Conference on Computer Vision in Rio de Janeiro, Brazil.
Funders: National Science Foundation, Afred P. Sloan Research Fellowship, Air Force Office of Scientific Research, Office of Naval Research.
Adapted from materials provided by University of California - San Diego.
“We think our paper is the first to bring external semantic context to the problem of object recognition,” said computer science professor Serge Belongie from UC San Diego.
The researchers show that the Google Labs tool called Google Sets can be used to provide external contextual information to automated object identifiers.
Google Sets generates lists of related items or objects from just a few examples. If you type in John, Paul and George, it will return the words Ringo, Beatles and John Lennon. If you type “neon” and “argon” it will give you the rest of the noble gasses.
“In some ways, Google Sets is a proxy for common sense. In our paper, we showed that you can use this common sense to provide contextual information that improves the accuracy of automated image labeling systems,” said Belongie.
The image labeling system is a three step process. First, an automated system splits the image up into different regions through the process of image segmentation. In the photo above, image segmentation separates the person, the court, the racket and the yellow sphere.
Next, an automated system provides a ranked list of probable labels for each of these image regions.
Finally, the system adds a dose of context by processing all the different possible combinations of labels within the image and maximizing the contextual agreement among the labeled objects within each picture.
It is during this step that Google Sets can be used as a source of context that helps the system turn a lemon into a tennis ball. In this case, these “semantic context constraints” helped the system disambiguate between visually similar objects.
In another example, the researchers show that an object originally labeled as a cow is (correctly) re-labeled as a boat when the other objects in the image – sky, tree, building and water – are considered during the post-processing context step. In this case, the semantic context constraints helped to correct an entirely wrong image label. The context information came from the co-occurrence of object labels in the training sets rather than from Google Sets.
The computer scientists also highlight other advances they bring to automated object identification. First, instead of doing just one image segmentation, the researchers generated a collection of image segmentations and put together a shortlist of stable image segmentations. This increases the accuracy of the segmentation process and provides an implicit shape description for each of the image regions.
Second, the researchers ran their object categorization model on each of the segmentations, rather than on individual pixels. This dramatically reduced the computational demands on the object categorization model.
In the two sets of images that the researchers tested, the categorization results improved considerably with inclusion of context. For one image dataset, the average categorization accuracy increased more than 10 percent using the semantic context provided by Google Sets. In a second dataset, the average categorization accuracy improved by about 2 percent using the semantic context provided by Google Sets. The improvements were higher when the researchers gleaned context information from data on co-occurrence of object labels in the training data set for the object identifier.
Right now, the researchers are exploring ways to extend context beyond the presence of objects in the same image. For example, they want to make explicit use of absolute and relative geometric relationships between objects in an image – such as “above” or “inside” relationships. This would mean that if a person were sitting on top of an animal, the system would consider the animal to be more likely a horse than a dog.
Reference: “Objects in Context,” by Andrew Rabinovich, Carolina Galleguillos, Eric Wiewiora and Serge Belongie from the Department of Computer Science and Engineering at the UCSD Jacobs School of Engineering. Andrea Vedaldi from the Department of Computer Science, UCLA.
The paper will be presented on Thursday 18 October 2007 at ICCV 2007 – the 11th IEEE International Conference on Computer Vision in Rio de Janeiro, Brazil.
Funders: National Science Foundation, Afred P. Sloan Research Fellowship, Air Force Office of Scientific Research, Office of Naval Research.
Adapted from materials provided by University of California - San Diego.
Fausto Intilla
martedì 16 ottobre 2007
Computer Security Can Double As Help For The Blind
Source:
Science Daily — Before you can post a comment to most blogs, you have to type in a series of distorted letters and numbers (a CAPTCHA) to prove that you are a person and not a computer attempting to add comment spam to the blog.
What if -- instead of wasting your time and energy typing something meaningless like SGO9DXG -- you could label an image or perform some other quick task that will help someone who is visually impaired do their grocery shopping?
In a position paper presented at Interactive Computer Vision (ICV) 2007 on October 15 in Rio de Janeiro, computer scientists from UC San Diego led by professor Serge Belongie outline a grid system that would allow CAPTCHAs to be used for this purpose -- and an endless number of other good causes.
"One of the application areas for my research is assistive technologyfor the blind. For example, there is an enormous amount of data that needs to be labeled for our grocery shopping aid to work. We are developing a wearable computer with a camera that can lead a visually impaired user to a desired product in a grocery store by analyzing the video stream. Our paper describes a way that people who are looking to prove that they are humans and not computers can help label still shots from video streams in real time," said Belongie.
The researchers call their system a "Soylent grid" which is a reference to the 1973 film Soylent Green (see more on this reference at the end of the article).
"The degree to which human beings could participate in the system (as remote sighted guides) ranges from none at all to virtually unlimited. If no human user is involved in the loop, only computer vision algorithms solve the identification problem. But in principle, if there were an unlimited number of humans in the loop, all the video frames could be submitted to a SOYLENT GRID, be solved immediately and sent back to the device to guide the user," the authors write in their paper.
From the front end, users who want to post a comment on a blog would be asked to perform a variety of tasks, instead of typing in a string of misshapen letters and numbers.
"You might be asked to click on the peanut butter jar or click the Cheetos bag in an image," said Belongie. "This would be one of the so called 'Where's Waldo' object detection tasks."
The task list also includes "Name that Thing" (object recognition), "Trace This" (image segmentation) and "Hot or Not" (choosing visually pleasing images).
"Our research on the personal shopper for the visually impaired -- called Grozi -- is a big motivation for this project. When we started the Grozi project, one of the students, Michele Merler -- who is now working on a Ph.D. at Columbia University -- captured 45 minutes of video footage from the campus grocery store and then endured weeks of manually intensive labor, drawing bounding boxes and identifying the 120 products we focused on. This is work the soylent grid could do," said Belongie.
From the back end, researchers and others who need images labeled would interact with clients (like a blog hosting company) that need to take advantage of the CAPTCHA and spam filtering capabilities of the grid.
"Getting this done is going to take an innovative collaboration between academia and industry. Calit2 could be uniquely instrumental in this project," said Belongie. "Right now we are working on a proposal that will outline exactly what we need -- access to X number of CAPTCHA requests in one week, for example. With this, we'll do a case study and demonstrate just how much data can be labeled with 99 percent reliability through the soylent grid. I'm hoping for people to say, 'Wow, I didn't know that kind of computation was available.'"
This work incorporates recent work from a variety of researchers, including computer scientist Luis von Ahn from Carnegie Mellon University. His reCAPTCHA project uses CAPTCHAs to digitize books.
Soylent Grid?
