[mpiwg-tools] ticket 383: variables with same name should have same meaning
mohror1 at llnl.gov
Fri Feb 7 15:17:09 CST 2014
This thread is to discuss the text we want to add to clarify that implementations should strive to provide a consistent view of variables across heterogeneous environments. In other words, variables with the same name should have "the same" meaning.
In our last call, we decided that the tools people need to figure out what makes sense for them and the implementors will see if they can provide what the tools people want. The current debate seems to be (from what I heard on the call):
1) If variables don't have the same exact meaning, they should have different names.
2) We should provide a high level set of variables that have the similar meanings across heterogenous environments (like the PAPI model, e.g., PAPI can report the number of floating point operations, but exactly what counts as a floating point op can vary across processor architectures).
So, let's discuss!
Kathryn Mohror, kathryn at llnl.gov<mailto:kathryn at llnl.gov>, http://scalability.llnl.gov/
Scalability Team @ Lawrence Livermore National Laboratory, Livermore, CA, USA
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mpiwg-tools