Using Point-to-Point Communications
Christopher Cameron, Brandon Barker, CAC Staff (original)
Cornell Center for Advanced Computing
Revisions: 5/2022, 1/2014, 10/2014 (original)
This topic gives an overview of how to use point-to-point messages effectively and avoid deadlock.
Objectives
After you complete this topic, you should be able to:
- Explain how deadlock occurs
- Identify situations where deadlock commonly occurs
- Describe four strategies for avoiding deadlock
- Explain which strategy to use in different situations
- Define the term persistent communication
- Explain how MPI objects are manipulated using persistent communication
- Describe a nonblocking message-passing routine in terms of persistent communication
- Demonstrate creating requests, sending/receiving messages, and cleaning up objects using persistent communication
Prerequisites
- Familiarity with HPC paradigms and MPI Basics.
- Experience programming in a high-level language such as Fortran or C would be helpful, but the reader can still follow along to understand the concepts and methodologies.
- A basic familiarity with parallel programming concepts
©
|
Cornell University
|
Center for Advanced Computing
|
Copyright Statement
|
Access Statement
CVW material development is supported by NSF OAC awards 1854828, 2321040, 2323116 (UT Austin) and 2005506 (Indiana University)
CVW material development is supported by NSF OAC awards 1854828, 2321040, 2323116 (UT Austin) and 2005506 (Indiana University)