Attachment 7381 Details for Bug 16516 – test case

test case

test4.c (text/x-c), 1.17 KB, created by Paul Zimmermann on 2014-01-31 22:30:58 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Paul Zimmermann

Created: 2014-01-31 22:30:58 UTC

Size: 1.17 KB

patch

obsolete

>#include <stdio.h>
>#include <math.h>
>#include <mpfr.h>
>#include <gnu/libc-version.h>
>#include <fenv.h>
>
>float
>ulp (float x, float y)
>{
>  float z = x - y, t, twoy;
>
>  twoy = y + y;
>  if (twoy != twoy + twoy) /* twoy is finite */
>    t = nextafterf (y, twoy);
>  else /* twoy is +Inf or -Inf */
>    t = nextafter (y, (float) 0);
>  t = y - t;
>  if (t < (float) 0)
>    t = -t;
>  /* now t = ulp(y) */
>  return z / t;
>}
>
>int
>main ()
>{
>  float x = 0x3.8b7f28p-128;
>  float y, z;
>  mpfr_t t;
>  int inex, runderflow;
>
>  printf("GNU libc version: %s\n", gnu_get_libc_version ());
>  printf("GNU libc release: %s\n", gnu_get_libc_release ());
>  fesetround (FE_TOWARDZERO);
>  feclearexcept (FE_ALL_EXCEPT);
>  y = erff (x);
>  runderflow = fetestexcept (FE_UNDERFLOW);
>  mpfr_set_emin (-125 - 24 + 1);
>  mpfr_set_emax (128);
>  mpfr_init2 (t, 24);
>  mpfr_set_flt (t, x, MPFR_RNDN);
>  inex = mpfr_erf (t, t, MPFR_RNDZ);
>  inex = mpfr_subnormalize (t, inex, MPFR_RNDZ);
>  z = mpfr_get_flt (t, MPFR_RNDD);
>  printf ("x=%a (%.9e)\n", x, x);
>  printf ("libc: erff(x)=%a (%.9e) underflow=%d\n", y, y, runderflow);
>  printf ("mpfr: erff(x)=%a (%.9e)\n", z, z);
>  printf ("ulp difference: %f\n", ulp (y, z));
>  return 0;
>}

Actions: View

Attachments on bug 16516: 7381