Questiable LCG

[Wolf-SO]

I really appreciate this API: _meta_inline int32 lcgRandom(_Inout_ uint32* state) as it is reentrant.

But I found the range of outputs generated by the function suspicious.

The classic implementation seems to be this:

// Source - https://stackoverflow.com/q/8569113
// Posted by Adam Stelmaszczyk, modified by community. See post 'Timeline' for change history
// Retrieved 2026-05-21, License - CC BY-SA 3.0

static unsigned long int next = 1;

int rand(void)  /* RAND_MAX assumed to be 32767. */
{
    next = next * 1103515245 + 12345;
    return (unsigned)(next/65536) % 32768;
}

And a discussion of another “classic” approach that is close to lcgRandom is this:

static unsigned int next = 0;
void srand(unsigned int seed) {
    next = seed;
}
/* RAND_MAX assumed to be 32767 */
int rand(void) {
    next = next * 1103515245 + 12345;
    return next % 32768;
}

Which is discussed in Computer Security Dawn Song (CS 161, Fall 2008, Notes 4) on page 2. I'd expect that a modulus of 2<sup>31</sup> would result in a comparable biased output. I mean, does it really change the low entropy in the lower bits?

Or to put it another way: In what ways is your implementation better than the following one?

_meta_inline int32 lcgRandom(_Inout_ uint32* state)
{
    return ((*state = *state * 1103515245 + 12345) & 0x7fffffff);
}