Tech's New Supercomputer Far Faster

Weng Fu and the new Tech supercomputer.

Virginia Tech crashed the supercomputing arena in 2003 with System X, a machine that placed the university among the world’s top computational research facilities. Now comes HokieSpeed, a new supercomputer that is up to 22 times faster and yet a quarter of the size of X, boasting a single-precision peak of 455 teraflops, or 455 trillion operations per second, and a double-precision peak of 240 teraflops, or 240 trillion operations per second.

That’s enough computational capability to place HokieSpeed at No. 96 on the most recent Top500 List, the industry-standard ranking of the world’s 500 fastest supercomputers. More intriguing is HokieSpeed’s energy efficiency, which ranks it at No. 11 in the world on the November 2011 Green500 List, a compilation of supercomputers that excel at using less energy to do more. On the Green500 List, HokieSpeed is the highest-ranked commodity supercomputer in the United States.

Located at Virginia Tech’s Corporate Research Center, HokieSpeed – the word “Hokie” originating from an old Virginia Tech sports cheer – contains 209 nodes, or separate computers, connected to one another in and across large metal racks, each roughly 6.5 feet tall, to create a single supercomputer that occupies half a row of racks in a vast university computer machine room. X took three times the rack space.

Each HokieSpeed node contains two 2.40-gigahertz Intel Xeon E5645 6-core central processing units, commonly called CPUs, and two NVIDIA M2050/C2050 448-core graphics processor units, or GPUs, which reside on a Supermicro 2026GT0TRF motherboard. That gives HokieSpeed more than 2,500 central processing unit cores and more than 185,000 graphics processor unit cores to compute with.

“HokieSpeed is a versatile heterogeneous supercomputing instrument, where each compute node consists of energy-efficient central-processing units and high-end graphics-processing units,” said Wu Feng, associate professor with the Virginia Tech College of Engineering’s computer science and electrical and computer engineering departments.

“This instrument will empower faculty, students, and staff across disciplines to tackle problems previously viewed as intractable or that required heroic efforts and significant domain-specific expertise to solve.”