Activity Number:
|
130
- Statistical Computation, Simulation, and Computer Experiments
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 31, 2017 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
Abstract #322943
|
|
Title:
|
Experiment Design and Analysis for Novel Computer Hardware Testing
|
Author(s):
|
Sarah Michalak* and Lisa Moore and John Daly and Nathan DeBardeleben and Gary Delp and Laura Monroe and Heather Quinn and James Rose
|
Companies:
|
Los Alamos National Laboratory and Los Alamos National Laboratory and University of Maryland and Los Alamos National Laboratory and Mayo Clinic and Los Alamos National Laboratory and Los Alamos National Laboratory and Mayo Clinic
|
Keywords:
|
Computational Resiliency ;
Computer Hardware ;
Experiment Design ;
Fault Tolerant Algorithm ;
Reliability
|
Abstract:
|
Increasing computational demands are making energy efficient computation more crucial. Novel compute hardware can be designed to be more energy efficient at the cost of increased faults. Using such hardware along with resilient or fault tolerant algorithms that can produce correct answers in the presence of faults may lead to increased energy efficiency of computation. Specifically, it may be possible to exploit the trade-off between the accuracy and energy efficiency of compute hardware in the presence of algorithmic resilience to find a "sweet spot" for which faulty hardware enables energy savings that persist despite any additional computation required by resilience mechanisms.
This talk presents results from early testing of fault tolerant approaches with novel compute hardware including experiment design and analysis of results.
|
Authors who are presenting talks have a * after their name.