HPC-05_mcastril_MPI_optimization

Last edition: 01/08/20 12:02:18 by mcastril

The PI is responsible to closely follow the progress of the action, and especially to contact NEMO project manager if the delay on preview (or review) are longer than the 2 weeks expected.

  1. Summary
  2. Preview
  3. Tests
  4. Review

Summary

Action HPC-05_mcastril_MPI_optimization
PI(S) Miguel Castrillo
Digest To reduce the overhead caused by MPI communications in NEMO
Dependencies None
Branch source:/NEMO/branches/{YEAR}/dev_r{REV}_{ACTION_NAME}
Previewer(s) Italo Epicoco
Reviewer(s) Italo Epicoco
Ticket #XXXX

Description

Assessment of the model sensitivity to the network latency, revealing the impact of internode messages on the model scalability. Different strategies will be evaluated, as reducing internode communication time and frequency, limiting the communication overhead in general, improving the balance between the number of messages and the packet size as well as studying the optimal distribution of the model subdomains according to the computing resources and network topology.

Implementation

The implementation will start with the classic scalability tests and performance analysis using state-of-the-art performance tools. From this information, an assessment will be done and we will propose the main directions in which we can work to take more profit from reducing communication's overhead. Then the changes will be performed and another study will evaluate the outcome of the actions and validate if they fulfilled the expectations.

Documentation updates

Preview

Tests

Review

Last modified 7 months ago Last modified on 2020-01-08T12:02:18+01:00