boost/compute/random/threefry_engine.hpp
//---------------------------------------------------------------------------// // Copyright (c) 2015 Muhammad Junaid Muzammil <mjunaidmuzammil@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_RANDOM_THREEFRY_HPP #define BOOST_COMPUTE_RANDOM_THREEFRY_HPP #include <algorithm> #include <boost/compute/types.hpp> #include <boost/compute/buffer.hpp> #include <boost/compute/kernel.hpp> #include <boost/compute/context.hpp> #include <boost/compute/program.hpp> #include <boost/compute/command_queue.hpp> #include <boost/compute/algorithm/transform.hpp> #include <boost/compute/detail/iterator_range_size.hpp> #include <boost/compute/utility/program_cache.hpp> #include <boost/compute/container/vector.hpp> #include <boost/compute/iterator/discard_iterator.hpp> namespace boost { namespace compute { /// \class threefry_engine /// \brief Threefry pseudorandom number generator. template<class T = uint_> class threefry_engine { public: static const size_t threads = 1024; typedef T result_type; /// Creates a new threefry_engine and seeds it with \p value. explicit threefry_engine(command_queue &queue) : m_context(queue.get_context()) { // setup program load_program(); } /// Creates a new threefry_engine object as a copy of \p other. threefry_engine(const threefry_engine<T> &other) : m_context(other.m_context), m_program(other.m_program) { } /// Copies \p other to \c *this. threefry_engine<T>& operator=(const threefry_engine<T> &other) { if(this != &other){ m_context = other.m_context; m_program = other.m_program; } return *this; } /// Destroys the threefry_engine object. ~threefry_engine() { } private: /// \internal_ void load_program() { boost::shared_ptr<program_cache> cache = program_cache::get_global_cache(m_context); std::string cache_key = std::string("threefry_engine_32x2"); // Copyright 2010-2012, D. E. Shaw Research. // All rights reserved. // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // * Redistributions of source code must retain the above copyright // notice, this list of conditions, and the following disclaimer. // * Redistributions in binary form must reproduce the above copyright // notice, this list of conditions, and the following disclaimer in the // documentation and/or other materials provided with the distribution. // * Neither the name of D. E. Shaw Research nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. const char source[] = "#define THREEFRY2x32_DEFAULT_ROUNDS 20\n" "#define SKEIN_KS_PARITY_32 0x1BD11BDA\n" "enum r123_enum_threefry32x2 {\n" " R_32x2_0_0=13,\n" " R_32x2_1_0=15,\n" " R_32x2_2_0=26,\n" " R_32x2_3_0= 6,\n" " R_32x2_4_0=17,\n" " R_32x2_5_0=29,\n" " R_32x2_6_0=16,\n" " R_32x2_7_0=24\n" "};\n" "static uint RotL_32(uint x, uint N)\n" "{\n" " return (x << (N & 31)) | (x >> ((32-N) & 31));\n" "}\n" "struct r123array2x32 {\n" " uint v[2];\n" "};\n" "typedef struct r123array2x32 threefry2x32_ctr_t;\n" "typedef struct r123array2x32 threefry2x32_key_t;\n" "threefry2x32_ctr_t threefry2x32_R(unsigned int Nrounds, threefry2x32_ctr_t in, threefry2x32_key_t k)\n" "{\n" " threefry2x32_ctr_t X;\n" " uint ks[3];\n" " uint i; \n" " ks[2] = SKEIN_KS_PARITY_32;\n" " for (i=0;i < 2; i++) {\n" " ks[i] = k.v[i];\n" " X.v[i] = in.v[i];\n" " ks[2] ^= k.v[i];\n" " }\n" " X.v[0] += ks[0]; X.v[1] += ks[1];\n" " if(Nrounds>0){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_0_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>1){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_1_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>2){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_2_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>3){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_3_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>3){\n" " X.v[0] += ks[1]; X.v[1] += ks[2];\n" " X.v[1] += 1;\n" " }\n" " if(Nrounds>4){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_4_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>5){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_5_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>6){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_6_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>7){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_7_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>7){\n" " X.v[0] += ks[2]; X.v[1] += ks[0];\n" " X.v[1] += 2;\n" " }\n" " if(Nrounds>8){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_0_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>9){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_1_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>10){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_2_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>11){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_3_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>11){\n" " X.v[0] += ks[0]; X.v[1] += ks[1];\n" " X.v[1] += 