openmp

just playing around with openmp. Look at this code fragments: #pragma omp parallel { for( i =0;i<n;i++) { doing something } } and for( i =0;i<n;i++) { #pragma omp parallel { doing something } } Why is the first one a lot more slower (around the factor 5) than the second one? From theory I thought that the first one must be faster, because the parallel region is only created once and not n-times like the second? Can someone explain this to me? The code i want to parallelise has the following structure: for(i=0;i<n;i++) //wont be parallelizable { for(j=i+1;j<n;j++) //will be parallelized { doing

Near the start of my c++ application, my main thread uses OMP to parallelize several for loops. After the first parallelized for loop, I see that the threads used remain in existence for the duration of the application, and are reused for subsequent OMP for loops executed from the main thread, using the command (working in CentOS 7): for i in $(pgrep myApplication); do ps -mo pid,tid,fname,user,psr -p $i;done Later in my program, I launch a boost thread from the main thread, in which I parallelize a for loop using OMP. At this point, I see an entirely new set of threads are created, which has

I've been trying to increase the performance of my OpenMP solution which often has to deal with nested loops on arrays. Although I've managed to bring it down to 37 from 59 seconds of the serial implementation (on an ageing dual-core Intel T6600) I'm worried that cache synch gets lots of CPU attention (when the CPU should be solving my problem!). I've been fighting to set up the profiler so I haven't verified that claim but my question stands regardless. According to this lecture on load balancing: Instead of doing work, the CPUs are busy fighting over the only used cache line in the program.

I updated my Xcode yesterday (version 9.0) and since then I cannot compile my code with clang anymore. It works great with with apple native compiler, but gives a compilation error with clang from macports. I will explain with more details now... I usually use clang 4.0 because it has openmp support and I change in Xcode by creating a user-defined setting as in the following figure. Image with how to use clang 4.0 from macports in Xcode This has been working perfectly for some time until I updated to Xcode 9.0. Now, I get the following error from clang compiler: cannot specify -o when

I do not understand why such code is not vectorized with gcc 4.4.6 int MyFunc(const float *pfTab, float *pfResult, int iSize, int iIndex) { for (int i = 0; i < iSize; i++) pfResult[i] = pfResult[i] + pfTab[iIndex]; } note: not vectorized: unhandled data-ref However, if I write the following code int MyFunc(const float *pfTab, float *pfResult, int iSize, int iIndex) { float fTab = pfTab[iIndex]; for (int i = 0; i < iSize; i++) pfResult[i] = pfResult[i] + fTab; } gcc succeeds auto-vectorize this loop if I add omp directive int MyFunc(const float *pfTab, float *pfResult, int iSize, int iIndex) {

I wonder if there is any technique to create parallel sections in OpenMp using a for-loop. For example, instead of creating n different #pragma omp sections, I want to create them using an n-iteration for-loop with some changing parameters for each section. #pragma omp parallel sections { #pragma omp section { /* Executes in thread 1 */ } #pragma omp section { /* Executes in thread 2 */ } #pragma omp section { /* Executes in thread n */ } } Hristo Iliev With explicit OpenMP tasks: #pragma omp parallel { // Let only one thread create all tasks #pragma omp single nowait { for (int i = 0; i < num