ABSTRACT

Memory Coherence Activity Prediction in Commercial Workloads

Stephen Somogyi, Thomas F. Wenisch, Nikolaos Hardavellas, Jangwoo Kim, Anastassia Ailamaki, Babak Falsafi

Carnegie Mellon University
Pittsburgh, PA 15213

Recent research indicates that prediction-based coherence optimi-zations offer substantial performance improvements for scientific applica-tions in distributed shared memory multiprocessors. Important commercial applications also show sensitivity to coherence latency, which will become more acute in the future as technology scales. Therefore it is important to in-vestigate prediction of memory coherence activity in the context of commer-cial workloads.

This paper studies a trace-based Downgrade Predictor (DGP) for predicting last stores to shared cache blocks, and a pattern-based Consumer Set Predic-tor (CSP) for predicting subsequent readers. We evaluate this class of predic-tors for the first time on commercial applications and demonstrate that our DGP correctly predicts 47%-76% of last stores. Memory sharing patterns in commercial workloads are inherently non-repetitive; hence CSP cannot at-tain high coverage. We perform an opportunity study of a DGP enhanced through competitive underlying predictors, and in commercial and scientific applications, demonstrate potential to increase coverage up to 14%.

FULL PAPER: pdf

Last updated 16 February, 2004