64-bit multiply with carry (MWC) for Intel x86-64

We don't have a uint128_t yet, but we can access rdx:rax. 

static uint64_t Q[R] = {}, c=1;

uint64_t mwc64() {
    uint64_t a = A;
    static int i = ~0;

    i = (i + 1) % R;
    asm(
        "mulq %%rdx\n\taddq %%rcx,%%rax\n\tadcq $0,%%rdx"
         : /*out*/ "=a" (Q[i]), "=d" (c)
         : /*in*/ "a" (Q[i]), "c" (c), "d" (a) );    
    return Q[i]; 
}

and some values for A and R:

A	R	period (approx)
18446744073709550874	1	2127
18446744073709547559	2	2191
18446744073709530150	4	2319
18446744073709525530	8	2575
18446744073709398669	16	21087
18446744073709219500	32	22111
18446744073707195115	64	24159
18446744073709214688	128	28255
18446744073688074570	256	216447