The researchers call their system a "Soylent grid" and titled their paper "Soylent Grid: it's Made of People! Both the grid name and paper name are references to the 1973 cult classic film Soylent Green, a dystopian science fiction film set in an overpopulated world in which the masses are reduced to eating different varieties of "soylent" -- a synthetic food that suggests both soybeans and lentils. The line from the movie that inspired the title of this paper comes is delivered when someone discovers that soylent green is actually made of cadavers from a government sponsored euthanasia program -- prompting the phrase "Soylent green, it's made of people!" The computer scientists are playing off this famous phrase with their title: "Soylent Grid: it's Made of People!" The idea being that people from all over the world need to jump through anti-spam hoops such as CAPTCHAs, and the power of these people can be harnessed through a grid structure to do some good in the world.
Article: "Soylent Grid: it's Made of People!" by Stephan Steinbach, Vincent Rabaud and Serge Belongie
Note: This story has been adapted from material provided by University of California - San Diego.
In a position paper presented at Interactive Computer Vision (ICV) 2007 on October 15 in Rio de Janeiro, computer scientists from UC San Diego led by professor Serge Belongie outline a grid system that would allow CAPTCHAs to be used for this purpose -- and an endless number of other good causes.
"One of the application areas for my research is assistive technologyfor the blind. For example, there is an enormous amount of data that needs to be labeled for our grocery shopping aid to work. We are developing a wearable computer with a camera that can lead a visually impaired user to a desired product in a grocery store by analyzing the video stream. Our paper describes a way that people who are looking to prove that they are humans and not computers can help label still shots from video streams in real time," said Belongie.
The researchers call their system a "Soylent grid" which is a reference to the 1973 film Soylent Green (see more on this reference at the end of the article).
"The degree to which human beings could participate in the system (as remote sighted guides) ranges from none at all to virtually unlimited. If no human user is involved in the loop, only computer vision algorithms solve the identification problem. But in principle, if there were an unlimited number of humans in the loop, all the video frames could be submitted to a SOYLENT GRID, be solved immediately and sent back to the device to guide the user," the authors write in their paper.
From the front end, users who want to post a comment on a blog would be asked to perform a variety of tasks, instead of typing in a string of misshapen letters and numbers.
"You might be asked to click on the peanut butter jar or click the Cheetos bag in an image," said Belongie. "This would be one of the so called 'Where's Waldo' object detection tasks."
The task list also includes "Name that Thing" (object recognition), "Trace This" (image segmentation) and "Hot or Not" (choosing visually pleasing images).
"Our research on the personal shopper for the visually impaired -- called Grozi -- is a big motivation for this project. When we started the Grozi project, one of the students, Michele Merler -- who is now working on a Ph.D. at Columbia University -- captured 45 minutes of video footage from the campus grocery store and then endured weeks of manually intensive labor, drawing bounding boxes and identifying the 120 products we focused on. This is work the soylent grid could do," said Belongie.
From the back end, researchers and others who need images labeled would interact with clients (like a blog hosting company) that need to take advantage of the CAPTCHA and spam filtering capabilities of the grid.
"Getting this done is going to take an innovative collaboration between academia and industry. Calit2 could be uniquely instrumental in this project," said Belongie. "Right now we are working on a proposal that will outline exactly what we need -- access to X number of CAPTCHA requests in one week, for example. With this, we'll do a case study and demonstrate just how much data can be labeled with 99 percent reliability through the soylent grid. I'm hoping for people to say, 'Wow, I didn't know that kind of computation was available.'"
This work incorporates recent work from a variety of researchers, including computer scientist Luis von Ahn from Carnegie Mellon University. His reCAPTCHA project uses CAPTCHAs to digitize books.
Soylent Grid?
The researchers call their system a "Soylent grid" and titled their paper "Soylent Grid: it's Made of People! Both the grid name and paper name are references to the 1973 cult classic film Soylent Green, a dystopian science fiction film set in an overpopulated world in which the masses are reduced to eating different varieties of "soylent" -- a synthetic food that suggests both soybeans and lentils. The line from the movie that inspired the title of this paper comes is delivered when someone discovers that soylent green is actually made of cadavers from a government sponsored euthanasia program -- prompting the phrase "Soylent green, it's made of people!" The computer scientists are playing off this famous phrase with their title: "Soylent Grid: it's Made of People!" The idea being that people from all over the world need to jump through anti-spam hoops such as CAPTCHAs, and the power of these people can be harnessed through a grid structure to do some good in the world.
Article: "Soylent Grid: it's Made of People!" by Stephan Steinbach, Vincent Rabaud and Serge Belongie
Note: This story has been adapted from material provided by University of California - San Diego.
Fausto Intilla
martedì 9 ottobre 2007
Quantum Computing Possibilites Enhanced With New Material
Source:
Science Daily — Scientists at Florida State University's National High Magnetic Field Laboratory and the university's Department of Chemistry and Biochemistry have introduced a new material that could be to computers of the future what silicon is to the computers of today.
The material -- a compound made from the elements potassium, niobium and oxygen, along with chromium ions -- could provide a technological breakthrough that leads to the development of new quantum computing technologies. Quantum computers would harness the power of atoms and molecules to perform memory and processing tasks on a scale far beyond those of current computers.
"The field of quantum information technology is in its infancy, and our work is another step forward in this fascinating field," said Saritha Nellutla, a postdoctoral associate at the magnet lab and lead author of the paper published in Physical Review Letters.
Semiconductor technology is close to reaching its performance limit. Over the years, processors have shrunk to their current size, with the components of a computer chip more than 1,000 times smaller than the thickness of a human hair. At those very small scales, quantum effects -- behaviors in matter that occur at the atomic and subatomic levels -- can start playing a role. By exploiting those behaviors, scientists hope to take computing to the next level.
In current computers, the basic unit of information is the "bit," which can have a value of 0 or 1. In so-called quantum computers, which currently exist only in theory, the basic unit is the "qubit" (short for quantum bit). A qubit can have not only a value of 0 or 1, but also all kinds of combinations of 0 and 1 -- including 0 and 1 at the same time -- meaning quantum computers could perform certain kinds of calculations much more effectively than current ones.
How scientists realize the promise of the theoretical qubit is not clear. Various designs and paths have been proposed, and one very promising idea is to use tiny magnetic fields, called "spins." Spins are associated with electrons and various atomic nuclei.
Magnet lab scientists used high magnetic fields and microwave radiation to "operate" on the spins in the new material they developed to get an indication of how long the spin could be controlled. Based on their experiments, the material could enable 500 operations in 10 microseconds before losing its ability to retain information, making it a good candidate for a qubit.
Putting this spin to work would usher in a technological revolution, because the spin state of an electron, in addition to its charge, could be used to carry, manipulate and store information.
