e5ca265a9c
new inputs with large errors for [a]cospi, [a]sinpi, [a]tanpi, atan2pi
...
These inputs were generated with the programs from
https://gitlab.inria.fr/zimmerma/math_accuracy ,
with rounding to nearest:
* for univariate binary32 functions by exhaustive search
* for other functions with the "threshold" parameter up to 10^6
2025-01-02 18:26:36 +01:00
b14224fb57
math: xfail some cospi tests for ibm128-libgcc
...
On powerpc math/test-ibm128-cospi shows:
testing long double (without inline functions)
Failure: cospi_downward (-0xf.ffffffffffffbffffffffffffcp+1020): Exception "Invalid operation" set
Failure: cospi_downward (-0xf.ffffffffffffbffffffffffffcp+1020): Exception "Overflow" set
Failure: cospi_downward (-0xf.ffffffffffffbffffffffffffcp+1020): errno set to 33, expected 0 (unchanged)
Failure: Test: cospi_downward (-0xf.ffffffffffffbffffffffffffcp+1020)
Result:
is: qNaN
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
Failure: Test: cospi_downward (0x3.fffffffffffffffcp+108)
Result:
is: 9.99999999999999999999999999995574e-01 0x1.ffffffffffffffffffffffff4c8p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 4.42501664022411309598141492088312e-30 0x1.670000000000000000000000000p-98
ulp : 179.5000
max.ulp : 4.0000
Failure: Test: cospi_downward (0x3.ffffffffffffffffffffffffffp+108)
Result:
is: 9.99999999999999999999999999996524e-01 0x1.ffffffffffffffffffffffff730p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 3.47591836363008326759542899077727e-30 0x1.1a0000000000000000000000000p-98
ulp : 141.0000
max.ulp : 4.0000
Failure: Test: cospi_towardzero (-0x3.fffffffffffffffcp+108)
Result:
is: 9.99999999999999999999999999852310e-01 0x1.ffffffffffffffffffffffe8990p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 1.47689552599346303944427057331536e-28 0x1.767000000000000000000000000p-93
ulp : 5991.0000
max.ulp : 4.0000
Failure: Test: cospi_towardzero (-0x3.ffffffffffffffffffffffffffp+108)
Result:
is: 9.99999999999999999999999999952569e-01 0x1.fffffffffffffffffffffff87c0p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 4.74302619264133348003801799876275e-29 0x1.e10000000000000000000000000p-95
ulp : 1924.0000
max.ulp : 4.0000
Failure: Test: cospi_towardzero (0x3.fffffffffffffffcp+108)
Result:
is: 9.99999999999999999999999999852310e-01 0x1.ffffffffffffffffffffffe8990p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 1.47689552599346303944427057331536e-28 0x1.767000000000000000000000000p-93
ulp : 5991.0000
max.ulp : 4.0000
Failure: Test: cospi_towardzero (0x3.ffffffffffffffffffffffffffp+108)
Result:
is: 9.99999999999999999999999999952569e-01 0x1.fffffffffffffffffffffff87c0p-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 4.74302619264133348003801799876275e-29 0x1.e10000000000000000000000000p-95
ulp : 1924.0000
max.ulp : 4.0000
Failure: Test: cospi_upward (-0x3.fffffffffffffffcp+108)
Result:
is: 9.99999999999999999999999999852323e-01 0x1.ffffffffffffffffffffffe899bp-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 1.47673235656615530277812119019587e-28 0x1.766568e20369c00000000000000p-93
ulp : 5990.3382
max.ulp : 4.0000
Failure: Test: cospi_upward (-0x3.ffffffffffffffffffffffffffp+108)
Result:
is: 9.99999999999999999999999999952583e-01 0x1.fffffffffffffffffffffff87cbp-1
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
difference: 4.74136253815267677203679334037676e-29 0x1.e0d4cf1e9076600000000000000p-95
ulp : 1923.3252
max.ulp : 4.0000
Failure: cospi_upward (0xf.ffffffffffffbffffffffffffcp+1020): Exception "Invalid operation" set
Failure: cospi_upward (0xf.ffffffffffffbffffffffffffcp+1020): Exception "Overflow" set
Failure: cospi_upward (0xf.ffffffffffffbffffffffffffcp+1020): errno set to 33, expected 0 (unchanged)
Failure: Test: cospi_upward (0xf.ffffffffffffbffffffffffffcp+1020)
Result:
is: qNaN
should be: 1.00000000000000000000000000000000e+00 0x1.000000000000000000000000000p+0
2024-12-05 13:47:52 -03:00
0ae0af68d8
Implement C23 cospi
...
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the cospi functions (cos(pi*x)).
Tested for x86_64 and x86, and with build-many-glibcs.py.
2024-12-04 10:20:44 +00:00
Paul Zimmermann
Adhemerval Zanella
Joseph Myers