- Boost-users - lists.boost.org

[Parallel BGL] Poor scalability of dijkstra and delta stepping parallel graph algorithms
by Alexander Frolov 21 Sep '16

21 Sep '16

Hi! I am evaluating scalability on various graph algorithms and found that base parallel SSSP algorithms in PBGL does not scale well in my cases. The plots for rmat (scale=20 and 22) are attached. The source code for the benchmark are attached as well. The main issue that makes me think that I'm doing smth wrong is that execution time of delta stepping algorithm is _increasing_ with node number. I tried to check the code but has not found anything suspicious. Can anybody help me with this issue? Best, Alexander

1 0

[Interprocess]: using condition variables with readers-writer lock (named_sharable_mutex)
by Chris Evans 21 Sep '16

21 Sep '16

Hi, I am trying to use condition variables to signify updated data in a managed_shared_memory segment. I have one "writer" and multiple "readers" of the shared state, so I am using a readers-writer lock as follows: using SharedMutex = bi::named_sharable_mutex; using ReadLock = bi::sharable_lock<SharedMutex>; using WriteLock = bi::scoped_lock<SharedMutex>; using NewEntryCondition = bi::named_condition_any; // With a "writer" block as follows: { WriteLock w_lock( sh_mut_ ); // Update the shared memory object sh_cond_.notify_all(); } // And a "reader" block as follows (many processes call the following simultaneously): { ReadLock r_lock( sh_mut_ ); sh_cond_.wait( r_lock, []() { std::cout << "checking predicate..." << std::endl; // predicate logic goes here... }); // Access object in shared memory here } Unfortunately, although the code compiles, the reader processes are never awakened from wait() and block forever. According to the following thread from 2011: http://lists.boost.org/boost-users/2011/02/66053.php "POSIX does not support condition variables with read-write locks (although I think Windows Vista does) so Interprocess does not offer any support for this." At the time, it appears named_condition_any didn't yet exist. Has the status of this changed in the last 5 years? Is there a suggested way of implementing readers-writer locks with condition variables using boost::interprocess? Thanks! -- Chris Evans Systems Engineer Azure Summit Technology, Inc. 3050 Chain Bridge Road, Suite 600 Fairfax, VA 22030

2 2

[compute] extending meta_kernel functionality
by Oswin Krause 20 Sep '16

20 Sep '16

Hi, I am trying to use boost.compute to extend my linear algebra library to GPUs. so far it helped me a lot with vectors as i can directly use the iterator interface. Matrices are harder. for matrices i need to overload the operator matrix(i,j)-> return the element (i,j). For this i have to return an expression which then offers operator<< with meta_kernel, for example template<class T, class IndexExpr1, class IndexExpr2> inline meta_kernel& operator<<(boost::compute::detail::meta_kernel &kernel, const matrix_element_expr<T, IndexExpr1, IndexExpr2> &expr) { return kernel << kernel.get_buffer_identifier<T>(expr.m_buffer, expr.m_address_space) << '[' << expr.m_expr1 <<"*" <<expr.m_stride1<<"+"<<expr.m_expr2<< "*"<<expr.m_stride2 << ']'; } Unfortunately all of this is in namespace boost::compute::detail, making it a no-go area. Therefore two questions: 1. How stable is i?/can meta_kernel be moved into a normal namespace? 2. Is there a different way to achieve this? (Note that the code above is only an example. I could hack it in this case using the return type of buffer_iterator but not in all cases unfortunately...)

1 0

boost::context "main" stack
by Michael Steinberg 20 Sep '16

20 Sep '16

Hello, is there some way to temporarily run a function on the main stack? I played around a bit, but failed ultimately since the first stub function to be called is getting a "null" context which I assume to be only a valid "return-context". We're facing a problem where a dialog run modally inside a context runs webkit, which triggers garbage collection, which fails miserably since it assumes we're on the "native" stack. If it were possible to run something "on-top" of the main stack, this could maybe be circumvented. Thank you, Michael

2 2

Insertion/Use of boost::multi_index
by Ram 19 Sep '16

19 Sep '16

Hi All , I have created a boost::multi_index container for my use and it compiles successfully. I would like to add my pointer objects to it. boost::multi_index does not seem to have an insert method defined which just takes one arguement which is what I would like the container to store. Maybe I didnt understand how boost::multi_index is meant to be used. Can you please explain to me how to insert an object which I wish the store in the container defined? How is it intended to be used? The container I defined is , typedef boost::multi_index_ClassX< ClassX*, boost::multi_index::indexed_by< boost::multi_index::hashed_unique< boost::multi_index::const_mem_fun<ParentOfClassX, Int64, &ParentOfClassX::getKey1>, boost::multi_index::const_mem_fun<ParentOfClassX, Int64, &ParentOfClassX::getKey2>, > > > MultiIndexedClassX; I would like to do an equivalent of, MultiIndexedClassX sample; ClassX a; sample.insert(&a); <- Since I am interested in storing a *X and I would like it to be indexed by various parameters of X. Also please let me know how to, 1. Insert *X into the container 2. Look up by a) Value of ParentOfClassX::Key1 b) Value of ParentOfClassX::Key2 Please also let me know if I have some basic understanding wrong since I am getting confused a lot in trying to use boost::multi_index. Thanks in advance, Ram

