Don't have Telegram yet? Try it now!
https://hgpu.org/?p=2729
Model-driven autotuning of sparse matrix-vector multiply on GPUs