Optimizing Sparse Matrix Operations on GPUs Using Merge Path | IEEE Conference Publication | IEEE Xplore