GE-SpMM: General-Purpose Sparse Matrix-Matrix Multiplication on GPUs for Graph Neural Networks | IEEE Conference Publication | IEEE Xplore