pyperformance
pyperformance copied to clipboard
`--shuffle`, `--shuffle-seed`
The current order is sorted:
https://github.com/python/pyperformance/blob/0ffe825f0bd7b74f3da7a616057edc399aa4c69f/pyperformance/run.py#L76
The rationale: with the increasing number of runtime optimizations (eg: JIT, branch predictors etc.) there might be a small penalty (or reward) based on order of execution.
I don't think we should be shuffling benchmarks.