I develop a parallel time-domain DAE simulation software using OpenMP and Fortran 2003. The main algorithm is:

while t<time_horizon do
   while not converged do
         Solve sub-domain DAE using a portion on X and X_1 to get DeltaX
         Correct X=X+DeltaX

X is the state variable vector and X_1 is the the vector at the previous time-step used for the discretization algorithm (Trapezoidal method).

The sequential implementation profiling shows that the memory copy takes 15% of the computational time, a significant amount...

I tried:

!$omp parallel workshare
!$omp end parallel worshare


!$omp parallel
   X_1(i*length/num_threads+1:(i+1)*length/num_threads)= X(i*length/num_threads+1:(i+1)*length/num_threads)
!$omp end parallel

This helped a little bit (dropped to 13%) but not much... Any thoughts? Any other way to keep the history? I keep the X of all subdomains as one vector as it's easier to handle (apply infinite norm, copy, etc.). Should I break it into one for each subdomain?

Thanks in advance!

  • $\begingroup$ This question is more about Fortran and OpenMP programming, you can probably get better answers in StackOverflow $\endgroup$ – RSFalcon7 Apr 5 '13 at 18:49

You do not need to copy the state vector. Assuming you are storing two copies X1 and X2, you can have the parallel region first work on X1, and then work on X2 on the next iteration. In Fortran, one easy way to do so is to move the parallel region into a subroutine, and to call it from within the main loop:

while t<time_horizon do
   call evolve(X1, X2)
   if (t>=time_horizon) then
      "swap x1 and x2"
   end if
   call evolve(X2, X1)
end while
  • $\begingroup$ The problem is that at the very beginning of the solution X1 and X2 need to be the same. X1 as the previous point used for the discretization and X2 as the initial point for the solution. At the end of the solution X1 still has the old point and X2 the new computed point. Maybe I don't understand something... $\endgroup$ – electrique Mar 26 '13 at 14:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.