Semi Conscious
Comment
Semiguru
And what will be the contributions to the US economy and welfare, considering ...
Texas-sized supercomputer rising in Austin
Dylan McGrath
11/19/2012 9:01 AM EST
In less than two months, officials at the Texas Advanced Computing Center (TACC) at the University of Texas-Austin will flip the switch on Stampede, a Dell PowerEdge C8220 supercomputer built in partnership with Dell and Intel.
Stampede, scheduled to be fully operational on Jan. 7, 2013, will feature 6,400 Dell C8220X nodes comprising 12,800 Intel Xeon E5 processors plus more than 6,000 Intel Xeon Phi coprocessors. According to TACC, the Xeon processors provide more than 2 petaflops of peak performance, while the Xeon Phi coprocessors will add more than 7 petaflops of additional performance.
Stampede—the successor to TACC's Ranger and Lonestar 4 systems—entered the semi-annual Top 500 supercomputers list lastweek at No. 7. Stampede—funded by the National Science Foundation (NSF)—was selected as the first installation to deploy Intel's Many Integrated Core (MIC) architecture Xeon Phi coprocessors. The supercomputer will be used by some of the U.S.'s top scientists for simulation and data analysis in research.
Weighing in at 500,000 pounds, Stampede features 182 racks, uses 5 megawatts of power, and boasts a peak performance approaching 10 petaflops.
According to Jay Boisseau, TACC's director, Stampede was designed to support comprehensive general science, as opposed to being narrowly-focused. "No one system can be everything to everybody," Boisseau said. "We have configured Stampede to be a hell of a system to most people."
Boisseau said he expects Stampede will be used on about 700 projects per year. The system is expected to be in use for at least four years, but could be extended by the NSF after that, he said.
For Stampede, TACC got early access to Intel's Xeon Phi coprocessors, codenamed Knights Corner. Intel announced last week at the Supercomputing 2012 conference that it has begun shipping the first Xeon Phi coprocessor, 5110P, with general availability expected in January. A second Xeon Phi family, 3100, is expected to be available in the first half of 2013.
Stampede, scheduled to be fully operational on Jan. 7, 2013, will feature 6,400 Dell C8220X nodes comprising 12,800 Intel Xeon E5 processors plus more than 6,000 Intel Xeon Phi coprocessors. According to TACC, the Xeon processors provide more than 2 petaflops of peak performance, while the Xeon Phi coprocessors will add more than 7 petaflops of additional performance.
Stampede—the successor to TACC's Ranger and Lonestar 4 systems—entered the semi-annual Top 500 supercomputers list lastweek at No. 7. Stampede—funded by the National Science Foundation (NSF)—was selected as the first installation to deploy Intel's Many Integrated Core (MIC) architecture Xeon Phi coprocessors. The supercomputer will be used by some of the U.S.'s top scientists for simulation and data analysis in research.
Weighing in at 500,000 pounds, Stampede features 182 racks, uses 5 megawatts of power, and boasts a peak performance approaching 10 petaflops.
According to Jay Boisseau, TACC's director, Stampede was designed to support comprehensive general science, as opposed to being narrowly-focused. "No one system can be everything to everybody," Boisseau said. "We have configured Stampede to be a hell of a system to most people."
Boisseau said he expects Stampede will be used on about 700 projects per year. The system is expected to be in use for at least four years, but could be extended by the NSF after that, he said. For Stampede, TACC got early access to Intel's Xeon Phi coprocessors, codenamed Knights Corner. Intel announced last week at the Supercomputing 2012 conference that it has begun shipping the first Xeon Phi coprocessor, 5110P, with general availability expected in January. A second Xeon Phi family, 3100, is expected to be available in the first half of 2013.
|
|
|
|
|
Navigate to related information
Semiguru
11/20/2012 11:53 PM EST
And what will be the contributions to the US economy and welfare, considering the tera- petaflops competition between the various computing centers over the past 2-3 decades???
Sign in to Reply