Optimizing and auto-tuning scale-free sparse matrix-vector multiplication on Intel Xeon Phi | IEEE Conference Publication | IEEE Xplore