Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Appears in MICRO-33 Compiler Controlled Value Prediction using Branch Predictor Based Confidence
 

Summary: Appears in MICRO-33
1
Compiler Controlled Value Prediction using Branch Predictor Based Confidence
Eric Larson and Todd Austin
Electrical Engineering and Computer Science
University of Michigan
{larsone,austin}@eecs.umich.edu
Abstract
Value prediction breaks data dependencies in a pro-
gram thereby creating instruction level parallelism that
can increase program performance. Hardware based
value prediction techniques have been shown to
increase speed, but at great cost as designs include pre-
diction tables, selection logic, and a confidence mecha-
nism. This paper proposes compiler-controlled value
prediction optimizations that obtain good speedups
while keeping hardware costs low. The branch predic-
tor is used to estimate the confidence of the value pre-
dictor for speculated instructions. This technique
obtains 4.6% speedup when completely implemented in

  

Source: Austin, Todd M. - Department of Electrical Engineering and Computer Science, University of Michigan
Larson, Eric - Department of Computer Science and Software Engineering, Seattle University

 

Collections: Computer Technologies and Information Sciences; Engineering