"This material is very promising," said Naresh Dalal, a professor of chemistry and biochemistry at FSU and one of the paper's authors. "But additional synthetic and magnetic characterization work is needed before it could be made suitable for use in a device."
Dalal also serves as an adviser to FSU chemistry graduate student Mekhala Pati, who created the material.
Note: This story has been adapted from material provided by Florida State University.
"The field of quantum information technology is in its infancy, and our work is another step forward in this fascinating field," said Saritha Nellutla, a postdoctoral associate at the magnet lab and lead author of the paper published in Physical Review Letters.
Semiconductor technology is close to reaching its performance limit. Over the years, processors have shrunk to their current size, with the components of a computer chip more than 1,000 times smaller than the thickness of a human hair. At those very small scales, quantum effects -- behaviors in matter that occur at the atomic and subatomic levels -- can start playing a role. By exploiting those behaviors, scientists hope to take computing to the next level.
In current computers, the basic unit of information is the "bit," which can have a value of 0 or 1. In so-called quantum computers, which currently exist only in theory, the basic unit is the "qubit" (short for quantum bit). A qubit can have not only a value of 0 or 1, but also all kinds of combinations of 0 and 1 -- including 0 and 1 at the same time -- meaning quantum computers could perform certain kinds of calculations much more effectively than current ones.
How scientists realize the promise of the theoretical qubit is not clear. Various designs and paths have been proposed, and one very promising idea is to use tiny magnetic fields, called "spins." Spins are associated with electrons and various atomic nuclei.
Magnet lab scientists used high magnetic fields and microwave radiation to "operate" on the spins in the new material they developed to get an indication of how long the spin could be controlled. Based on their experiments, the material could enable 500 operations in 10 microseconds before losing its ability to retain information, making it a good candidate for a qubit.
Putting this spin to work would usher in a technological revolution, because the spin state of an electron, in addition to its charge, could be used to carry, manipulate and store information.
"This material is very promising," said Naresh Dalal, a professor of chemistry and biochemistry at FSU and one of the paper's authors. "But additional synthetic and magnetic characterization work is needed before it could be made suitable for use in a device."
Dalal also serves as an adviser to FSU chemistry graduate student Mekhala Pati, who created the material.
Note: This story has been adapted from material provided by Florida State University.
Fausto Intilla
mercoledì 3 ottobre 2007
Technology Could Enable Computers To 'Read The Minds' Of Users
Source:
"New evaluation techniques that monitor user experiences while working with computers are increasingly necessary," said Robert Jacob, computer science professor and researcher. "One moment a user may be bored, and the next moment, the same user may be overwhelmed. Measuring mental workload, frustration and distraction is typically limited to qualitatively observing computer users or to administering surveys after completion of a task, potentially missing valuable insight into the users' changing experiences."
Sergio Fantini, biomedical engineering professor, in conjunction with Jacob's human-computer interaction (HCI) group, is studying functional near-infrared spectroscopy (fNIRS) technology that uses light to monitor brain blood flow as a proxy for workload stress a user may experience when performing an increasingly difficult task. A $445,000 grant from the National Science Foundation will allow the interdisciplinary team to incorporate real-time biomedical data with machine learning to produce a more in-tune computer user experience.
Lighting up the brain
"fNIRS is an emerging non-invasive, lightweight imaging tool which can measure blood oxygenation levels in the brain," said Fantini, also an associate dean for graduate education at Tufts' School of Engineering.
The fNIRS device, which looks like a futuristic headband, uses laser diodes to send near-infrared light through the forehead at a relatively shallow depth--only two to three centimeters--to interact with the brain's frontal lobe. Light usually passes through the body's tissues, except when it encounters oxygenated or deoxygenated hemoglobin in the blood. Light waves are absorbed by the active, blood-filled areas of the brain and any remaining light is diffusely reflected to the fNIRS detectors.
"fNIRS, like MRI, uses the idea that blood flow changes to compensate for the increased metabolic demands of the area of the brain that's being used," said Erin Solovey, a graduate researcher at the School of Engineering.
"We don't know how specific we can be about identifying users' different emotional states," said Fantini. "However, the particular area of the brain where the blood flow change occurs should provide indications of the brain metabolic changes and by extension workload, which could be a proxy for emotions like frustration."
In the initial experiments, Jacob and Fantini's groups determined how accurately fNIRS could register users' workload. While wearing the fNIRS device, test subjects viewed a multicolored cube consisting of eight smaller cubes with two, three or four different colors. As the cube rotated onscreen, subjects counted the number of colored squares in a series of 30 tasks. The fNIRS device and subsequent user surveys reflected greater difficulty as users kept track of increasing numbers of colors. The fNIRS data agreed with user surveys up to 83 percent of the time.
The Tufts group will present its initial results on using fNIRS to detect the user workload experience at the Association for Computing Machinery (ACM) symposium on user interface software and technology, to be held Oct. 7 through 10 in Newport, R.I.
"It seems that we can predict, with relatively high confidence, whether the subject was experiencing no workload, low workload, or high workload," said Leanne Hirshfield, a graduate researcher and lead author on the poster paper to be presented at the ACM symposium.
Note: This story has been adapted from material provided by Tufts University.
Fausto Intilla
www.oloscience.com
giovedì 27 settembre 2007
'Printers' That Can Make 3-D Solid Objects Soon To Enter Mainstream
Source:
Such technology already exists and is maturing rapidly so that high-tech designers and others can share solid designs almost as quickly as sending a fax. The systems available are based on bath of liquid plastic which is solidified by laser light. The movements of the laser are controlled by a computer that reads a digitized 3D map of the solid object or design.
Writing in the Inderscience publication International Journal of Technology Marketing, US researchers discuss how this technology might eventually move into the mainstream allowing work environments to 3-D print equipment, whether that is plastic paperclips, teacups, or components that can be joined to make sophisticated devices, perhaps bolted together with printed nuts and bolts.
Physicist Phil Anderson of the School of Theoretical and Applied Science working with Cherie Ann Sherman of the Anisfield School of Business, both at Ramapo College of New Jersey, in Mahwah, New Jersey, explain how this technology, which is known formally as 'rapid prototyping' could revolutionize the way people buy goods.
It will allow them to buy or obtain a digital file representing a physical product electronically and then produce the object at a time and place convenient to them. The technology will be revolutionary in the same way that music downloads have shaken up the music industry. "This technology has the potential to generate a variety of new business models, which would enhance the average consumer's lifestyle," say the paper's authors.
The team discusses the current advanced applications of rapid prototyping which exist in the military where missing and damaged components can be produced at the site of action. Education too can make use of 3-D printing to allow students to make solid their experimental designs.
Also, product developers can share tangible prototypes by transferring the digitized design without the delay of shipping a solid object between sites, which may be separated by thousands of miles. The possibilities for consumer goods, individualized custom products, replacement components, and quick fixes for broken objects, are almost unlimited, the authors suggest.
