wallscratch·2 years ago·discussCan anyone comment on how efficient the Warp code is compared to manually written / fine-tuned CUDA?