Branch misprediction rule #262
Conversation
| yield new Finding( | ||
| `Branch Misprediction is ${diff}% different between '${ruleOpts.base_run}' and '${ruleOpts.this_run}'. ` + | ||
| `This can significantly impact performance since every time the CPU predicts incorrectly it has to flush ` + | ||
| `the current set of instructions it was working on and start over by fetching new instructions from the correct place.`, |
There was a problem hiding this comment.
This is probably too verbose. Should limit it to a single line if possible I think. For more detail we may want to link to other AWS documentation.
There was a problem hiding this comment.
I agree but I would like that at same time the user can understand what these metrics are about.
It would be nice to have a link to which we can refer to so that we can keep it here short but user can get more details if needed
There was a problem hiding this comment.
I agree such documentation should exist somewhere, but I don't agree that it should be on the "findings" page which is supposed to be a summary of what was flagged.
There was a problem hiding this comment.
I recall there is an optional extra argument to the Finding intended for this sort of extra detail (but I don't recall how / where it is displayed).
If more than 20% difference between two runs, it should trigger the rule.
Tested with local APerf profiles. Here the output: