I discovered a problem in glibc-2.2.4-11.2.src.rpm concerning QNaN and
SNaN (quiet not-a-number and signalling not-a-number). It seems that
NaN's are always interpreted the Intel-way.
In glibc-2.2.4/soft-fp/quad.h it says:
#define _FP_QNANBIT_Q \
((_FP_W_TYPE)1 << (_FP_FRACBITS_Q-2) % _FP_W_TYPE_SIZE)
But this is not correct on MIPS processors, where SNaNs has the bit set
and QNaNs has the bit cleared. Changing the definition above to 0 doesn't
seem to solve the problem (I probably haven't found the root of the
problem).
The following piece of code gives the wrong result on a MIPS processor:
-------------------- cut here --------------
#include <math.h>
typedef union {
long long ll;
double d;
} t_number;
int main()
{
t_number x, z;
x.ll = 0x7ff7ffffffffffff; /* QNaN */
z.d = sqrt(x.d);
printf("%e %016llx\n", z.d, z.ll);
}
-------------------- cut here --------------
The result is a signalling NaN:
nan 7ff8000000000000
(I expected 0x7ff7ffffffffffff)
/Kjeld
On Fri, 14 Sep 2001, H . J . Lu wrote:
> Here is a patch for sysdeps/mips/fpu/libm-test-ulps. BTW, I got many
> math failures on Linux/mipsel. Has anyone else seen them?
>
>
> H.J.
> ----
> 2001-09-14 H.J. Lu <hjl@gnu.org>
>
> * sysdeps/mips/fpu/libm-test-ulps: Updated.
>
> --- sysdeps/mips/fpu/libm-test-ulps.mips Fri Apr 27 21:25:17 2001
> +++ sysdeps/mips/fpu/libm-test-ulps Fri Sep 14 11:01:52 2001
> @@ -7,7 +7,7 @@ ifloat: 2
> Test "asin (0.5) == pi/6":
> float: 2
> ifloat: 2
> -Test "asin (0.7) == 0.7753974966107530637":
> +Test "asin (0.7) == 0.77539749661075306374035335271498708":
> double: 1
> float: 2
> idouble: 1
> @@ -175,12 +175,12 @@ idouble: 1
> Test "Imaginary part of: cexp (-2.0 - 3.0 i) == -0.13398091492954261346140525546115575 - 0.019098516261135196432576240858800925 i":
> float: 1
> ifloat: 1
> -Test "Real part of: cexp (0.7 + 1.2 i) == 0.7296989091503236012 + 1.8768962328348102821 i":
> +Test "Real part of: cexp (0.7 + 1.2 i) == 0.72969890915032360123451688642930727 + 1.8768962328348102821139467908203072 i":
> double: 1
> float: 1
> idouble: 1
> ifloat: 1
> -Test "Imaginary part of: cexp (0.7 + 1.2 i) == 0.7296989091503236012 + 1.8768962328348102821 i":
> +Test "Imaginary part of: cexp (0.7 + 1.2 i) == 0.72969890915032360123451688642930727 + 1.8768962328348102821139467908203072 i":
> float: 1
> ifloat: 1
>
> @@ -249,7 +249,7 @@ float: 1
> ifloat: 1
>
> # cos
> -Test "cos (0.7) == 0.7648421872844884262":
> +Test "cos (0.7) == 0.76484218728448842625585999019186495":
> double: 1
> float: 1
> idouble: 1
> @@ -374,7 +374,7 @@ double: 2
> float: 1
> idouble: 2
> ifloat: 1
> -Test "exp10 (0.7) == 5.0118723362727228500":
> +Test "exp10 (0.7) == 5.0118723362727228500155418688494574":
> float: 1
> ifloat: 1
> Test "exp10 (3) == 1000":
> @@ -451,6 +451,21 @@ ifloat: 2
> Test "j0 (8.0) == 0.17165080713755390609":
> float: 1
> ifloat: 1
> +Test "j0 (4.0) == -3.9714980986384737228659076845169804197562E-1":
> +double: 1
> +float: 1
> +idouble: 1
> +ifloat: 1
> +ildouble: 1
> +ldouble: 1
> +Test "j0 (-4.0) == -3.9714980986384737228659076845169804197562E-1":
> +double: 1
> +float: 1
> +idouble: 1
> +ifloat: 1
> +ildouble: 1
> +ldouble: 1
> +
>
> # j1
> Test "j1 (10.0) == 0.043472746168861436670":
> @@ -563,7 +578,7 @@ idouble: 1
> ifloat: 1
>
> # sincos
> -Test "sincos (0.7, &sin_res, &cos_res) puts 0.76484218728448842626 in cos_res":
> +Test "sincos (0.7, &sin_res, &cos_res) puts 0.76484218728448842625585999019186495 in cos_res":
> double: 1
> float: 1
> idouble: 1
> @@ -573,7 +588,7 @@ double: 1
> float: 0.5
> idouble: 1
> ifloat: 0.5
> -Test "sincos (M_PI_6l*2.0, &sin_res, &cos_res) puts 0.866025403784438646764 in sin_res":
> +Test "sincos (M_PI_6l*2.0, &sin_res, &cos_res) puts 0.86602540378443864676372317075293616 in sin_res":
> double: 1
> float: 1
> idouble: 1
> @@ -583,7 +598,7 @@ double: 0.2758
> float: 0.3667
> idouble: 0.2758
> ifloat: 0.3667
> -Test "sincos (pi/6, &sin_res, &cos_res) puts 0.866025403784438646764 in cos_res":
> +Test "sincos (pi/6, &sin_res, &cos_res) puts 0.86602540378443864676372317075293616 in cos_res":
> float: 1
> ifloat: 1
>
> @@ -605,6 +620,13 @@ double: 1
> float: 1
> idouble: 1
> ifloat: 1
> +Test "tanh (-0.7) == -0.60436777711716349631":
> +double: 1
> +float: 1
> +idouble: 1
> +ifloat: 1
> +ildouble: 1
> +ldouble: 1
>
> # tgamma
> Test "tgamma (-0.5) == -2 sqrt (pi)":
>
--
_ _ ____ ___ Mailto:kjelde@mips.com
|\ /|||___)(___ MIPS Denmark Direct: +45 44 86 55 85
| \/ ||| ____) Lautrupvang 4 B Switch: +45 44 86 55 55
TECHNOLOGIES DK-2750 Ballerup Fax...: +45 44 86 55 56
Denmark http://www.mips.com/
