-
Notifications
You must be signed in to change notification settings - Fork 95
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
sequential_distribute: Subdivide mesh :Distribute issue #218
Comments
@Girishchandra-Yendargaye The initial creation of the distributed domain is memory heavy. A sequential domain is created on processor 0 and then we run the structure through metis to partition the structure (again just on processor 0). This essentially doubles the memory. And then the partitions are communicated to the other processors. What can be done if you are going to be running a number of simulations with the same underlying domain, is to do the partitioning once on a large memory machine or computational node, and then use that distributed structure for all subsequent simulations. |
Apologies for the intervention (and for not using the users-list), but I have a follow-up question: |
@Girishchandra-Yendargaye Have a look at the code in the folder anuga/simulation and anuga/parallel/parallel_api.py There is code in there using sequential_distribute to do the creation of the the distribution without necessarily running the evolve. |
@stoiver Thank you sir for reply I am running only pickling part still it is failing...Below is code ####Output is Can you please suggest something. |
@stoiver One more issue while sequential_distribute Even if memory is available it is failing |
Distribution is failing..
sequential_distribute: Subdivide mesh
Error! ***Memory allocation failed for TRINODALMETIS: nind.
Why does distribute require additional memory
The text was updated successfully, but these errors were encountered: