c++ - Can a thread-local copy of select elements be created of a shared 2D array in a parallel region? (Shared, private, barrier: OPenMP) -


i have 2-d grid of nxn elements. in 1 iteration, i'm calculating value of 1 element averaging values of neighbors. is:

    for(int i=0;i<n;i++)         for(int j=0;j<n;j++)             grid[i][j] = (grid[i-1][j] + grid[i][j-1] + grid[i+1][j] + grid[i][j+1])/4.0;

and need run above nested loop iter number of iterations. need following:

  1. i need threads calculate average, wait till threads have finished calculating , update grid in 1 go.
  2. the loop iter iterations run sequentially, during every iteration, value of grid[i][j] every i , j should calculated in parallel.

in order have following ideas , questions:

  1. maybe make grid shared , put copy of select 4 elements of grid needed calculating grid[i][j] making 4 elements private thread. (basically grid shared threads, there local copy of 4 iteration-specific elements in every thread too.) is possible?
  2. would barrier in fact needed threads finish , start onto next iteration?

i'm new openmp way of thinking , i'm utterly lost in simple problem. i'd grateful if resolve confusion.

  1. in practice, you'd want have (much) fewer threads grid points, each thread calculating whole bunch of points (for example, 1 row). there overhead associated starting openmp (or other kind of) threads, , program memory-bound rather cpu-bound anyway. starting thread per grid point defeat whole purpose of parallelizing computation. hence, idea #1 not recommended (i not quite sure understood correctly though; maybe not proposing).

  2. i recommend (also pointed out others in op comments) allocate twice memory needed store grid values , use 2 pointers swapped between iterations: 1 points memory holding previous iteration values read only, other 1 new iteration values write-only. note swap pointers, not copy memory. after iteration done, can copy final result desired location.

  3. yes, need synchronize threads between iterations, in openmp done implicitly opening parallel region within iteration loop (there implicit barrier @ end of parallel region):

    for (int iter = 0; iter < niter; ++iter) {     #pragma omp parallel     {         // range of points current thread         // loop on thread's points , apply stencil     } }

    or, using parallel for construct:

    const int np = n*n; (int iter = 0; iter < niter; ++iter) {     #pragma omp parallel     (int ip = 0; ip < np; ++ip)     {         const int = ip / n;         const int j = ip % n;         // apply stencil [i,j]     } }

    the second version auto-distribute work evenly between available threads, want. in first have manually.


Comments

Post a Comment

Popular posts from this blog

Is there a better way to structure post methods in Class Based Views -

i'm using class based views in django , never specify url in ajax calls , omit action parameter in form because know these requests go through post method of class based view. notice view - profile page make many requests depending on user updating on page - post method become convoluted , if/else mess example: def post(self, request, *args, **kwargs): # handle user changing available date if request.post.get('availabledate') != none: self.updateavailabledate(request) return jsonresponse({'result':'success'}) if request.post.get('newprojectname') != none: creative_user = creativeuserprofile.objects.get(id = request.user.id) project = project.create_project(creative_user, request.post['newprojectname']) project.save() return jsonresponse({'projectid': project.id }) if request.post.get('projectimage') != none , request.files['file']: proj
Read more

performance - Why is XCHG reg, reg a 3 micro-op instruction on modern Intel architectures? -

i'm doing micro-optimization on performance critical part of code , came across sequence of instructions (in at&t syntax): add %rax, %rbx mov %rdx, %rax mov %rbx, %rdx i thought had use case xchg allow me shave instruction , write: add %rbx, %rax xchg %rax, %rdx however, dimay found agner fog's instruction tables , xchg 3 micro-op instruction 2 cycle latency on sandy bridge, ivy bridge, broadwell, haswell , skylake. 3 whole micro-ops , 2 cycles of latency! 3 micro-ops throws off 4-1-1-1 cadence , 2 cycle latency makes worse original in best case since last 2 instructions in original might execute in parallel. now... cpu might breaking instruction micro-ops equivalent to: mov %rax, %tmp mov %rdx, %rax mov %tmp, %rdx where tmp anonymous internal register , suppose last 2 micro-ops run in parallel latency 2 cycles. given register renaming occurs on these micro-architectures, though, doesn't make sense me done way. why wouldn't register renamer
Read more

jquery - Responsive Navbar with Sub Navbar -

good day, forgive me nav vocabulary not best. if question confusing apologize. i'm working on navigation bar needs centered on page @ 768px , above. items in nav open (or perhaps toggle) subnav lives directly below. these nav should appear centered on page. i have been able work through responsive portion of nav bar, having main items align vertically , sub nav menu of each item roll out below, , gave them indent separation. working well. however, issue not responsive side, behavior @ width greater 768px. happens toggling menu's sub navigation pushes other main items down page below sub nav. here's code: $(document).ready(function () { //stick in fixed 100% height behind navbar don't wrap $('#slide-nav.navbar-inverse').after($('<div class="inverse" id="navbar-height-col"></div>')); $('#slide-nav.navbar-default').after($('<div id="navbar-height-col"></d
Read more