From the business perspective, e-commerce sites will essentially become digital download sites with physical stores, retail employees, and shipping eliminated. It is only a matter of time before the 'killer application,' the 3-D equivalent of the mp3 music file, one might say, arrives to make owning a 3-D printer as necessary to the modern lifestyle as owning a microwave oven, a TV, or indeed a personal computer.
Note: This story has been adapted from a news release issued by Inderscience Publishers.
Fausto Intilla
www.oloscience.com
venerdì 7 settembre 2007
Computer Scientists Take The 'Why' Out Of WiFi
Source:
Science Daily — “People expect WiFi to work, but there is also a general understanding that it’s just kind of flakey,” said Stefan Savage, one of the UCSD computer science professors who led development of an automated, enterprise-scale WiFi troubleshooting system for UCSD’s computer science building. The system is described in a paper presented in August in Kyoto, Japan at ACM SIGCOMM, one of the world’s premier networking conferences.
“If you have a wireless problem in our building, our system automatically analyzes the behavior of your connection – each wireless protocol, each wired network service and the many interactions between them. In the end, we can say ‘it’s because of this that your wireless is slow or has stopped working’ – and we can tell you immediately,” said Savage.
For humans, diagnosing problems in the now ubiquitous 802.11-based wireless access networks requires a huge amount of data, expertise and time. In addition to the myriad complexities of the wired network, wireless networks face the additional challenges of shared spectrum, user mobility and authentication management. Finally, the interaction between wired and wireless networks is itself a source of many problems.
“Wireless networks are hooked on to the wired part of the Internet with a bunch of ‘Scotch tape and bailing wire’ – protocols that really weren’t designed for WiFi,” explained Savage. “If one of these components has a glitch, you may not be able to use the Internet even though the network itself is working fine.”
There are so many moving pieces, so many things you can not see. Within this soup, everything has to work just right. When it doesn’t, trying to identify which piece wasn’t working is tough and requires sifting through a lot of data. For example, someone using the microwave oven two rooms away may cause enough interference to disrupt your connection.
“Today, if you ask your network administrator why it takes minutes to connect to the network or why your WiFi connection is slow, they’re unlikely to know the answer,” explained Yu-Chung Cheng, a computer science Ph.D. student at UCSD and lead author on the paper. “Many problems are transient – they’re gone before you can even get an admin to look at them – and the number of possible reasons is huge,” explained Cheng, who recently defended his dissertation and will join Google this fall.
“Few organizations have the expertise, data or tools to decompose the underlying problems and interactions responsible for transient outages or performance degradations,” the authors write in their SIGCOMM paper.
The computer scientists from UCSD’s Jacobs School of Engineering presented a set of modeling techniques for automatically characterizing the source of such problems. In particular, they focus on data transfer delays unique to 802.11 networks – media access dynamics and mobility management latency.
The UCSD system runs 24 hours a day, constantly churning through the flood of data relevant to the wireless network and catching transient problems.
“We’ve created a virtual wireless expert who is always at work,” said Cheng.
Within the UCSD Computer Science building, all the wireless help-desk issues go through the new automated system, which has been running for about 9 months. The data collection has been going on for almost 2 years.
One of the big take-away lessons is that there is no one thing that affects wireless network performance. Instead, there are a lot of little things that interact and go wrong in ways you might not anticipate.
“I look at this as an engineering effort. In the future, I think that enterprise wireless networks will have sophisticated diagnostics and repair capabilities built in. How much these will draw from our work is hard to tell today. You never know the impact you are going to have when you do the work,” said Savage. “In the meantime, our system is the ultimate laboratory for testing new wireless gadgets and new approaches to building wireless systems. We just started looking at WiFi-based Voice-Over-IP (VOIP) phones. We learn something new every week.”
Paper citation: "Automating Cross-Layer Diagnosis of Enterprise Wireless Networks," by Yu-Chung Cheng, Mikhail Afanasyev, Patrick Verkaik, Jennifer Chiang, Alex C. Snoeren, Stefan Savage, and Geoffrey M. Voelker from the Department of Computer Science and Engineering at UCSD's Jacobs School of Engineering; Péter Benkö from the Traffic Analysis and Network Performance Laboratory (TrafficLab) at Ericsson Research, Budapest, Hungary
Funding was provided by UCSD Center for Networked Systems (CNS), Ericsson, National Science Foundation (NSF), and a UC Discovery Grant.
Note: This story has been adapted from a news release issued by University of California - San Diego.
Fausto Intilla
www.oloscience.com
martedì 4 settembre 2007
Internet Map Looks Like A Digital Dandelion
Source:
Science Daily — What looks like the head of a digital dandelion is a map of the Internet generated by new algorithms from computer scientists at UC San Diego. This map features Internet nodes – the red dots – and linkages – the green lines.
But it is no ordinary map. It is a (mostly) randomly generated graph that retains the essential characteristics of a specific corner of the Internet but doubles the number of nodes.
On August 30 in Kyoto, Japan at ACM SIGCOMM, the premier computer networking conference, UCSD computer scientists presented techniques for producing annotated, Internet router graphs of different sizes – based on observations of Internet characteristics.
The graph annotations include information about the relevant peer-to-peer business relationships that help to determine the paths that packets of information take as they travel across the Internet. Generating these kinds of graphs is critical for a wide range of computer science research.
“Defending against denial of service attacks and large-scale worm outbreaks depends on network topology. Our work allows computer scientists to experiment with a range of random graphs that match Internet characteristics. This work is also useful for determining the sensitivity of particular techniques – like routing protocols and congestion controls – to network topology and to variations in network topology,” said Priya Mahadevan, the first author on the SIGCOMM 2007 paper. Mahadevan just completed her computer science Ph.D. at UCSD’s Jacobs School of Engineering. In October, she will join Hewlett Packard Laboratories in Palo Alto, CA.
“We’re saying, ‘here is what the Internet looks like, and here is our recreation of it on a larger scale.’ Our algorithm produces random graphs that maintain the important interconnectivity characteristics of the original. The goal is to produce a topology generator capable of outputting a range of annotated Internet topologies of varying sizes based on available measurements of network connectivity and characteristics,” said Amin Vahdat, the senior author on the paper, a computer science professor at UCSD and the Director of UCSD’s Center for Networked Systems (CNS) – an industrial/academic collaboration investigating emerging issues in computing systems that are both very large (planetary scale) and very small (the scale of wireless sensor networks).
The authors are making the source code for their topology generator publicly available and hope that it will benefit a range of studies.
“The techniques we have developed for characterizing and recreating Internet characteristics are generally applicable to a broad range of disciplines that consider networks, including physics, biology, chemistry, neuroscience and sociology,” said Vahdat.
