This is the mail archive of the
mailing list for the GCC project.
Re: [gfortran] Implement irand(), rand(), and srand() for G77 compatibility
On Mon, May 31, 2004 at 12:50:41AM +0100, Paul Brook wrote:
> > irand(i) returns a value in the range [1,GFC_RAND_M-1].
> > irand(i) - 1 is in the range [0,GFC_RAND_M-2]. So, the
> > division above is in the range [0,1).
> You're forgetting that real*4 has significantly less than 32 bits precision,
> so is unable to accurately represent integers of this magnitude. After
> rounding the result will in the range [0-1]
Thanks for the wack with the clue stick.
The simple fix is
return (GFC_REAL_4) ((double)(prefix(irand)(i) - 1) / (double)GFC_RAND_M1);
I'm not sure if this works on amd64, alpha, sparc64, etc.
A more complicate fix is
#define PRECS 32
GFC_REAL_4 div32 (GFC_INTEGER_4 num, GFC_INTEGER_4 dem)
mpf_t n, d;
mpf_init2 (mpf_t n, PRECS);
mpf_init2 (mpf_t d, PRECS);
mpf_set_ui (n, num);
mpf_set_ui (d, dem);
mpf_div (n, n, d);
val = (GFC_REAL_4) mpf_get_d(n);
/* Return a REAL in the range [0,1). */
GFC_REAL_4 prefix(rand) (GFC_INTEGER_4 *i)
val = div32(prefix(irand)(i) - 1, GFC_RAND_M1);
For rand(), I think the simple fix is in order. Also, it seems
that libgfortran does not use gmp.