Scaling my AMReX code #3193
Unanswered
attarnasir
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hey AMReX folks,
I was trying to solve a 2-D problem with explicit time-stepping using AMReX, and my main aim was to show the code was scaling. I ran my code for two grid sizes 1) 1000x1000 and 2) 2000x2000. The execution time for the 1000x1000 grid was 6.39 mins, and for the 2000x2000 grid was 14.45 mins. The results match in both cases. I was wondering how this was possible. In a 1000x1000 grid, if my code is 6.39 minutes, then for a 2000x2000 grid (which is a four times increase in the number of points and thus a four times increase in the number of computations), it should ideally take close to 25.56 mins (6.39x4) assuming linear scaling. But how is AMReX being so efficient and delivering an execution time lower than the linear scaling time? What is happening here? I’m curious to know.
Important information:
Any advice would be greatly appreciated!
Thank You
Beta Was this translation helpful? Give feedback.
All reactions