|Title:||Studying new ways for improving adaptive history length branch predictors||Authors:||Falcón, Ayose
Santana, Oliverio J.
|UNESCO Clasification:||330406 Arquitectura de ordenadores||Keywords:||Branch prediction
Dynamic history length
|Issue Date:||2002||Journal:||Lecture Notes in Computer Science||Conference:||4th International Symposium on High Performance Computing, ISHPC 2002||Abstract:||Pipeline stalls due to branches limit processor performance significantly. This paper provides an in depth evaluation of Dynamic History Length Fitting, a technique that changes the history length of a two-level branch predictor during the execution, trying to adapt to its different phases. We analyse the behaviour of DHLF compared with fixed history length gshare predictors, and contribute showing two factors that explain DHLF behaviour: Opportunity Cost and Warm-up Cost. Additionally, we evaluate the use of profiling for detecting future improvements. Using this information, we show that new heuristics that minimise both opportunity cost and warm-up cost could outperform significantly current variable history length techniques. Especially at program start-up, where the algorithm tries to learn the behaviour of the program to better predict future branches, the use of profiling reduces considerably the cost produced by continuous history length changes.||URI:||http://hdl.handle.net/10553/43889||ISBN:||978-3-540-43674-4
|ISSN:||0302-9743||DOI:||10.1007/3-540-47847-7_23||Source:||Zima H.P., Joe K., Sato M., Seo Y., Shimasaki M. (eds) High Performance Computing. ISHPC 2002. Lecture Notes in Computer Science, vol 2327. Springer, Berlin, Heidelberg|
|Appears in Collections:||Actas de congresos|
checked on Jul 4, 2020
Items in accedaCRIS are protected by copyright, with all rights reserved, unless otherwise indicated.