When the University of Florida was looking to avoid a potential $100,000 energy surcharge, as well as accelerate the second phase of its high-performance computing deployment by as much as a year, it found the answer in dual-core enabled servers from Rackable Systems Inc.
The combination of Rackable's high-density servers that are half the depth of conventional rack-mount equipment, which allows for installation from both the front and back of a chassis, and dual-core Opteron processors from Advanced Micro Devices Inc., has let the university increase its computing capacity by 1-1/2 to two times what it had anticipated this year, says Charles Taylor, senior HPC systems engineer at the university.
"It has been such a win because you lose nothing in terms of scalability and you get twice the computational power at almost no cost in terms of power and cooling," Taylor says. "It's a very hard thing to beat."
To meet the university's requirements to provide computational capabilities for use in such applications as basic sciences to applied engineering and drug development, the University of Florida planned a three-phased project to create as much as 9 teraflops of total performance.
The project begun with phase one implementation in February 2004: A 384-processor cluster based on Xeon processors in servers from "one of the big four equipment vendors," Taylor says.
What skills do network managers really need to properly secure industrial networks? What new protocols, frameworks, and regulations are important? And what conferences and certifications can help? Here are five tips to get started.
A full-stack approach to retail edge offers retailers a way to optimize operations and adapt to changes in a post-pandemic world.
Network management tool sprawl is getting in the way of network management. It’s time for IT to do something about it.