Citation: Orbis: Rescaling Degree Correlations to Generate Annotated Internet Topologies, Priya Mahadevan, Calvin Hubble, Bradley Huffaker, Dimitri Krioukov, and Amin Vahdat, Proceedings of the ACM SIGCOMM Conference, Kyoto, Japan, August 2007.
Funding was provided by the National Science Foundation (NSF) and UCSD’s Center for Networked Systems (CNS)
Note: This story has been adapted from a news release issued by University of California - San Diego.
On August 30 in Kyoto, Japan at ACM SIGCOMM, the premier computer networking conference, UCSD computer scientists presented techniques for producing annotated, Internet router graphs of different sizes – based on observations of Internet characteristics.
The graph annotations include information about the relevant peer-to-peer business relationships that help to determine the paths that packets of information take as they travel across the Internet. Generating these kinds of graphs is critical for a wide range of computer science research.
“Defending against denial of service attacks and large-scale worm outbreaks depends on network topology. Our work allows computer scientists to experiment with a range of random graphs that match Internet characteristics. This work is also useful for determining the sensitivity of particular techniques – like routing protocols and congestion controls – to network topology and to variations in network topology,” said Priya Mahadevan, the first author on the SIGCOMM 2007 paper. Mahadevan just completed her computer science Ph.D. at UCSD’s Jacobs School of Engineering. In October, she will join Hewlett Packard Laboratories in Palo Alto, CA.
“We’re saying, ‘here is what the Internet looks like, and here is our recreation of it on a larger scale.’ Our algorithm produces random graphs that maintain the important interconnectivity characteristics of the original. The goal is to produce a topology generator capable of outputting a range of annotated Internet topologies of varying sizes based on available measurements of network connectivity and characteristics,” said Amin Vahdat, the senior author on the paper, a computer science professor at UCSD and the Director of UCSD’s Center for Networked Systems (CNS) – an industrial/academic collaboration investigating emerging issues in computing systems that are both very large (planetary scale) and very small (the scale of wireless sensor networks).
The authors are making the source code for their topology generator publicly available and hope that it will benefit a range of studies.
“The techniques we have developed for characterizing and recreating Internet characteristics are generally applicable to a broad range of disciplines that consider networks, including physics, biology, chemistry, neuroscience and sociology,” said Vahdat.
Citation: Orbis: Rescaling Degree Correlations to Generate Annotated Internet Topologies, Priya Mahadevan, Calvin Hubble, Bradley Huffaker, Dimitri Krioukov, and Amin Vahdat, Proceedings of the ACM SIGCOMM Conference, Kyoto, Japan, August 2007.
Funding was provided by the National Science Foundation (NSF) and UCSD’s Center for Networked Systems (CNS)
Note: This story has been adapted from a news release issued by University of California - San Diego.
Fausto Intilla
domenica 2 settembre 2007
I.B.M. Researchers Advancing Computer Processing Ability
By JOHN MARKOFF
Published: August 31, 2007
Published: August 31, 2007
SAN FRANCISCO, Aug. 30 — Researchers at I.B.M. laboratories say they have made progress toward storing information and computing at the level of individual atoms.
The scientists documented their work in two papers appearing on Friday in the journal Science. Both papers are focused on new understanding of the behavior of magnetism at the tiny scale of nanotechnology, where scientists hope to develop electronics made from components that are far smaller than today’s transistors and wires.
In one paper the researchers describe a technique for reading and writing digital ones and zeroes onto a handful of atoms, or even individual atoms. The second paper describes the ability to use a single molecule as a switch, replicating the behavior of today’s transistors.
The papers are the latest indication that computing technology is beginning to emerge that could replace today’s microelectronics materials in the next decade.
R. Stanley Williams, a Hewlett-Packard physicist, said this week that his group had begun manufacturing prototypes of a silicon chip that combines both conventional microelectronics and molecular scale components. Their first hybrid device is a circuit called a field programmable gate array, or F.P.G.A., using molecular-scale components as the configuration circuitry, an approach that will save tremendous space in the chip design.
A team of I.B.M. researchers at the company’s Almaden Research Center in San Jose, Calif., were able to use a scanning tunneling microscope to observe the magnetic orientation of iron and manganese atoms at low temperatures. Controlling magnetic direction is a crucial technique that is used in reading and writing digital information on magnetic storage disks like standard hard drives.
In addition to the potential storage applications, the researchers noted that atomic-scale magnetic structures are also of scientific interest because they may be harnessed for quantum computing, a technology that would be far faster than current computers for some specialized uses.
A second group of I.B.M. scientists in Zurich were able to place two hydrogen atoms in an ultrathin insulating film and switch them back and forth between two states, creating the equivalent of the ones and zeroes used in standard chips. They were also able to use the same switching process to inject an electric charge into one molecule and link the effect to a neighboring molecule. That suggests it might be possible to extend the effect into a fabric of trillions of atom-size switches in the future.
The laboratory advances are far from being ready to commercialize, but they provide hope for the electronics industry, which has grown steadily because of the continuous shrinking in size and falling cost of components for more than four decades.
The scientists documented their work in two papers appearing on Friday in the journal Science. Both papers are focused on new understanding of the behavior of magnetism at the tiny scale of nanotechnology, where scientists hope to develop electronics made from components that are far smaller than today’s transistors and wires.
In one paper the researchers describe a technique for reading and writing digital ones and zeroes onto a handful of atoms, or even individual atoms. The second paper describes the ability to use a single molecule as a switch, replicating the behavior of today’s transistors.
The papers are the latest indication that computing technology is beginning to emerge that could replace today’s microelectronics materials in the next decade.
R. Stanley Williams, a Hewlett-Packard physicist, said this week that his group had begun manufacturing prototypes of a silicon chip that combines both conventional microelectronics and molecular scale components. Their first hybrid device is a circuit called a field programmable gate array, or F.P.G.A., using molecular-scale components as the configuration circuitry, an approach that will save tremendous space in the chip design.
A team of I.B.M. researchers at the company’s Almaden Research Center in San Jose, Calif., were able to use a scanning tunneling microscope to observe the magnetic orientation of iron and manganese atoms at low temperatures. Controlling magnetic direction is a crucial technique that is used in reading and writing digital information on magnetic storage disks like standard hard drives.
In addition to the potential storage applications, the researchers noted that atomic-scale magnetic structures are also of scientific interest because they may be harnessed for quantum computing, a technology that would be far faster than current computers for some specialized uses.
A second group of I.B.M. scientists in Zurich were able to place two hydrogen atoms in an ultrathin insulating film and switch them back and forth between two states, creating the equivalent of the ones and zeroes used in standard chips. They were also able to use the same switching process to inject an electric charge into one molecule and link the effect to a neighboring molecule. That suggests it might be possible to extend the effect into a fabric of trillions of atom-size switches in the future.
