Happy With Four Cores? Stanford Researchers Are Working With A Million!
Quad-core systems are amazing, right? What about a million-core system! Yep, you read it right! Stanford Engineering's Center for Turbulence Research (CTR) has broken the million-core supercomputer barrier with the newly installed Sequoia IBM Bluegene/Q systems at Lawrence Livemore National Laboratories.
Joseph Nicholas, a research associate, worked on the newly installed system and showed for the first time that it is possible to perform simulation over a million-core system. Sequoia, one of  the most powerful supercomputers in the world, has got 1,572,864 processors (cores) and 1.6 petabytes of memory connected by a high-speed five-dimensional torus interconnect. Nicholas used the system to predict the noise generated by a supersonic jet engine.
 #-Link-Snipped-#
According to Parviz Moin, Director of CTR, using supercomputers with hundreds of thousands of computing cores, engineers have been able to model jet engines and the noise they produce with accuracy and speed. Though it all looks too pretty, the computation becomes a lot complex with a million cores. At this level, previously innocuous parts of the computer code can suddenly become bottleneck.
via #-Link-Snipped-#
Joseph Nicholas, a research associate, worked on the newly installed system and showed for the first time that it is possible to perform simulation over a million-core system. Sequoia, one of  the most powerful supercomputers in the world, has got 1,572,864 processors (cores) and 1.6 petabytes of memory connected by a high-speed five-dimensional torus interconnect. Nicholas used the system to predict the noise generated by a supersonic jet engine.
 #-Link-Snipped-#
According to Parviz Moin, Director of CTR, using supercomputers with hundreds of thousands of computing cores, engineers have been able to model jet engines and the noise they produce with accuracy and speed. Though it all looks too pretty, the computation becomes a lot complex with a million cores. At this level, previously innocuous parts of the computer code can suddenly become bottleneck.
via #-Link-Snipped-#
0