| Abstract |
|---|
| Many scientific applications handle compressed sparse matrices. Cache behaviorduring the execution of codes with irregular access patterns, such asthose generated by this type of matrices, has not been widely studied.In this work a probabilistic model for the prediction of thenumber of misses on a direct mapped cache memory consideringsparse matrices with an uniform distribution ispresented. As an example of the potential usability of such types of models, and takinginto account the state of the art with respect to high performancesuperscalar and/or superpipelined CPUs with a multilevel memoryhierarchy, we have modeled the cache behavior of an optimized sparsematrix-dense matrix product algorithm including blocking at the memoryand register levels. |
| Contact |
| Ramsn Doallo Dept. Electrsnica e Sistemas,Facultade de Informatica,Universidade da Coruqa,Campus de Elviqa, s/n,15071, A Coruqa,SPAIN, doallo@udc.es |