The laboratory advances are far from being ready to commercialize, but they provide hope for the electronics industry, which has grown steadily because of the continuous shrinking in size and falling cost of components for more than four decades.
Fausto Intilla
lunedì 27 agosto 2007
World's Highest-resolution Computer Display Reaches 220 Million Pixels In Resolution
Source:
Science Daily — Engineers at the University of California, San Diego have constructed the highest-resolution computer display in the world – with a screen resolution up to 220 million pixels.
The system located at the UCSD division of the California Institute for Telecommunications and Information Technology (Calit2) is also linked via optical fiber to Calit2’s building at UC Irvine, which boasts the previous record holder. The combination – known as the Highly Interactive Parallelized Display Space (HIPerSpace) – can deliver real-time rendered graphics simultaneously across 420 million pixels to audiences in Irvine and San Diego.
“We don’t intend to stop there,” said Falko Kuester, Calit2 professor for visualization and virtual reality and associate professor of structural engineering in UCSD’s Jacobs School of Engineering. “HIPerSpace provides a unique environment for visual analytics and cyberinfrastructure research and we are now seeking funding to double the size of the system at UC San Diego alone to reach half a billion pixels with a one gigapixel distributed display in sight.”
Kuester is the chief architect of the systems deployed in both Calit2 buildings. Until last week, UC Irvine’s Highly Interactive Parallelized Display Wall (HIPerWall) – built in 2005 with funding from the National Science Foundation (NSF) – held the record of 200 million pixels for a tiled display system. It is located in the Calit2 Center of Graphics, Visualization and Imaging Technology (GRAVITY), which Kuester directs. When Kuester’s group moved to UCSD in 2006 they began work on the next generation of massively tiled display walls, which now serve as a prototype for ultra-high resolution OptIPortal tiled displays developed by the NSF-funded OptIPuter project (led by Calit2 director Larry Smarr).
The new HIPerSpace system between Irvine and San Diego is joined together via high-performance, dedicated optical networking that clocks in at up to two gigabits per second (2Gbps). The systems use the same type of graphics rendering technology, from industry partner NVIDIA. The “graphics super cluster” being developed at UCSD consists of 80 NVIDIA Quadro FX 5600 graphics processing units (GPUs). “The graphics and computational performance of these cards is quite astounding,” said Kuester. “Putting the theoretical computational performance of the cluster at almost 40 teraflops. To put that into context, the top-rated supercomputer in the world five years ago was operating at that same speed. While these are purely theoretical numbers, the comparison clearly hints at capabilities of this new cluster that go far beyond generating impressive visual information.”
The processing power will come in handy for the kinds of large-scale applications that are likely to make use of the HIPerSpace system. Calit2 will make the displays available to teams of scientists or engineers dealing with very large data sets, from multiple gigabytes to terabytes, notably in the Earth sciences, climate prediction, biomedical engineering, genomics, and brain imaging. “The higher-resolution displays allow researchers to take in both the broad view of the data and the minutest details, all at the same time,” said Kuester. “HIPerSpace also allows us to experiment on the two campuses with distributed teams that can collaborate and share insights derived from a better understanding of complex results. This capability will allow researchers at two UC campuses to collaborate more intensively with each other, and eventually with other campuses, thanks to the rapid rollout of OptIPortals outside of California.”
In San Diego, the OptIPortal is deployed on the second floor of Atkinson Hall, next to the offices of the NEES Cyberinfrastructure Center (NEESit), which supports the NSF-funded George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES) and its 15 sites around the country. “Structural engineering simulations require a massive amount of data processing and visualization, especially if you need to crunch data coming in from all of the NEES participating sites,” said Kuester. “We are also using the system for a large-scale, global seismicity visualization using data collected over the past thirty years.”
“I am excited that UC Irvine’s HIPerWall is now networked to its larger sibling,” said Stephen Jenks, professor of electrical engineering and computer science at UC Irvine and a participant in Calit2 at UCI. “With the high-speed OptIPuter network between the two Calit2 buildings, we will be able to virtually join the display walls at a distance of nearly 100 miles, so they can work together to show different parts of a huge data set or each can replicate parts of the other. We look forward to exploring remote collaboration technology and how the two systems can help researchers understand data better than ever before.”
UCSD’s HIPerSpace OptIPortal is similar to the HIPerWall because both are tiled display systems, but with different hardware. Irvine’s version is constructed with 50 Apple 30-inch Cinema Displays, powered by 25 Power Mac G5s running the Mac OS X operating system. UCSD’s Linux-based OptIPortal consists of 55 Dell displays driven by 18 Dell XPS personal computers. The system at UCSD uses the San Diego Supercomputer Center’s new 64-bit version of grid-computing middleware known as ROCKS released in early August and Calit2’s Cluster GL for heterogeneous systems (CGLX) framework, which is capable of supporting both systems concurrently.
"The usability of high-performance visualization clusters such as HiPerSpace is bound tightly to the accessibility of its resources, so cumbersome script configuration and specially-written software are no longer viable,” said Calit2 postdoctoral researcher Kai-Uwe Doerr. “The visualization software developed here at Cailt2 was designed to provide an efficient and transparent mechanism to grant access to available graphics resources and make the transition of a desktop application to a cluster seamless and uncomplicated – with minimal or no changes to the original code." Doerr and Kuester are part of a large team making HIPerSpace a reality. Others at UC San Diego include So Yamaoka; Daniel Knoblauch; Jason Kimball; Kevin Ponto; Tung-Ju Hsieh; Andres Fernandez Munuera; Tom DeFanti; Greg Dawe; Stephen Jenks; and Duy-Quoc Lai, who will start graduate school at UC Irvine this fall. Other UCI researchers on the HIPerWall end of the project include Harry Mangalam, Frank Wessel, Charlie Zender, Soroosh Sorooshian, Jean-Luc Gaudiot and Sung-Jin Kim.
Note: This story has been adapted from a news release issued by University of California, San Diego.
“We don’t intend to stop there,” said Falko Kuester, Calit2 professor for visualization and virtual reality and associate professor of structural engineering in UCSD’s Jacobs School of Engineering. “HIPerSpace provides a unique environment for visual analytics and cyberinfrastructure research and we are now seeking funding to double the size of the system at UC San Diego alone to reach half a billion pixels with a one gigapixel distributed display in sight.”
Kuester is the chief architect of the systems deployed in both Calit2 buildings. Until last week, UC Irvine’s Highly Interactive Parallelized Display Wall (HIPerWall) – built in 2005 with funding from the National Science Foundation (NSF) – held the record of 200 million pixels for a tiled display system. It is located in the Calit2 Center of Graphics, Visualization and Imaging Technology (GRAVITY), which Kuester directs. When Kuester’s group moved to UCSD in 2006 they began work on the next generation of massively tiled display walls, which now serve as a prototype for ultra-high resolution OptIPortal tiled displays developed by the NSF-funded OptIPuter project (led by Calit2 director Larry Smarr).
