Accelerating rational drug design, large-scale genomic search, facilitating CS education, human computer interaction, and for cancer research.

HokieSpeed is the highest-ranked commodity

supercomputer in the U.S. on the Green500, topping out at 928.96 MFLOPS/watt. Feng serves as the

principal investigator for this project, which was supported

by the Major Research Instrumentation (MRI)

program at National Science Foundation (NSF).

synergy.cs.vt.edu www.cs.vt.edu

At the Synergy Lab, we conduct basic and applied

research that explores a breadth of complementary intellectual activities that span the high-performance

& enterprise-wide spectrum – from systems software to middleware to applications software to tools – in

order to empower cyber-scientists and cyber-engineers of tomorrow.

http://goo.gl/AbjU2

Virginia Tech: HokieSpeed

accel.cs.vt.edu www.mpiblast.org myvice.cs.vt.edu sss.cs.vt.edu

Overview

(12/20/2011) Virginia Tech unveils HokieSpeed at #11 on the Green500 and #96 on the TOP500

http://goo.gl/gpojE http://goo.gl/UAL5X http://goo.gl/uYCAI

Prof. Wu Feng of Virginia Tech talks about their use of

GPU

Virginia Tech Supercomputer Aims

to Accelerate Research

VT nears completion of HokieSpeed

Research Areas Green

Computing Measurement, modeling, prediction, and management of power and energy; metrics, methodologies, and workloads; benchmarking; Green500.

Networking Architecture, performance measurement and evaluation, protocols, routing and fundamentals.

Renaissance

Systems Systems software and tools; adaptive run-time systems, parallel computing, monitoring and measurement, emerging platforms in all forms and scales.

www.green500.org www.chrec.org

Exploring AMD Fusion

Demonstrating seamless

disconnection/reconnection

over network

Benchmarking GPUs in a

compute cluster

Power metering

Research Publications

Visit research group website http://synergy.cs.vt.edu/publications.php

Making sense of 24TB of gene sequence data

Helping kids explore

computer science

synergy.cs.vt.edu

The Supercomputing in Small Spaces (SSS) project began in September 2001 with a 24-node Bladed Beowulf cluster dubbed MetaBlade, which consumed only 400 watts of power and occupied five square feet of space. By April 2002, the SSS project unveiled its 240-node "Green Destiny" cluster, which consumed as little as 3.2 kilowatts (i.e., two hairdryers) while still occupying only five square feet of space and delivering over 100 Gflops on the LINPACK benchmark, which would have placed it at #393 on the TOP500 List at the time. Due to the audacity of Green Destiny, it ended up being featured in The New York Times, CNN, and PC World to name a few and currently resides in the Computer History Museum.

Since 2002, the SSS project has evolved in two different directions: (1) a low-power, architectural approach and (2) a power-aware, software-based

approach.

Those who bring synergy to the lab

Accelerators ‘R Us Since the 1960s, the general-purpose processor (also known as the central processing unit or CPU) has served as the brains in computing instruments. Recent trends, however, have exposed the CPU as a "jack of all (computing) trades, master of none." To address this, heterogeneous computing systems with multiple types of brains have emerged to herald a new age in supercomputing. Building on our expertise in this area at Virginia Tech, we seek to tackle all aspects of accelerator-based parallel computing from systems software to middleware and libraries to

applications.

Faculty and Staff

Students

W. Feng M. Gardner H. Lin

Ashwin

Vignesh Marwa Umar

Konstantinos Kenneth

Nabeel Paul Tom

Lokendra Balaji

Parang Shucai

Jing

Green500 The Green500 provides rankings of the most energy-efficient supercomputers in the world. We raise awareness about power consumption, promote alternative total cost of ownership performance metrics, and ensure that supercomputers only simulate climate change and not create it.

Supercomputing in Small Spaces

Lee

Those who fund our research

mpiBLAST mpiBLAST is a freely available, open-source, parallel implementation of NCBI BLAST. By efficiently utilizing distributed computational resources through database fragmentation, query segmentation, intelligent scheduling, and parallel I/O, mpiBLAST improves NCBI BLAST performance by several orders of magnitude while scaling to hundreds of processors. mpiBLAST is also portable across many different platforms and operating systems.

MyVICE Our aim is to introduce the power of (parallel) computing to the classroom at an early age. Feng started off with teaching a class of 2nd graders LOGO and transitioned to "programming with pictures" through Storytelling Alice and more recently, Scratch for

grades 3 – 8.

Mariam