2 10

Help Required, trouble while using boost::multi_index
by Ram 19 Sep '16

19 Sep '16

Hi, I am trying to create a ClassX using "boost::multi_index" and I was successfully able to create a ClassX like, typedef boost::multi_index_ClassX< ClassX*, boost::multi_index::indexed_by< boost::multi_index::hashed_unique< boost::multi_index::const_mem_fun<ParentOfClassX, Int64, &ParentOfClassX::getKey1>, boost::multi_index::const_mem_fun<ParentOfClassX, Int64, &ParentOfClassX::getKey2>, > > > MultiIndexedClassX; This compiles perfectly with no problems. I am writing the code and compiling with Visual Studio 2005 and have included, #include <boost/multi_index_container.hpp> #include "boost/multi_index/hashed_index_fwd.hpp" #include <boost/multi_index/mem_fun.hpp> and have defined namespaces, using namespace boost::multi_index; using boost::multi_index::multi_index_container; using boost::multi_index::hashed_unique; using boost::multi_index::indexed_by; But when I create an object of that type like, MultiIndexedClassX sample; I get a big list compilation errors like, *c:\code\common\lib\boost\boost\multi_index\detail\node_type.hpp(44) : error C2903: 'node_class' : symbol is neither a class template nor a function template* * c:\code\common\lib\boost\boost\mpl\aux_\preprocessed\plain\apply_wrap.hpp(49) : see reference to class template instantiation 'boost::multi_index::detail::index_node_applier::apply<IndexSpecifierIterator,Super>' being compiled* * with* * [* * IndexSpecifierIterator=boost::mpl::v_iter<boost::mpl::vector1<boost::multi_index::hashed_unique<boost::multi_index::const_mem_fun<ParentOfClassX,GKey,ParentOfClassX::getKey1>,boost::multi_index::const_mem_fun<ParentOfClassX,Vs64,ParentOfClassX::getKey2>>>,0>,* * Super=boost::multi_index::detail::index_node_base<ClassX *,std::allocator<ClassX *>>* * ]* * c:\code\common\lib\boost\boost\mpl\aux_\preprocessed\plain\bind.hpp(207) : see reference to class template instantiation 'boost::mpl::apply_wrap2<F,T1,T2>' being compiled* * with* * [* * F=boost::multi_index::detail::index_node_applier,* * T1=boost::mpl::v_iter<boost::mpl::vector1<boost::multi_index::hashed_unique<boost::multi_index::const_mem_fun<ParentOfClassX,GKey,ParentOfClassX::getKey1>,boost::multi_index::const_mem_fun<ParentOfClassX,Vs64,ParentOfClassX::getKey2>>>,0>,* * T2=boost::multi_index::detail::index_node_base<ClassX *,std::allocator<ClassX *>>* * ]* * c:\code\common\lib\boost\boost\mpl\aux_\preprocessed\plain\apply_wrap.hpp(49) : see reference to class template instantiation 'boost::mpl::bind2<F,T1,T2>::apply<U1,U2>' being compiled* * with* * [* * F=boost::multi_index::detail::index_node_applier,* * T1=boost::mpl::_2,* * T2=boost::mpl::_1,* * U1=boost::multi_index::detail::index_node_base<ClassX *,std::allocator<ClassX *>>,* * U2=boost::mpl::v_iter<boost::mpl::vector1<boost::multi_index::hashed_unique<boost::multi_index::const_mem_fun<ParentOfClassX,GKey,ParentOfClassX::getKey1>,boost::multi_index::const_mem_fun<ParentOfClassX,Vs64,ParentOfClassX::getKey2>>>,0>* * ]* * c:\code\common\lib\boost\boost\mpl\aux_\preprocessed\plain\apply.hpp(63) : see reference to class template instantiation 'boost::mpl::apply_wrap2<F,T1,T2>' being compiled* * with* * [* * F=boost::mpl::bind2<boost::multi_index::detail::index_node_applier,boost::mpl::_2,boost::mpl::_1>,* * T1=boost::multi_index::detail::index_node_base<ClassX *,std::allocator<ClassX *>>,* * T2=boost::mpl::v_iter<boost::mpl::vector1<boost::multi_index::hashed_unique<boost::multi_index::const_mem_fun<ParentOfClassX,GKey,ParentOfClassX::getKey1>,boost::multi_index::const_mem_fun<ParentOfClassX,Vs64,ParentOfClassX::getKey2>>>,0>* * ]* * c:\code\common\lib\boost\boost\mpl\aux_\preprocessed\plain\reverse_iter_fold_impl.hpp(58) : see reference to class template instantiation 'boost::mpl::apply2<F,T1,T2>' being compiled* * with* * [* * F=boost::mpl::bind2<boost::multi_index::detail::index_node_applier,boost::mpl::_2,boost::mpl::_1>,* * T1=boost::multi_index::detail::index_node_base<ClassX *,std::allocator<ClassX *>>,* * T2=boost::mpl::v_iter<boost::mpl::vector1<boost::multi_index::hashed_unique<boost::multi_index::const_mem_fun<ParentOfClassX,GKey,ParentOfClassX::getKey1>,boost::multi_index::const_mem_fun<ParentOfClassX,Vs64,ParentOfClassX::getKey2>>>,0>* * ]* Thanks, Ram

