1<html> 2<head> 3<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> 4<title>Frequently Asked Questions</title> 5<link rel="stylesheet" href="../../../../../doc/src/boostbook.css" type="text/css"> 6<meta name="generator" content="DocBook XSL Stylesheets V1.79.1"> 7<link rel="home" href="../index.html" title="Chapter 1. Boost.Compute"> 8<link rel="up" href="../index.html" title="Chapter 1. Boost.Compute"> 9<link rel="prev" href="performance.html" title="Performance"> 10</head> 11<body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"> 12<table cellpadding="2" width="100%"><tr> 13<td valign="top"><img alt="Boost C++ Libraries" width="277" height="86" src="../../../../../boost.png"></td> 14<td align="center"><a href="../../../../../index.html">Home</a></td> 15<td align="center"><a href="../../../../../libs/libraries.htm">Libraries</a></td> 16<td align="center"><a href="http://www.boost.org/users/people.html">People</a></td> 17<td align="center"><a href="http://www.boost.org/users/faq.html">FAQ</a></td> 18<td align="center"><a href="../../../../../more/index.htm">More</a></td> 19</tr></table> 20<hr> 21<div class="spirit-nav"> 22<a accesskey="p" href="performance.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a> 23</div> 24<div class="section"> 25<div class="titlepage"><div><div><h2 class="title" style="clear: both"> 26<a name="boost_compute.faq"></a><a class="link" href="faq.html" title="Frequently Asked Questions">Frequently Asked Questions</a> 27</h2></div></div></div> 28<h4> 29<a name="boost_compute.faq.h0"></a> 30 <span class="phrase"><a name="boost_compute.faq.how_do_i_report_a_bug__issue__or_feature_request_"></a></span><a class="link" href="faq.html#boost_compute.faq.how_do_i_report_a_bug__issue__or_feature_request_">How 31 do I report a bug, issue, or feature request?</a> 32 </h4> 33<p> 34 Please submit an issue on the GitHub issue tracker at <a href="https://github.com/boostorg/compute/issues" target="_top">https://github.com/boostorg/compute/issues</a>. 35 </p> 36<h4> 37<a name="boost_compute.faq.h1"></a> 38 <span class="phrase"><a name="boost_compute.faq.where_can_i_find_more_documentation_"></a></span><a class="link" href="faq.html#boost_compute.faq.where_can_i_find_more_documentation_">Where can 39 I find more documentation?</a> 40 </h4> 41<div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "> 42<li class="listitem"> 43 The main documentation is here: <a href="http://boostorg.github.io/compute/" target="_top">http://boostorg.github.io/compute/</a> 44 </li> 45<li class="listitem"> 46 The README is here: <a href="https://github.com/boostorg/compute/blob/master/README.md" target="_top">https://github.com/boostorg/compute/blob/master/README.md</a> 47 </li> 48<li class="listitem"> 49 The wiki is here: <a href="https://github.com/boostorg/compute/wiki" target="_top">https://github.com/boostorg/compute/wiki</a> 50 </li> 51<li class="listitem"> 52 The contributor guide is here: <a href="https://github.com/boostorg/compute/blob/master/CONTRIBUTING.md" target="_top">https://github.com/boostorg/compute/blob/master/CONTRIBUTING.md</a> 53 </li> 54<li class="listitem"> 55 The reference is here: <a href="http://boostorg.github.io/compute/compute/reference.html" target="_top">http://boostorg.github.io/compute/compute/reference.html</a> 56 </li> 57</ul></div> 58<h4> 59<a name="boost_compute.faq.h2"></a> 60 <span class="phrase"><a name="boost_compute.faq.where_is_the_best_place_to_ask_questions_about_the_library_"></a></span><a class="link" href="faq.html#boost_compute.faq.where_is_the_best_place_to_ask_questions_about_the_library_">Where 61 is the best place to ask questions about the library?</a> 62 </h4> 63<p> 64 The mailing list at <a href="https://groups.google.com/forum/#!forum/boost-compute" target="_top">https://groups.google.com/forum/#!forum/boost-compute</a>. 65 </p> 66<h4> 67<a name="boost_compute.faq.h3"></a> 68 <span class="phrase"><a name="boost_compute.faq.what_compute_devices__e_g__gpus__are_supported_"></a></span><a class="link" href="faq.html#boost_compute.faq.what_compute_devices__e_g__gpus__are_supported_">What 69 compute devices (e.g. GPUs) are supported?