David R. Kaeli

VCSR-TC: A Sparse Matrix-Matrix Optimization Strategy for Leveraging Modern GPUs with Tensor Cores

Sparse–dense matrix multiplication (SpMM) underpins graph neural networks, recommendation systems, and pruned-transformer inference, yet remains memory-bound on modern GPUs because nonzero column indices induce scattered, noncoalesced accesses to the dense operand B. Recent tensor-core SpMM kernels reconstruct the spar...