[WIP] New caliper modifier implementation #618
Draft
+78
−79
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Outdated suggestions cannot be applied.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.
Work in progress. This PR changes the caliper modifier implementation (Solves #603)
In the existing caliper modifier, we define caliper modes to collect specific metrics (e.g. cuda, mpi, time etc)
The current benchpark implementation sets these modes at experiment init time. Each of these modes creates an environment variable in the experiment bash script to collect the desired metrics. However, this approach does not facilitate benchmark-specific changes as the caliper modifier is processed during the Ramble setup phase.
In the new impementation, we use the caliper modes specified on experiment init to directly set the required variables in the experiment ramble.yaml. The generated ramble.yaml is functionally equivalent to the one generated by the existing caliper modifier. However, this approach allows for changes to the caliper config at benchmark init time.
Adding/modifying a benchmark (docs: Adding a Benchmark)
application.py
or inrepo/benchmark_name/application.py
will appear in the docs catalogueexperiments/benchmark_name/experiment.py
to define a single node and multi-node experiments