Efficient sparse matrix-vector multiplication on cache-based GPUs | IEEE Conference Publication | IEEE Xplore