2 5

[Graph] Is there an active maintainer for Boost.Graph?
by Rohan Shah 19 Sep '16

19 Sep '16

Hi all. I'm wondering whether Boost.Graph currently has an active maintainer. I only ask because there is a fairly active github repository ( https://github.com/boostorg/graph) but it seems that important commits fail to make it into the main release on the website. For example, I would consider this (https://github.com/boostorg/graph/commit/ 8ece8fac15560cf38cf3ffbb1d56708ef8e586a2) commit to be fairly critical - Without it part of depth_first_search seems to be somewhat broken, and that's a core algorithm. And that commit is from 2014! A conversation on #boost IRC suggested I mention this problem here. Note: I am not volunteering myself to help fix this - I'm simply not qualified for it. Sorry.

1 0

[boost] [bjam] combinations of options
by Lorenzo Caminiti 19 Sep '16

19 Sep '16

Hello all, When I use bjam the following will compile my code with combinations of MSVC, GCC, and CLang for both shared and static linking: $ bjam toolset=msvc,gcc,clang link=shared,static # (1) Is there a way to define a variant, target, or something else in the Jamfiles so that I can shortcut the above by typing just the variant target at the command line: $ bjam all_toolset_link_combinations # equivalent to (1) above I am asking this just for convenience because I have a feature like toolset but with 30+ values so it'd be nice to shortcut it to a single target on the bjam command line. Thank you. --Lorenzo

2 4

[test] Init framework before dataset generation
by Florian Lindner 16 Sep '16

16 Sep '16

Hello, I try to use boost test with data test cases: std::vector<float> fibonacci() { std::cout << "Data Set" << std::endl; [...] } BOOST_DATA_TEST_CASE(datatest, boost::unit_test::data::make(fibonacci()), mapping) { std::cout << "Datatype" << typeid(mapping).name() << std::endl; } The problem of that is, that fibonacci is initialized before anything else. The order of initializations is: Data Set main init_unit_test Global Fixture That works fine here, but my fibonacci uses MPI which needs to initialized by MPI_Init. I want to do that initialization in the global fixture or int main or init_unit_tests, once for the entire test runner. Is there anything I can do about that and belay initialization of Data Set after one of the others? Thanks, Florian

2 2

[release] Boost 1.62.0. Beta 2
by Rene Rivera 14 Sep '16

14 Sep '16

The packages for 1.62.0 beta 2 are now available: <https://sourceforge.net/projects/boost/files/boost/1.62.0.beta.2/> The SHA256 checksums are as follows: 95feae3620b3e604e93f14d97195fc04a4b67063692859af8df19657d41b82d5 boost_1_62_0_b2.7z 27bc9b328ff73d7dbf80cd9d4b5f79b0878221e5d8e4b6c47444b135edc1080a boost_1_62_0_b2.tar.bz2 3939d5efa683449e87b374d7e92e3eeb6490063ddb2877919f0fad26e3a0f53e boost_1_62_0_b2.tar.gz cff9ab5db0928244c4c836cf3da5140e6e93db7756e2791ba99a21104ad3aa72 boost_1_62_0_b2.zip As always, the release managers would appreciate it if you download the archive of your choice and give building and using it a try. Please report both success and failure, and anything else that is noteworthy. Thanks! -- The release managers -- -- Rene Rivera -- Grafik - Don't Assume Anything -- Robot Dreams - http://robot-dreams.net -- rrivera/acm.org (msn) - grafikrobot/aim,yahoo,skype,efnet,gmail

3 2