DocumentCode
3443369
Title
Improving Branch Prediction Related WCET Abstract Interpretation
Author
Louise, Stephane
Author_Institution
LIST, CEA, Gif-sur-Yvette, France
Volume
2
fYear
2011
fDate
28-31 Aug. 2011
Firstpage
130
Lastpage
133
Abstract
In this paper we refine a Markovian model previously applied to modelize a subset of dynamic branch predictors related WCET delays, in order to take the finiteness of Branch History Table (BHT) into account. The base of the model allows the modeling of a deterministic or non deterministic automaton or set of automata thanks to a Markovian model. We previously applied this formalism to cache related delay and cache related preemption delays and simple dynamic branch prediction with infinite BHT. Now we would like to take into account the finiteness of the BHT in the previous model in order to estimate extend our model of static analysis of Worst Case Execution Time (WCET). For illustration, we chose the case of a dynamic prediction based on a two bit saturating counter, with a LRU policy for replacement policy.
Keywords
cache storage; deterministic automata; program diagnostics; LRU policy; Markovian model; bit saturating counter; branch history table; branch prediction related WCET abstract interpretation; cache related delay; cache related preemption delays; deterministic automaton; dynamic branch predictors related WCET delays; infinite BHT; nondeterministic automaton; replacement policy; static analysis; worst case execution time; Analytical models; Benchmark testing; Concurrent computing; History; Pipelines; Predictive models; Radiation detectors; Branch predictor; Markovian model; WCET; static analysis;
fLanguage
English
Publisher
ieee
Conference_Titel
Embedded and Real-Time Computing Systems and Applications (RTCSA), 2011 IEEE 17th International Conference on
Conference_Location
Toyama
ISSN
1533-2306
Print_ISBN
978-1-4577-1118-3
Type
conf
DOI
10.1109/RTCSA.2011.74
Filename
6029873
Link To Document