PPOPP
2010
ACM
14 years 2 months ago
2010
ACM
Loop vectorization, a key feature exploited to obtain high performance on Single Instruction Multiple Data (SIMD) vector architectures, is significantly hindered by irregular memo...