Insights on the Potential Deadlock
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
The results of the defect analysis in Potential Deadlock has led us to the following observations.
Contents |
[edit] Insights on Training
Since there was a clear difference between those who did and didn’t avoid the potential deadlock, it would be interesting to understand the reason. For example, if the absence of the defect was due to the programmer’s awareness of potential deadlocks, it should be possible to “teach” this defect to effectively reduce its occurrence.
[edit] Tool needs
Since deadlock is a very common problem, the detection of deadlock is supported by a number of existing defect detection tools such as Umpire, MPI_Check, and Marmot. Although the potential deadlock is also in their scope, we were unable to apply them to the defective code samples we collected, since they were either not currently available, they were not applicable to C programs, or they have not supported potential deadlocks in the implementation yet. This indicates that HPC developers may have difficulty finding tools for this purpose.
|
No pages link to $1. |
