User has to know what to do with the errors that comparison script reported. If it's an ambiguous result, should they run again? Collect more data? This issue can be closed by a PR that prints suggested steps in nvbench comparison script.
User has to know what to do with the errors that comparison script reported. If it's an ambiguous result, should they run again? Collect more data?
This issue can be closed by a PR that prints suggested steps in nvbench comparison script.