February 2020 paper
spECK: accelerating GPU sparse matrix-matrix multiplication through lightweight analysis