The new HIPerSpace system between Irvine and San Diego is joined together via high-performance, dedicated optical networking that clocks in at up to two gigabits per second (2Gbps). The systems use the same type of graphics rendering technology, from industry partner NVIDIA. The “graphics super cluster” being developed at UCSD consists of 80 NVIDIA Quadro FX 5600 graphics processing units (GPUs). “The graphics and computational performance of these cards is quite astounding,” said Kuester. “Putting the theoretical computational performance of the cluster at almost 40 teraflops. To put that into context, the top-rated supercomputer in the world five years ago was operating at that same speed. While these are purely theoretical numbers, the comparison clearly hints at capabilities of this new cluster that go far beyond generating impressive visual information.”
The processing power will come in handy for the kinds of large-scale applications that are likely to make use of the HIPerSpace system. Calit2 will make the displays available to teams of scientists or engineers dealing with very large data sets, from multiple gigabytes to terabytes, notably in the Earth sciences, climate prediction, biomedical engineering, genomics, and brain imaging. “The higher-resolution displays allow researchers to take in both the broad view of the data and the minutest details, all at the same time,” said Kuester. “HIPerSpace also allows us to experiment on the two campuses with distributed teams that can collaborate and share insights derived from a better understanding of complex results. This capability will allow researchers at two UC campuses to collaborate more intensively with each other, and eventually with other campuses, thanks to the rapid rollout of OptIPortals outside of California.”
In San Diego, the OptIPortal is deployed on the second floor of Atkinson Hall, next to the offices of the NEES Cyberinfrastructure Center (NEESit), which supports the NSF-funded George E. Brown, Jr. Network for Earthquake Engineering Simulation (NEES) and its 15 sites around the country. “Structural engineering simulations require a massive amount of data processing and visualization, especially if you need to crunch data coming in from all of the NEES participating sites,” said Kuester. “We are also using the system for a large-scale, global seismicity visualization using data collected over the past thirty years.”
“I am excited that UC Irvine’s HIPerWall is now networked to its larger sibling,” said Stephen Jenks, professor of electrical engineering and computer science at UC Irvine and a participant in Calit2 at UCI. “With the high-speed OptIPuter network between the two Calit2 buildings, we will be able to virtually join the display walls at a distance of nearly 100 miles, so they can work together to show different parts of a huge data set or each can replicate parts of the other. We look forward to exploring remote collaboration technology and how the two systems can help researchers understand data better than ever before.”
UCSD’s HIPerSpace OptIPortal is similar to the HIPerWall because both are tiled display systems, but with different hardware. Irvine’s version is constructed with 50 Apple 30-inch Cinema Displays, powered by 25 Power Mac G5s running the Mac OS X operating system. UCSD’s Linux-based OptIPortal consists of 55 Dell displays driven by 18 Dell XPS personal computers. The system at UCSD uses the San Diego Supercomputer Center’s new 64-bit version of grid-computing middleware known as ROCKS released in early August and Calit2’s Cluster GL for heterogeneous systems (CGLX) framework, which is capable of supporting both systems concurrently.
"The usability of high-performance visualization clusters such as HiPerSpace is bound tightly to the accessibility of its resources, so cumbersome script configuration and specially-written software are no longer viable,” said Calit2 postdoctoral researcher Kai-Uwe Doerr. “The visualization software developed here at Cailt2 was designed to provide an efficient and transparent mechanism to grant access to available graphics resources and make the transition of a desktop application to a cluster seamless and uncomplicated – with minimal or no changes to the original code." Doerr and Kuester are part of a large team making HIPerSpace a reality. Others at UC San Diego include So Yamaoka; Daniel Knoblauch; Jason Kimball; Kevin Ponto; Tung-Ju Hsieh; Andres Fernandez Munuera; Tom DeFanti; Greg Dawe; Stephen Jenks; and Duy-Quoc Lai, who will start graduate school at UC Irvine this fall. Other UCI researchers on the HIPerWall end of the project include Harry Mangalam, Frank Wessel, Charlie Zender, Soroosh Sorooshian, Jean-Luc Gaudiot and Sung-Jin Kim.
Note: This story has been adapted from a news release issued by University of California, San Diego.
Fausto Intilla
sabato 25 agosto 2007
MIT aims to optimize chip designs (Model could reduce fabrication costs)
Source:
Anne Trafton,
News Office, August 16, 2007
The computer chips inside high-speed communication devices have become so small that tiny variations that appear during chip fabrication can make a big difference in performance.
Those variations can cause fluctuations in circuit speed and power so the chips don't meet their original design specifications, says MIT Professor Duane Boning, whose research team is working to predict the variation in circuit performance and maximize the number of chips working within the specifications.
The researchers have recently developed a model to characterize the variation in one kind of chip. The model could be used to estimate the ability to manufacture a circuit early in the development stages, helping to optimize chip designs and reduce costs.
"We're getting closer and closer to some of the limits on size, and variations are increasing in importance," says Boning, a professor of electrical engineering and computer science (EECS) and associate head of the department. "It's becoming much more difficult to reduce variation in the manufacturing process, so we need to be able to deal with variation and compensate for it or correct it in the design."
Boning and EECS graduate student Daihyun Lim's model characterizes variation in radio frequency integrated circuits (RFICs), which are used in devices that transfer large amounts of data very rapidly, such as high-definition TV receivers.
The researchers published their results in two papers in February and June. They also presented a paper on the modeling of variation in integrated circuits at this year's International Symposium on Quality Electronic Design.
RFIC chips are essential in many of today's high-speed communication and imaging devices. Shrinking the size of a chip's transistors to extremely small dimensions (65 nanometers, or billionths of a meter), improves the speed and power consumption of the RFIC chips, but the small size also makes them more sensitive to small and inevitable variations produced during manufacturing.
"The extremely high speeds of these circuits make them very sensitive to both device and interconnect parameters," said Boning, who is also affiliated with MIT's Microsystems Technology Laboratories. "The circuit may still work, but with the nanometer-scale deviations in geometry, capacitance or other material properties of the interconnect, these carefully tuned circuits don't operate together at the speed they're supposed to achieve."
Every step of chip manufacturing can be a source of variation in performance, said Lim. One source that has become more pronounced as chips have shrunk is the length of transistor channels, which are imprinted on chips using lithography.
"Lithography of very small devices has its optical limitation in terms of resolution, so the variation of transistor channel length is inevitable in nano-scale lithography," said Lim.
