multithreading - Memory leak when using OpenMP -

the below test case runs out of memory on 32 bit machines (throwing std::bad_alloc) in loop following "post mt section" message when openmp used, however, if #pragmas openmp commented out, code runs through completion fine, appears when memory allocated in parallel threads, not free correctly , run out of memory.

question whether there wrong memory allocation , deletion code below or bug in gcc v4.2.2 or openmp? tried gcc v4.3 , got same failure.

int main(int argc, char** argv) {     std::cout << "start " << std::endl;      {             std::vector<std::vector<int*> > nts(100);             #pragma omp parallel             {                     #pragma omp                     for(int begin = 0; begin < int(nts.size()); ++begin) {                             for(int = 0; < 1000000; ++i) {                                     nts[begin].push_back(new int(5));                             }                     }             }      std::cout << "  pre delete " << std::endl;             for(int begin = 0; begin < int(nts.size()); ++begin) {                     for(int j = 0; j < nts[begin].size(); ++j) {                             delete nts[begin][j];                     }             }     }     std::cout << "post mt section" << std::endl;     {             std::vector<std::vector<int*> > nts(100);             int begin, i;             try {               for(begin = 0; begin < int(nts.size()); ++begin) {                     for(i = 0; < 2000000; ++i) {                             nts[begin].push_back(new int(5));                     }               }             } catch (std::bad_alloc &e) {                     std::cout << e.what() << std::endl;                     std::cout << "begin: " << begin << " i: " << << std::endl;                     throw;             }             std::cout << "pre delete 1" << std::endl;              for(int begin = 0; begin < int(nts.size()); ++begin) {                     for(int j = 0; j < nts[begin].size(); ++j) {                             delete nts[begin][j];                     }             }     }      std::cout << "end of prog" << std::endl;      char c;     std::cin >> c;      return 0; } 

changing first openmp loop 1000000 2000000 cause same error. indicates out of memory problem openmp stack limit.

try setting openmp stack limit unlimit in bash with

ulimit -s unlimited 

you can change openmp environment variable omp_stacksize , setting 100mb or more.

update 1: change first loop to

{     std::vector<std::vector<int*> > nts(100);     #pragma omp schedule(static) ordered     for(int begin = 0; begin < int(nts.size()); ++begin) {         for(int = 0; < 2000000; ++i) {             nts[begin].push_back(new int(5));         }     }      std::cout << "  pre delete " << std::endl;     for(int begin = 0; begin < int(nts.size()); ++begin) {         for(int j = 0; j < nts[begin].size(); ++j) {             delete nts[begin][j]         }     } } 

then, memory error @ i=1574803 on main thread.

update 2: if using intel compiler, can add following top of code , solve problem (providing have enough memory overhead).

std::cout << "previous stack size " << kmp_get_stacksize_s() << std::endl; kmp_set_stacksize_s(1000000000); std::cout << "now stack size " << kmp_get_stacksize_s() << std::endl; 

update 3: completeness, mentioned member, if performing numerical computation, best preallocate in single new float[1000000] instead of using openmp 1000000 allocations. applies allocating objects well.