</a> 70 </h4> 71<p> 72 Any device which implements the OpenCL standard is supported. This includes 73 GPUs from NVIDIA, AMD, and Intel as well as CPUs from AMD and Intel and other 74 accelerator cards such as the Xeon Phi. 75 </p> 76<h4> 77<a name="boost_compute.faq.h4"></a> 78 <span class="phrase"><a name="boost_compute.faq.can_you_compare_boost_compute_to_other_gpgpu_libraries_such_as_thrust__bolt_and_vexcl_"></a></span><a class="link" href="faq.html#boost_compute.faq.can_you_compare_boost_compute_to_other_gpgpu_libraries_such_as_thrust__bolt_and_vexcl_">Can 79 you compare Boost.Compute to other GPGPU libraries such as Thrust, Bolt and 80 VexCL?</a> 81 </h4> 82<p> 83 Thrust implements a C++ STL-like API for GPUs and CPUs. It is built with multiple 84 backends. NVIDIA GPUs use the CUDA backend and multi-core CPUs can use the 85 Intel TBB or OpenMP backends. However, thrust will not work with AMD graphics 86 cards or other lesser-known accelerators. I feel Boost.Compute is superior 87 in that it uses the vendor-neutral OpenCL library to achieve portability across 88 all types of compute devices. 89 </p> 90<p> 91 Bolt is an AMD specific C++ wrapper around the OpenCL API which extends the 92 C99-based OpenCL language to support C++ features (most notably templates). 93 It is similar to NVIDIA's Thrust library and shares the same failure, lack 94 of portability. 95 </p> 96<p> 97 VexCL is an expression-template based linear-algebra library for OpenCL. The 98 aims and scope are a bit different from the Boost Compute library. VexCL is 99 closer in nature to the Eigen library while Boost.Compute is closer to the 100 C++ standard library. I don't feel that Boost.Compute really fills the same 101 role as VexCL. In fact, the recent versions of VexCL allow to use Boost.Compute 102 as one of the backends, which makes the interaction between the two libraries 103 a breeze. 104 </p> 105<p> 106 Also see this StackOverflow question: <a href="http://stackoverflow.com/questions/20154179/differences-between-vexcl-thrust-and-boost-compute" target="_top">http://stackoverflow.com/questions/20154179/differences-between-vexcl-thrust-and-boost-compute</a> 107 </p> 108<h4> 109<a name="boost_compute.faq.h5"></a> 110 <span class="phrase"><a name="boost_compute.faq.why_not_write_just_write_a_new_opencl_back_end_for_thrust_"></a></span><a class="link" href="faq.html#boost_compute.faq.why_not_write_just_write_a_new_opencl_back_end_for_thrust_">Why 111 not write just write a new OpenCL back-end for Thrust?</a> 112 </h4> 113<p> 114 It would not be possible to provide the same API that Thrust expects for OpenCL. 115 The fundamental reason is that functions/functors passed to Thrust algorithms 116 are actual compiled C++ functions whereas for Boost.Compute these form expression 117 objects which are then translated into C99 code which is then compiled for 118 OpenCL. 119 </p> 120<h4> 121<a name="boost_compute.faq.h6"></a> 122 <span class="phrase"><a name="boost_compute.faq.why_not_target_cuda_and_or_support_multiple_back_ends_"></a></span><a class="link" href="faq.html#boost_compute.faq.why_not_target_cuda_and_or_support_multiple_back_ends_">Why 123 not target CUDA and/or support multiple back-ends?</a> 124 </h4> 125<p> 126 CUDA and OpenCL are two very different technologies. OpenCL works by compiling 127 C99 code at run-time to generate kernel objects which can then be executed 128 on the GPU. CUDA, on the other hand, works by compiling its kernels using a 129 special compiler (nvcc) which then produces binaries which can executed on 130 the GPU. 131 </p> 132<p> 133 OpenCL already has multiple implementations which allow it to be used on a 134 variety of platforms (e.g. NVIDIA GPUs, Intel CPUs, etc.). I feel that adding 135 another abstraction level within Boost.Compute would only complicate and bloat 136 the library. 137 </p> 138<h4> 139<a name="boost_compute.faq.h7"></a> 140 <span class="phrase"><a name="boost_compute.faq.is_it_possible_to_use_ordinary_c___functions_functors_or_c__11__lambdas_with_boost_compute_"></a></span><a class="link" href="faq.html#boost_compute.faq.is_it_possible_to_use_ordinary_c___functions_functors_or_c__11__lambdas_with_boost_compute_">Is 141 it possible to use ordinary C++ functions/functors or C++11 lambdas with Boost.Compute?