The researchers' model looks at how variation affects three different properties of circuits--capacitance, resistance and transistor turn-on voltage. Those variations cannot be measured directly, so Lim took an indirect approach: He measured the speed of the chip's circuits under different amounts of applied current and then used a mathematical model to estimate the electrical parameters of the circuits.
To the researchers' surprise, they found correlations between some of the variations in each of the three properties, but not in others. For example, when capacitance was high, resistance was low. However, the transistor threshold voltage was nearly independent of the parasitic capacitance and resistance. The different degrees of correlation should be considered in the statistical simulation of the circuit performance during design for more accurate prediction of manufacturing yield, said Lim.
The research was funded by the MARCO/DARPA Focus Center Research Program's Interconnect Focus Center and Center for Circuits and Systems Solutions, and by IBM, National Semiconductor and Samsung Electronics.
Fausto Intilla
mercoledì 15 agosto 2007
Computers Expose The Physics Of NASCAR
Source:
Science Daily — It's an odd combination of Navier-Stokes equations and NASCAR driving.Computer scientists at the University of Washington have developed software that is incorporated in new technology allowing television audiences to instantaneously see how air flows around speeding cars. The algorithm, first presented at a computer graphics conference last August, was since used by sports network ESPN and sporting-technology company Sportvision Inc. to create a new effect for racing coverage.
The fast-paced innovation hit prime time in late July when ESPN used the Draft Track technology to visualize the air flow behind cars in the Allstate 400 at the Brickyard, a NASCAR race at the Indianapolis Motor Speedway. Zoran Popovi, an associate professor in the UW's department of computer science and engineering, and two students wrote the code that dramatically speeds up real-time fluid dynamics simulations. Working with ESPN, a Chicago-based company named Sportvision developed the application for NASCAR competition. The Draft Track application calculates air flow over the cars and then displays it as colors trailing behind the car. Green, blue, yellow and red correspond to different speeds and directions for air flow when two or more cars approach one another while driving at speeds upward of 200 miles per hour. "What ESPN wanted to do is tell the story for the viewer of how drafting works because it's such a big part of the event," said Rick Cavallaro, chief scientist at Sportvision. "How the drivers use drafting to save gas, pick up speed, et cetera." The UW researchers' breakthrough was figuring out how to simulate and display complex systems very quickly. Studios such as Pixar already use physical laws, such as the Navier-Stokes fluids equations, in their animations. This allows the studios to create realistic pictures of how smoke curls, how a fire's flames lick, and even how hair or fabric blows in the wind. But these calculations take hours to run on many high-performance computers. And increasing the speed of the image is only one challenge of moving to a real-time setting. "The studios shoot a two-second special effect and if it doesn't work they just change the parameters and try again," Popovi said. "But in a real-time context the simulation has to run indefinitely, and for an unforeseen set of inputs." To make the simulation work in real time and be interactive, "you kind of need to rethink the math problem," he said. "The method that ended up being used is drastically different from what people have done before." The new algorithm first simulates all the ways that smoke, fire -- or in this case, modified stock cars -- can behave. Then it runs the simulation for a reduced number of physically possible parameters. This allows the model to run a million times faster than before. The researchers presented the work at the SIGGRAPH computer graphics conference in August 2006. Popovi imagined that the first applications would be introducing interactive simulations in video games that would allow players to drive through a smoky fire, interact with the weather in a flight simulator, or drive racecars in a virtual wind tunnel. Other research results from his lab were licensed to the game industry and then adopted in video games. But in March, Sportvision approached the researchers to see whether it could license the software for use in NASCAR visualizations. The two parties agreed to a nonexclusive, open-source agreement where the company would be allowed to use the technique. "What's interesting is how the flow from the car in front is affecting the cars behind," Popovi said. "When there are two cars behind, then the interaction becomes very complex." Sportvision creates technology to enhance sports coverage. It introduced the glowing puck for National Hockey League telecasts in the mid 1990s and later came up with the yellow lines that drag a virtual highlighter over the first-down line in football. The company has already developed add-ons for ESPN's NASCAR coverage, placing Global Positioning System receivers, inertial measurement systems and telemetry on each car that can determine each car's speed and position several times a second. Now company engineers will use data from those sensors to model and display the air flowing over the cars. "This is certainly not an application that had occurred to me," admitted Adrien Treuille, a doctoral student who co-authored the software. He, like Popovi, said he had not previously been a NASCAR fan. The group hoped the work might be used for realistic training simulations such as firefighters entering a smoke-filled building. "But once [Sportvision] called us and started describing what they wanted to do," Treuille recalled, "We said, 'Yes, that would totally work.'" Article: "Model reduction for real-time fluids".
Note: This story has been adapted from a news release issued by University of Washington.
Fausto Intilla
NCAR Adds Resources To TeraGrid
Source:
Science Daily — Researchers who use the TeraGrid, the nation's most comprehensive and advanced infrastructure for open scientific research, can now leverage the computing resources of a powerful, 2048-processor BlueGene/L system at the National Center for Atmospheric Research (NCAR).
NCAR plans to provide up to 4.5 million processor-hours of BlueGene/L computing annually to researchers who have received computing grants from the National Science Foundation (NSF).The operational integration of TeraGrid with the BlueGene/L system, nicknamed "frost," involved extensive preparation by NCAR's Computational and Information Systems Laboratory (CISL). Engineers deployed the necessary networking infrastructure, then established connectivity to NCAR's data storage systems, and merged the local resource accounting system with the TeraGrid. "We are excited to be at a point where all our hard work and preparation pays off, and to provide the TeraGrid community with access to this valuable collaborative resource," says Richard Loft, NCAR TeraGrid principal investigator. NCAR is also testing experimental systems and services on the TeraGrid. These include the wide-area versions of general parallel file systems from IBM and Cluster File Systems, as well as a remote data visualization capability based on the VAPOR tool, an open source application developed by NCAR, the University of California, Davis, and Ohio State University under the sponsorship of NSF.NCAR's frost system, which is operated in partnership with the University of Colorado, will be the second BlueGene/L system on the TeraGrid, joining the San Diego Supercomputer Center's 6,144 processor system. With the addition of frost, the TeraGrid has more than 250 teraflops of computing capability and more than 30 petabytes of online and archival data storage, with rapid access and retrieval over high-performance networks.About the TeraGridThe TeraGrid, sponsored by the National Science Foundation Office of Cyberinfrastructure, is a partnership of people, resources, and services that enables discovery in U.S. science and engineering. Through coordinated policy, grid software, and high-performance network connections, the TeraGrid integrates a distributed set of high-capability computational, data-management and visualization resources to make research more productive.
Note: This story has been adapted from a news release issued by National Center for Atmospheric Research.
Fausto Intilla
Iscriviti a:
Post (Atom)