3;\n" " }\n" " if(Nrounds>12){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_4_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>13){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_5_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>14){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_6_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>15){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_7_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>15){\n" " X.v[0] += ks[1]; X.v[1] += ks[2];\n" " X.v[1] += 4;\n" " }\n" " if(Nrounds>16){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_0_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>17){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_1_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>18){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_2_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>19){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_3_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>19){\n" " X.v[0] += ks[2]; X.v[1] += ks[0];\n" " X.v[1] += 5;\n" " }\n" " if(Nrounds>20){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_4_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>21){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_5_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>22){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_6_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>23){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_7_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>23){\n" " X.v[0] += ks[0]; X.v[1] += ks[1];\n" " X.v[1] += 6;\n" " }\n" " if(Nrounds>24){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_0_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>25){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_1_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>26){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_2_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>27){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_3_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>27){\n" " X.v[0] += ks[1]; X.v[1] += ks[2];\n" " X.v[1] += 7;\n" " }\n" " if(Nrounds>28){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_4_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>29){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_5_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>30){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_6_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>31){ X.v[0] += X.v[1]; X.v[1] = RotL_32(X.v[1],R_32x2_7_0); X.v[1] ^= X.v[0]; }\n" " if(Nrounds>31){\n" " X.v[0] += ks[2]; X.v[1] += ks[0];\n" " X.v[1] += 8;\n" " }\n" " return X;\n" "}\n" "__kernel void generate_rng(__global uint *ctr, __global uint *key, const uint offset) {\n" " threefry2x32_ctr_t in;\n" " threefry2x32_key_t k;\n" " const uint i = get_global_id(0);\n" " in.v[0] = ctr[2 * (offset + i)];\n" " in.v[1] = ctr[2 * (offset + i) + 1];\n" " k.v[0] = key[2 * (offset + i)];\n" " k.v[1] = key[2 * (offset + i) + 1];\n" " in = threefry2x32_R(20, in, k);\n" " ctr[2 * (offset + i)] = in.v[0];\n" " ctr[2 * (offset + i) + 1] = in.v[1];\n" "}\n"; m_program = cache->get_or_build(cache_key, std::string(), source, m_context); } public: /// Generates Threefry random numbers using both the counter and key values, and then stores /// them to the range [\p first_ctr, \p last_ctr). template<class OutputIterator> void generate(OutputIterator first_ctr, OutputIterator last_ctr, OutputIterator first_key, OutputIterator last_key, command_queue &queue) { const size_t size_ctr = detail::iterator_range_size(first_ctr, last_ctr); const size_t size_key = detail::iterator_range_size(first_key, last_key); if(!size_ctr || !size_key || (size_ctr != size_key)) { return; } kernel rng_kernel = m_program.create_kernel("generate_rng"); rng_kernel.set_arg(0, first_ctr.get_buffer()); rng_kernel.set_arg(1, first_key.get_buffer()); size_t offset = 0; for(;;){ size_t count = 0; size_t size = size_ctr/2; if(size > threads){ count = (std::min)(static_cast<size_t>(threads), size - offset); } else { count = size; } rng_kernel.set_arg(2, static_cast<const uint_>(offset)); queue.enqueue_1d_range_kernel(rng_kernel, 0, count, 0); offset += count; if(offset >= size){ break; } } } template<class OutputIterator> void generate(OutputIterator first_ctr, OutputIterator last_ctr, command_queue &queue) { const size_t size_ctr = detail::iterator_range_size(first_ctr, last_ctr); if(!size_ctr) { return; } boost::compute::vector<uint_> vector_key(size_ctr, m_context); vector_key.assign(size_ctr, 0, queue); kernel rng_kernel = m_program.create_kernel("generate_rng"); rng_kernel.set_arg(0, first_ctr.get_buffer()); rng_kernel.set_arg(1, vector_key); size_t offset = 0; for(;;){ size_t count = 0; size_t size = size_ctr/2; if(size > threads){ count = (std::min)(static_cast<size_t>(threads), size - offset); } else { count = size; } rng_kernel.set_arg(2, static_cast<const uint_>(offset)); queue.enqueue_1d_range_kernel(rng_kernel, 0, count, 0); offset += count; if(offset >= size){ break; } } } private: context m_context; program m_program; }; } // end compute namespace } // end boost namespace #endif // BOOST_COMPUTE_RANDOM_THREEFRY_HPP