Add missing lambda wrappers for builtin OpenCL funcs #660
Conversation
jszuppe
force-pushed
the
pr_lambda_funcs
branch
from
90d106e to
b3d8e49
Compare
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(abs_diff) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(add_sat) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(hadd) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(hradd) |
There was a problem hiding this comment.
6.13.3 Integer functions, 4th function is:rhadd not hradd
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(exp) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(exp2) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(exp10) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(fabs) |
There was a problem hiding this comment.
6.13.2 Math functions, expm1 is missing
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(fmax) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(fmin) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(fmod) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(hypot) |
There was a problem hiding this comment.
6.13.2 Math functions: fract and frexp are missing
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(log1p) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(logb) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_TERNARY_FUNCTION(mad) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(nextafter) |
There was a problem hiding this comment.
6.13.2 Math functions: maxmag, minmag, modf & nan are missing.
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(pown) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(powr) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_BINARY_FUNCTION(remainder) | ||
| BOOST_COMPUTE_LAMBDA_WRAP_UNARY_FUNCTION_T(rint) |
There was a problem hiding this comment.
6.13.2 Math function: remquo is missing
There was a problem hiding this comment.
Excellent work Jakub. It was a much bigger change than I was expecting. I particularly liked that grouped and ordered the functions as in the OpenCL spec, it made reviewing much easier.
I only reviewed the changes for completeness against the OpenCL language spec. I've put comments where I think that functions are missing.
Also what about the optional half_* and native_* functions. Should they also be included?
|
Thanks for the comments, I'll look at them later today. |
jszuppe
force-pushed
the
pr_lambda_funcs
branch
from
b3d8e49 to
e8897de
Compare
|
I just need to add |
| /// vector_type_producer<int_, 1>::type == int_ | ||
| /// \endcode | ||
| template<class Scalar, size_t N> | ||
| struct vector_type_producer |
There was a problem hiding this comment.
Isn't this the same as make_vector_type<T, N> from <boost/compute/type_traits/make_vector_type.hpp>?
There was a problem hiding this comment.
Oh, yes, you're right. I couldn't find it. I'll remove it.
|
Thanks for adding the rest of these! Yeah, I don't think I skipped them on purpose, just never got around to adding them. |
jszuppe
force-pushed
the
pr_lambda_funcs
branch
from
e8897de to
0705750
Compare
|
I'll add |
jszuppe
force-pushed
the
pr_lambda_funcs
branch
from
0705750 to
e25bb7d
Compare
|
Merged! Thanks for adding all these! |
|
I've just tested the merged changes in the develop branch against the original failing code for issue #659 and I'm delighted to report that it's definitely fixed the issue for normalize and atan2. Thanks again Jakub. |
|
Glad, I could help. |
4 participants
I added a few missing lambda wrappers for OpenCL builtin function (see #659), but I noticed that a lot more is missing. @kylelutz, did you skip them on purpose or should I add all of them?