An international team of researchers, led by scientists at The Genome Institute at Washington University School of Medicine in St. Louis and the University of … Read more
Johns Hopkins biomedical engineers have teamed up with clinicians to create a new drug-delivery strategy for a type of central vision loss caused by blood … Read more
An anti-cancer drug has been proven to be equally as effective in treating the most common cause of blindness in older adults as a more … Read more
The latest Green500 Listshows two continuing trends despite seven of the world’s Top 10 greenest supercomputers changing up from the previous June 2011 rankings: Many … Read more
Intel-rival Advanced Micro Devices got a nice science win Monday when Sandia National Laboratory and Cray Inc. said they would build a supercomputer capable 40 trillion calculations per second using AMD’s forthcoming Opteron processor. Ten thousand of them, to be precise. Total cost: $90 million. Sandia says it will use the computing heavyweight for “modeling and simulation of complex problems that were only recently thought impractical, if not impossible.”
The New York Times and the Associated Press take long looks at flip sides of the coming Intel-Advanced Micro Devices war over 64-bit processors. In “Intel’s Huge Bet Turns Iffy,” The Times examines the company’s enormous investment of time and money — 10 years and $5 billion — into the chip, which it co-developed with Hewlett-Packard. The Infineon, or more precisely the Infineon 2 (an earlier version 1 was largely considered a flop) handles enormous quantities of data, but also uses lots of electricity. While the former is nice, the Times says, companies that run big server farms are increasingly mindful about all the juice needed to keep them running and cooled. AMD’s own 64-bit entry, the Hammer, uses less electricity, the Times notes ominously. But as the AP reports, there are no guarantees for AMD either. One of Hammer’s big selling points is that it is backward compatible with current x86 software, meaning anything you’re running now on a Windows box and more. That could make it an appealing crossover product, tempting for use in corporate servers and consumer desktops alike. But with a soft economy and many people happy with the speed they’ve already got, anything short of a groundswell adoption could be a major bummer for the perennial no. 2.
Apple in recent years has sought to close the megahertz gap with Intel and Advanced Micro Devices by selling high-end machines that come with two processors instead of one. Thus, a dual-processor 1.2GHz G4 can reasonably claim to compete with a 2.5GHz Intel box. But Intel reckons there’s more than one way to build a two-brained beast and is experimenting with putting two processor cores on a single piece of silicon. It’s a process that CNET’s News.com says over the next decade will improve performance and reduce power consumption. One advantage is the ability to better dissipate heat. When a series of calculations has Core A chugging along at full steam, its transistors can hit temperatures that start to degrade performance. When that happens, the chip can “hop” some of the work over to Core B for an overall performance improvement. Another approach is to have Core A and Core B specialize in different things, similarly spreading the number-crunching responsibility around.
Computer chips have been shrinking for years. But who stops to consider that that’s only been possible because the stuff on the chips, like circuits, transistors and memory have shrunk too? To keep the trend going, Germany’s Infineon has joined Advanced Micro Devices and United Microelectronics Corp. to develop technology to produce the tiny structures needed inside chips. As the number of elements on a chip doubles approximately every year, “chipmakers are under pressure to develop new microelements to fit on (them),” Reuters reports. Currently, the size of the smallest element on a chip is 130nm. The three-way alliance will focus on developing a 65nm and 45nm manufacturing process.
Intel this week launches the Itanium 2 server processor. The company is betting big that it can convince companies which build servers to use the new chip as ubiquitously as PC makers now employ its lower-end CPUs. But the Wall Street Journal (subscription required) reports Intel has its work cut out. By Intel’s own admission, most corporate clients are already plenty happy with the oomph their existing 32-bit processors provide. And for folks that need supercomputer-level power, clustering many weaker, but far cheaper chips may provide an appealing cost-benefit ratio. Yes, there is that segment — which in revenue terms is substantial — willing to pay a premium for 64-bit processors. But in that arena Intel will compete with entrenched products from IBM, Hewlett-Packard and Sun. Finally, if and when all those dragons are slain, pokey old Advanced Micro Devices is coming round the bend with a new line of “Hammer” chips that feature both 32- and 64-bit capability. Five years from now it’ll be a kick to read a “The Soul of a New Machine“-style chronicle of how Intel handles the coming slugfest.