FM / detail

/usr/include/boost/compute/detail

Name	Size	Mode	Actions
assert_cl_success.hpp	851	0644	edit dl rm
buffer_value.hpp	4062	0644	edit dl rm
cl_versions.hpp	3130	0644	edit dl rm
device_ptr.hpp	5394	0644	edit dl rm
diagnostic.hpp	5196	0644	edit dl rm
duration.hpp	1641	0644	edit dl rm
getenv.hpp	936	0644	edit dl rm
get_object_info.hpp	8316	0644	edit dl rm
global_static.hpp	1437	0644	edit dl rm
is_buffer_iterator.hpp	960	0644	edit dl rm
is_contiguous_iterator.hpp	3720	0644	edit dl rm
iterator_plus_distance.hpp	1662	0644	edit dl rm
iterator_range_size.hpp	1414	0644	edit dl rm
iterator_traits.hpp	1158	0644	edit dl rm
literal.hpp	1444	0644	edit dl rm
lru_cache.hpp	3467	0644	edit dl rm
meta_kernel.hpp	32043	0644	edit dl rm
mpl_vector_to_tuple.hpp	2270	0644	edit dl rm
nvidia_compute_capability.hpp	2073	0644	edit dl rm
parameter_cache.hpp	7191	0644	edit dl rm
path.hpp	2100	0644	edit dl rm
print_range.hpp	2588	0644	edit dl rm
read_write_single_value.hpp	2542	0644	edit dl rm
sha1.hpp	1545	0644	edit dl rm
variadic_macros.hpp	1836	0644	edit dl rm
vendor.hpp	1491	0644	edit dl rm
work_size.hpp	1242	0644	edit dl rm

Edit: /usr/include/boost/compute/detail/work_size.hpp (1242B)

//---------------------------------------------------------------------------//
// Copyright (c) 2013-2014 Kyle Lutz <kyle.r.lutz@gmail.com>
//
// Distributed under the Boost Software License, Version 1.0
// See accompanying file LICENSE_1_0.txt or copy at
// http://www.boost.org/LICENSE_1_0.txt
//
// See http://boostorg.github.com/compute for more information.
//---------------------------------------------------------------------------//

#ifndef BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP
#define BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP

#include <cmath>

namespace boost {
namespace compute {
namespace detail {

// Given a total number of values (count), a number of values to
// process per thread (vtp), and a number of threads to execute per
// block (tpb), this function returns the global work size to be
// passed to clEnqueueNDRangeKernel() for a 1D algorithm.
inline size_t calculate_work_size(size_t count, size_t vpt, size_t tpb)
{
    size_t work_size = static_cast<size_t>(std::ceil(float(count) / vpt));
    if(work_size % tpb != 0){
        work_size += tpb - work_size % tpb;
    }
    return work_size;
}

} // end detail namespace
} // end compute namespace
} // end boost namespace

#endif // BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP

//---------------------------------------------------------------------------// // Copyright (c) 2013-2014 Kyle Lutz // // Distributed under the Boost Software License, Version 1.0 // See accompanying file LICENSE_1_0.txt or copy at // http://www.boost.org/LICENSE_1_0.txt // // See http://boostorg.github.com/compute for more information. //---------------------------------------------------------------------------// #ifndef BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP #define BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP #include namespace boost { namespace compute { namespace detail { // Given a total number of values (count), a number of values to // process per thread (vtp), and a number of threads to execute per // block (tpb), this function returns the global work size to be // passed to clEnqueueNDRangeKernel() for a 1D algorithm. inline size_t calculate_work_size(size_t count, size_t vpt, size_t tpb) { size_t work_size = static_cast(std::ceil(float(count) / vpt)); if(work_size % tpb != 0){ work_size += tpb - work_size % tpb; } return work_size; } } // end detail namespace } // end compute namespace } // end boost namespace #endif // BOOST_COMPUTE_DETAIL_WORK_SIZE_HPP