- 1 Vapor IO Brings OpenDCRE to General Availability
- 2 VMware Takes the Wraps Off vRealize Automation and vRealize Business
- 3 Microsoft Previews Hyper-V Containers for Windows Server 2016
- 4 Mirantis Led FUEL Project Gets Installed Under OpenStack Big Tent
- 5 Red Hat Enterprise Linux 7.2 Adds Security, DR Features
IBM Supercomputer Shatters Own Speed Record
An IBM Blue Gene supercomputer has nearly doubled its previous world-beating speed, performing at 135.3 trillion floating points per second, said the Department of Energy (DoE). Blue Gene/L now simulates the nuclear arms stockpile at more than 135 teraflops.
The DoE's National Nuclear Security Administration (NNSA), which uses the machine to help maintain the U.S. nuclear weapons stockpile, said the new speed trounced the supercomputer's previous speed of 70.7 teraflops.
When it is finished this summer, BlueGene/L will have 65,000 nodes for a total of 131,072 processors, said Don Johnston, a spokesman for Lawrence Livermore National Laboratory in California, which is where the system is being tested.
The previous fastest speed of 70.7 was championed by NNSA last November with the release of the latest Linpack computing benchmark.
At 70.7 teraflops last fall, scientists at Lawrence Livermore were able to perform molecular dynamics simulations of 16 million atoms with the greatest accuracy ever recorded. BlueGene/L also helped scientists studying the effects of voids in metal failure perform such simulations with more than 2.1 billion atoms.
But the fact that the researchers were able to boost the supercomputer's speed so considerably from November to March speaks to the ability for such machines to be considerably scaled to perform processing-intensive tasks. When it is finished this summer, BlueGene/L is expected to run more than 64,000 IBM Power processors at 360 teraflops.
To Big Blue's delight, BlueGene/L currently sits comfortably atop the Top500 supercomputing list as the most powerful machine in the world. NASA's SGI Altix, code-named Voltaire, is currently second at 51.9 teraflops.
While the numbers are impressive for technology vendors, the research the machines enable is more important; it includes such areas as weather modeling, meteorology, and simulating the effects of space on humans and objects.
This article was originally published on internetnews.com.