c - Why is my OpenMP implementation slower than a single threaded implementation? (Followup) -
this follow why openmp implementation slower single threaded implementation? .
i have adhered answer provided, , used tasking instead of pragmas speed code. however, compared sequential (same) program, both programs run equally fast. witness no speed up.
the reworked code here: http://pastebin.com/3sfanec4
i removed pragmas , replaced tasking pragmas recursive procedures.
am doing wrong? should seeing linear speed up. guys think?
thanks!
first - still have "#pragma end critical" should removed. isn't causing problem, incorrect. second - said in other question posted, might have think how parallelizing code see speedup, replacing other pragmas task pragmas may not speed up. third - haven't put tasks parallel region, not running in parallel @ all. , can't add parallel region around tasks or going doing same tasks multiple times.
Comments
Post a Comment