Fragmented Messages
From HPCBugBase
HPCBugBase Menu
Submit feedback
Overview
Index
- Defect types (defect patterns)
- Specific defects (individual defects that belong to a defect type)
- Instances (code examples)
- Articles (various info)
- Templates
- Show all categories
- Show all pages
Index by Languages
Contents |
[edit] Fault Description
In general, messages between processes should be aggregated into the chunks of sufficient size to avoid the overhead of connection handshaking and message headers.
[edit] Statistics (Frequency)
[edit] Other Findings and Contexts
Interaction of computation and communication can influence the choice of an appropriate message size. A typical situation is that data becomes ready for sending as computation progresses. Waiting for sufficient amount of data to be become available can delay the communication. A tradeoff between the penalty of delay and the overhead of fragemented messages is dependent on application.
|
Pages referring to this entry: Performance |
