Add parallel utilization #73
Open
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Describe what the pull request does
Implementation of a utilization field for parallel profile types. This is used so that the parallel job then uses multiple cores of each host (#72). This is an optional field that by default is assumed to be 1. This does not change the behavior of the code when "core" is not defined for the host. When core is set, the parallel job is assumed to use the whole node (which seems to be intuitive).
Approach
The number of cores to use is calculated based off utilization in the same way as the usage trace. The computation and host vector are then repeated that number of times. The communication matrix is 0 extended, as if only the first core on the host is communicating across hosts.
Limitations
This could lead to higher overheads when core is set to a high number. Also, currently I support a single utilization number, which was enough for my purposes. It may be convenient to support a utilization per host vector, especially in the parallel profile type.
Alternatives
Originally, I intended to use the
s4u::Exec::set_boundmethod. However, this is not supported for theL07model.Checklist
Branch name.
Branch content.