Romu – Fine random number generators

Re:Romu – Fine random number generators by edflsafoiewq on Saturday February 29, 2020 @04:56AM (#22449874)

re the ILP argument ("app continues running with no delay because it already has the random number it needs"): can't any RNG take advantage of that by caching the next value to return? ie. turning this

    state_t s;
    uint rng_next() {
        X;
        return Y;
    }

into

    state_t s;
    uint next;
    uint rng_next() {
        uint n = next;
        X;
        next = Y;
        return n;
    }

[ Parent | on HN ]

Re:Romu – Fine random number generators by MarkOverton on Sunday March 01, 2020 @01:27AM (#22455474)

I mention this idea in the paper (which is linked at the top of the website). The problem is that this technique increases register pressure when the generator is inlined, which in turn increases spills, which reduces performance.

[ Parent | on HN ]