Software data prefetching to reduce data reloading overhead

In this paper, we propose a software data prefetching mechanism to insert prefetching instructions efficiently into a loop. Proposed mechanism analyzes the array reference pattern, and inserts prefetching instructions into the loop according to the results of the analysis. Analyzing the loop, it considers cache size and the number of array elements that will be used to exploit reuse property. However, in case that it is hard to benefit by the reuse due to the large number of array elements, proposed mechanism reorganizes the loop properly to reduce the overhead of reloading array elements that had been pushed out from the cache.