| NSF Org: |
CCF Division of Computing and Communication Foundations |
| Recipient: |
|
| Initial Amendment Date: | July 21, 2016 |
| Latest Amendment Date: | July 21, 2016 |
| Award Number: | 1643056 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Almadena Chtchelkanova
achtchel@nsf.gov (703)292-7498 CCF Division of Computing and Communication Foundations CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | August 1, 2016 |
| End Date: | July 31, 2018 (Estimated) |
| Total Intended Award Amount: | $299,970.00 |
| Total Awarded Amount to Date: | $299,970.00 |
| Funds Obligated to Date: |
|
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
201 PRESIDENTS CIR SALT LAKE CITY UT US 84112-9049 (801)581-6903 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
UT US 84112-9205 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): | Software & Hardware Foundation |
| Primary Program Source: |
|
| Program Reference Code(s): |
|
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
Numerical algorithms used in Cyber Physical Systems, decision-making systems, financial processing, and other HPC applications that use real numbers are prone to introduce computational errors because of a well-known reason: real numbers do not exist in computers, and we must use floating-point data types to approximate such computations. As data movement costs energy, the lowest precision of floating-point data must be allocated without compromising the computational integrity. This project implements methods to reduce the amount of energy consumed by numerical computations running on computing devices at all scales including supercomputers for scientific research all the way to embedded and mobile devices finding uses in many walks of real life including medical devices and robots. A key thrust of the work is to perform energy reduction through reduced transfers between computing units. The project studies how the number of bits used to represent data introduce errors in computations, and whether these errors affect the correctness of results.
The PIs propose to develop new formal methods tools to automatically estimate error bounds, develop auto-tuning compilers to carefully select precision, and build new superoptimizers to generate more efficient code. These new technologies will be applied to improve software in the domains of machine learning and high-performance computing. The PIs shall develop suitable criteria for errors in high performance computing systems and machine learning systems. They will develop tools that allocate precision optimally while staying within the bounds of acceptable answers. Their tools will be released to a community of researchers interested in working toward exascale computing, and deploying machine learning applications in safety-critical devices. This work represents a synergistic combination of PI skills ranging through high performance computing, machine learning, formal methods, and compiler technologies.
PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH
Note:
When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external
site maintained by the publisher. Some full text articles may not yet be available without a
charge during the embargo (administrative interval).
Some links on this page may take you to non-federal websites. Their policies may differ from
this site.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
When developing computer representations of continuous quantities (real numbers), a designer allocates enough data bits to model the data with the requisite precision. However, allocating more than the required number of bits leads to excessive memory consumption (especially the cache memory that is a precious resource). It also increases energy consumption which is proportional to the amount of data moved. In this research project, we studied the precision allocation problem from a number of different perspectives. We chose two primary application domains: machine learning, and high performance computing.
Our first major finding is that conducting rigorous precision allocation requires tools for accurate roundoff error analysis. Our second major finding is that the nature of the application -- machine learning or high performance computing -- has a significant impact on the impact of precision allocation on application behavior. Our third major finding is that even within a domain such as high performance computing, the nature of the task carried out -- whether it involves discrete decisions or involves primarily arithmetic operations --decides how one can optimally trim precision in order to reduce energy costs without affecting application behavior. In the domain of machine learning, the decision space is is even more complex, and depends on whether precision tuning is being done while training or during deployment of a trained system.
The major tool related contribution of this project is a tool called FPTaylor that has been engineered to high standards, and has been downloaded and studied by a number of groups. A comprehensive journal article presents a detailed comparative study of FPTaylor against other tools in its class. The project was instrumental in training five PhD students and one MS student in this area of critical national importance.
Last Modified: 09/01/2018
Modified by: Ganesh L Gopalakrishnan
Please report errors in award information by writing to: awardsearch@nsf.gov.