</a> 142 </h4> 143<p> 144 Unfortunately no. OpenCL relies on having C99 source code available at run-time 145 in order to execute code on the GPU. Thus compiled C++ functions or C++11 lambdas 146 cannot simply be passed to the OpenCL environment to be executed on the GPU. 147 </p> 148<p> 149 This is the reason why I wrote the Boost.Compute lambda library. Basically 150 it takes C++ lambda expressions (e.g. _1 * sqrt(_1) + 4) and transforms them 151 into C99 source code fragments (e.g. “input[i] * sqrt(input[i]) + 4)”) 152 which are then passed to the Boost.Compute STL-style algorithms for execution. 153 While not perfect, it allows the user to write code closer to C++ that still 154 can be executed through OpenCL. 155 </p> 156<p> 157 Also check out the BOOST_COMPUTE_FUNCTION() macro which allows OpenCL functions 158 to be defined inline with C++ code. An example can be found in the monte_carlo 159 example code. 160 </p> 161<h4> 162<a name="boost_compute.faq.h8"></a> 163 <span class="phrase"><a name="boost_compute.faq.what_is_the_command_queue_argument_that_appears_in_all_of_the_algorithms_"></a></span><a class="link" href="faq.html#boost_compute.faq.what_is_the_command_queue_argument_that_appears_in_all_of_the_algorithms_">What 164 is the command_queue argument that appears in all of the algorithms?</a> 165 </h4> 166<p> 167 Command queues specify the context and device for the algorithm's execution. 168 For all of the standard algorithms the command_queue parameter is optional. 169 If not provided, a default command_queue will be created for the default GPU 170 device and the algorithm will be executed there. 171 </p> 172<h4> 173<a name="boost_compute.faq.h9"></a> 174 <span class="phrase"><a name="boost_compute.faq.how_can_i_print_out_the_contents_of_a_buffer_vector_on_the_gpu_"></a></span><a class="link" href="faq.html#boost_compute.faq.how_can_i_print_out_the_contents_of_a_buffer_vector_on_the_gpu_">How 175 can I print out the contents of a buffer/vector on the GPU?</a> 176 </h4> 177<p> 178 This can be accompilshed easily using the generic boost::compute::copy() algorithm 179 along with std::ostream_iterator<T>. For example: 180 </p> 181<p> 182</p> 183<pre class="programlisting"><span class="identifier">std</span><span class="special">::</span><span class="identifier">cout</span> <span class="special"><<</span> <span class="string">"vector: [ "</span><span class="special">;</span> 184<span class="identifier">boost</span><span class="special">::</span><span class="identifier">compute</span><span class="special">::</span><span class="identifier">copy</span><span class="special">(</span> 185 <span class="identifier">vector</span><span class="special">.</span><span class="identifier">begin</span><span class="special">(),</span> <span class="identifier">vector</span><span class="special">.</span><span class="identifier">end</span><span class="special">(),</span> 186 <span class="identifier">std</span><span class="special">::</span><span class="identifier">ostream_iterator</span><span class="special"><</span><span class="keyword">int</span><span class="special">>(</span><span class="identifier">std</span><span class="special">::</span><span class="identifier">cout</span><span class="special">,</span> <span class="string">", "</span><span class="special">),</span> 187 <span class="identifier">queue</span> 188<span class="special">);</span> 189<span class="identifier">std</span><span class="special">::</span><span class="identifier">cout</span> <span class="special"><<</span> <span class="string">"]"</span> <span class="special"><<</span> <span class="identifier">std</span><span class="special">::</span><span class="identifier">endl</span><span class="special">;</span> 190</pre> 191<p> 192 </p> 193<h4> 194<a name="boost_compute.faq.h10"></a> 195 <span class="phrase"><a name="boost_compute.faq.does_boost_compute_support_zero_copy_memory_"></a></span><a class="link" href="faq.html#boost_compute.faq.does_boost_compute_support_zero_copy_memory_">Does 196 Boost.Compute support zero-copy memory?</a> 197 </h4> 198<p> 199 Zero-copy memory allows OpenCL kernels to directly operate on regions of host 200 memory (if supported by the platform). 