Green Computing for a Clean Tomorrow

1
Green Computing for a Clean Tomorrow
Prof. Wu FENG, feng_at_cs.vt.edu College of
Engineering, Depts. of CS ECE
Goal Deliver high performance while reducing
power energy consumption and improving
reliability.
Approaches
Motivation

• Power Consumption Heat Generation Hurt
  Reliability, Availability, Total Cost of
  Ownership
• Electrical Power for Computing Costs
• Earth Simulator 12 MW/year ? 10M/year
• Worlds Processors 1.3 GW/year ? 1B/year
• Hiding in Plain Sight, Google Seeks More Power,
  
• The New York Times, June 14, 2006.
• Computing Contributes to Global Warming

• Low-Power, High-Performance Computing
• Green Destiny A 240-Node Supercomputer in 5 Sq.
  Ft. with a 3.2-kW Power Envelope
• Reliability

3.2 kW
MTBI mean time between interrupt MTBF mean
time between failure MTTR mean time to restore
2001 to 2006
40
30.0 kW
4
Just the processors (i.e., CPUs) in PCs 40
Hoover Dams (Estimated power consumption of PCs
120 Hoover Dams)

• Power-Aware, High-Performance Computing

Observation
Self-Adapting Software for Energy Efficiency
Conserve power energy WHILE programs run.
Arrenhius Equation (applied to microelectronics)
Twenty Years of Empirical Data For every 10C
increase in temperature, the failure rate of the
system doubles.

• Many commodity technologies support dynamic
  voltage frequency scaling (DVFS), which allows
  changes to the processor voltage and frequency at
  run time.
• A computing system can trade off processor
  performance for power reduction.
• Power a V2f, where V is the supply voltage of the
  processor and f is its frequency.
• Processor performance a frequency.
• Approach Intelligent DVFS Scheduling
• Determine when to adjust the voltage-frequency
  setting and what to adjust it to.

Hypothesis
Reduce power consumption ? Reduce system
temperature ? Reduce failure rate
The Project Supercomputing in Small Spaces
NAS/NPB 3.2 MPI, C.16

• Improve efficiency, reliability, availability,
• and usability of computing systems.
• Sacrifice a bit of raw speed to reduce power
  energy consumption.
• Improve overall throughput as the system will
  always be available, i.e., effectively no
  downtime.
• Reduce total cost of ownership increase return
  on investment.
• Crude Analogy
• Formula One Race Car Wins raw performance but
  reliability is so poor that it requires frequent
  maintenance. Throughput low.
• Honda S2000 Loses raw performance but high
  reliability results in high throughput (i.e.,
  miles driven/month ? answers/month).

Parallel Codes
Sequential Codes
relative time / relative energy with respect to
total execution time and system energy usage
Energy savings and performance improvement!
Selected Publications

• Making a Case for a Green500 List, 20th IEEE
  Intl Parallel Distributed Processing Symp.,
  Apr. 2006.
• A Power-Aware Run-Time System for
  High-Performance Computing, SC05, Nov. 2005.
• The Importance of Being Low Power in
  High-Performance Computing, CTWatch Quarterly
  (NSF), 1(3)12-20, Aug. 2005.
• Green Destiny and Its Evolving Parts,
  Innovative Supercomputer Architecture Award, 19th
  Intl Supercomputer Conf., Jun. 2004.
• Green Destiny mpiBLAST Bioinfomagic, 10th
  Intl Conf. on Parallel Computing (ParCo), Sept.
  2003.
• Honey, I Shrunk the Beowulf! 31st Intl Conf.
  on Parallel Processing, Aug. 2002.

Featured in The New York Times, CNN, and BBC
News Now in The Computer History Museum
Laboratory