On Fri, 2015-03-27 at 11:13 +0000, Bruno Jimenez wrote: > On Wed, 2015-03-25 at 13:33 -0400, Jan Vesely wrote: > > tested on clover and intel ocl-sdk > > > > Signed-off-by: Jan Vesely <jan.ves...@rutgers.edu> > > --- > > generated_tests/gen_cl_math_builtins.py | 11 +++++++++++ > > 1 file changed, 11 insertions(+) > > > > diff --git a/generated_tests/gen_cl_math_builtins.py > > b/generated_tests/gen_cl_math_builtins.py > > index d86b53a..59b7f18 100644 > > --- a/generated_tests/gen_cl_math_builtins.py > > +++ b/generated_tests/gen_cl_math_builtins.py > > @@ -39,6 +39,7 @@ CLC_VERSION_MIN = { > > 'asinpi' : 10, > > 'atan' : 10, > > 'atan2' : 10, > > + 'atan2pi' : 10, > > 'atanh' : 10, > > 'atanpi': 10, > > 'cbrt' : 10, > > @@ -153,6 +154,16 @@ tests = { > > ], > > 'tolerance' : 6 > > }, > > + 'atan2pi' : { > > + 'arg_types' : [F, F, F], > > + 'function_type': 'ttt', > > + 'values' : [ > > + [atan2(0.0, 0.0)/pi, atan2(1.2345, 10.0)/pi, atan2(35671470.0, > > 0.1)/pi], # Result > > + [0.0, 1.2345, 35671470.0 > > ], # Arg0 > > + [0.0, 10.0, 0.1 ] > > # Arg1 > > + > Hi, > > If I remember correctly, atan2 also accepts negative imputs for x and y, > and returns the correct angle (remember that tan(x) == tan(x + pi)). > Should they be also tested?
I agree. This is just a simple adaptation of atan2 tests. There's also additional edge case behavior described in 7.5.1 that should be tested too. jan > > -Bruno > > > ], > > + 'tolerance' : 6 > > + }, > > 'atanh' : { > > 'arg_types' : [F, F], > > 'function_type': 'ttt', > > -- Jan Vesely <jan.ves...@rutgers.edu>
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Piglit mailing list Piglit@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/piglit