A Comparison of Compiler Tiling Algorithms

Linear algebra codes contain data locality which can be exploited by tiling multiple loop nests. Several approaches to tiling have been suggested for avoiding connict misses in low associativity caches. We propose a new technique based on intra-variable padding and compare its performance with existing techniques. Results show padding improves performance of matrix multiply by over 100% in some cases over a range of matrix sizes. Comparing the eecacy of diierent tiling algorithms, we discover rectangular tiles are slightly more eecient than square tiles. Overall, tiling improves performance from 0-250%. Copying tiles at run time proves to be quite eeective.