201 </p> 202<p> 203 Boost.Compute supports zero-copy memory in multiple ways. The low-level interface 204 is provided by allocating <code class="computeroutput">buffer</code> 205 objects with the <code class="computeroutput"><span class="identifier">CL_MEM_USE_HOST_PTR</span></code> 206 flag. The high-level interface is provided by the <code class="computeroutput"><a class="link" href="../boost/compute/mapped_view.html" title="Class template mapped_view">mapped_view<T></a></code> 207 class which provides a std::vector-like interface to a region of host-memory 208 and can be used directly with all of the Boost.Compute algorithms. 209 </p> 210<h4> 211<a name="boost_compute.faq.h11"></a> 212 <span class="phrase"><a name="boost_compute.faq.is_boost_compute_thread_safe_"></a></span><a class="link" href="faq.html#boost_compute.faq.is_boost_compute_thread_safe_">Is 213 Boost.Compute thread-safe?</a> 214 </h4> 215<p> 216 The low-level Boost.Compute APIs offer the same thread-safety guarantees as 217 the underyling OpenCL library implementation. However, the high-level APIs 218 make use of a few global static objects for features such as automatic program 219 caching which makes them not thread-safe by default. 220 </p> 221<p> 222 To compile Boost.Compute in thread-safe mode define <code class="computeroutput"><span class="identifier">BOOST_COMPUTE_THREAD_SAFE</span></code> 223 before including any of the Boost.Compute headers. By default this will require 224 linking your application/library with the Boost.Thread library. 225 </p> 226<h4> 227<a name="boost_compute.faq.h12"></a> 228 <span class="phrase"><a name="boost_compute.faq.what_applications_libraries_use_boost_compute_"></a></span><a class="link" href="faq.html#boost_compute.faq.what_applications_libraries_use_boost_compute_">What 229 applications/libraries use Boost.Compute?</a> 230 </h4> 231<p> 232 Boost.Compute is used by a number of open-source libraries and applications 233 including: 234 </p> 235<div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "> 236<li class="listitem"> 237 ArrayFire (<a href="http://arrayfire.com" target="_top">http://arrayfire.com</a>) 238 </li> 239<li class="listitem"> 240 Ceemple (<a href="http://www.ceemple.com" target="_top">http://www.ceemple.com</a>) 241 </li> 242<li class="listitem"> 243 Odeint (<a href="http://headmyshoulder.github.io/odeint-v2" target="_top">http://headmyshoulder.github.io/odeint-v2</a>) 244 </li> 245<li class="listitem"> 246 VexCL (<a href="https://github.com/ddemidov/vexcl" target="_top">https://github.com/ddemidov/vexcl</a>) 247 </li> 248</ul></div> 249<p> 250 If you use Boost.Compute in your project and would like it to be listed here 251 please send an email to Kyle Lutz (kyle.r.lutz@gmail.com). 252 </p> 253<h4> 254<a name="boost_compute.faq.h13"></a> 255 <span class="phrase"><a name="boost_compute.faq.how_can_i_contribute_"></a></span><a class="link" href="faq.html#boost_compute.faq.how_can_i_contribute_">How 256 can I contribute?</a> 257 </h4> 258<p> 259 We are actively seeking additional C++ developers with experience in GPGPU 260 and parallel-computing. 261 </p> 262<p> 263 Please send an email to Kyle Lutz (kyle.r.lutz@gmail.com) for more information. 264 </p> 265<p> 266 Also see the <a href="https://github.com/boostorg/compute/blob/master/CONTRIBUTING.md" target="_top">contributor 267 guide</a> and check out the list of issues at: <a href="https://github.com/boostorg/compute/issues" target="_top">https://github.com/boostorg/compute/issues</a>. 268 </p> 269</div> 270<table xmlns:rev="http://www.cs.rpi.edu/~gregod/boost/tools/doc/revision" width="100%"><tr> 271<td align="left"></td> 272<td align="right"><div class="copyright-footer">Copyright © 2013, 2014 Kyle Lutz<p> 273 Distributed under the Boost Software License, Version 1.0. (See accompanying 274 file LICENSE_1_0.txt or copy at <a href="http://www.boost.org/LICENSE_1_0.txt" target="_top">http://www.boost.org/LICENSE_1_0.txt</a>) 275 </p> 276</div></td> 277</tr></table> 278<hr> 279<div class="spirit-nav"> 280<a accesskey="p" href="performance.html"><img src="../../../../../doc/src/images/prev.png" alt="Prev"></a><a accesskey="u" href="../index.html"><img src="../../../../../doc/src/images/up.png" alt="Up"></a><a accesskey="h" href="../index.html"><img src="../../../../../doc/src/images/home.png" alt="Home"></a> 281</div> 282</body> 283</html> 284
