Welcome!
To use the personalized features of this site, please log in or register.
If you have forgotten your username or password, we can help.
My Menu
Saved Items
Abstract

We present a novel pre-runtime scheduling method for uni-processors which precisely incorporates the effects of task switching on the processor cache into its decisions. Tasks are modelled as a sequence of non preemtable segments with precedence constraints. The cache behavior of each task segment is statically determined by interpretation. For the sake of efficiency, the scheduling algorithm uses a heuristically guided search strategy. Each time a new task segment is added to a partial schedule, its worst case execution time is calculated based on the cache state at the end of the preceding partial schedule.

Fulltext Preview (Small, Large)
Image of the first page of the fulltext


Export this chapter
Export this chapter as RIS | Text
 
Remote Address: 38.107.191.108 • Server: mpweb03
HTTP User Agent: CCBot/1.0 (+http://www.commoncrawl.org/bot.html)