nostr:npub1pzyjfwn57mnh0y778ctqs05qkyqdk99k589y0v5mdqcx76jy0ass6apz3v nostr:npub1cv874nnmd6tgm7l8s5gz9eh37tl4ed5kwd76txn3h4e4kacx3tssmucvxj nostr:npub1knzsux7p6lzwzdedp3m8c3c92z0swzc0xyy5glvse58txj5e9ztqpac73p I suspect one could relatively easily be more clever about programming. Has anyone tried to implement it for a GPU, for example? It feels intuitively like the sort of thing that could make good use of that style of parallel programming.