Fast, vectorized morton code encoding. Completely portable and only relies on compiler optimizations.

morton.c

#include <stdint.h>
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>

static inline uint64_t split(uint64_t x, int log_bits) {
    uint64_t mask = (UINT64_C(1) << (1 << log_bits)) - 1;
    for (int i = log_bits, n = 1 << log_bits; i > 0; --i, n >>= 1) {
        mask = (mask | (mask << n)) & ~(mask << (n / 2));
        x = (x | (x << n)) & mask;
    }
    return x;
}

uint64_t encode(uint64_t x, uint64_t y, uint64_t z, int log_bits) {
    return split(x, log_bits) | (split(y, log_bits) << 1) | (split(z, log_bits) << 2);
}

int main(int argc, char** argv) {
    size_t n_iters = 500;
    uint64_t res = 0;
    for (size_t iter = 0; iter < n_iters; ++iter) {
        for (uint64_t i = 0; i < 256 * 256 * 256; ++i)
            res += encode(i & 0xFF, (i >> 8) & 0xFF, (i >> 16) & 0xFF, 5);
    }
    printf("%"PRIu64"\n", res);
    return 0;
}