Holistic design approaches have gained in prominence in microprocessor design research, as they show tremendous potential in improving multiple design objectives such as energy efficiency and reliability. However, holistic techniques encompassing multiple design layers, which are traditionally decoupled, are facing severe challenges from the economic feasibility standpoint. On the other hand, to reduce the design costs, several industry leaders are advocating techniques to effectively involve design automation in multiple layers of computer system design.

This research combines two promising directions for future computer designs: holistic perspective and embracing design automation at multiple design layers. The research builds a foundation for realizing an end-to-end holistic design, where design choices from multiple layers are seamlessly integrated. Joint design space exploration is performed in logic and physical design, while considering system level architectural design choices and their implication on workload execution. Two central themes are explored under this paradigm: (a) improving system wide energy efficiency, and (b) improving system reliability focusing on intermittent timing fault vulnerability. Innovations in computer design technology that surpass boundaries between multiple design layers can be modeled and investigated using this framework, stimulating research on far-reaching problems.

Research in the area of cross-layer design will enable circuit designers and system architects to increase collaborative design to develop affordable, energy-efficient and reliable computer systems. Major research insights will be integrated in courses taught by the PIs to train computer engineers. Skill sets disseminated through teaching will create new specialties and lead to more efficient future computer systems.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Network Systems (CNS)
Type
Standard Grant (Standard)
Application #
1117425
Program Officer
M. Mimi McClure
Project Start
Project End
Budget Start
2011-08-15
Budget End
2014-07-31
Support Year
Fiscal Year
2011
Total Cost
$386,000
Indirect Cost
Name
Utah State University
Department
Type
DUNS #
City
Logan
State
UT
Country
United States
Zip Code
84322