I am planning on benchmarking some code based on the erl flag +sbwt, and I was wondering if that could have an unintended effect benchee’s output. Should I use a different benchmarking method for something at this level?
It won’t mess with Benchee itself, but you’ll have to sort of generate your own output by combining different runs. I’m not sure if that stuff can be configured at runtime, but if it can then you could conceivable use a before hook to configure the scheduler and then get a good output that way.
Awesome, glad this won’t bork Benchee between runs. Yeah, sadly I’m going to have to do a good amount of copy/pasting between changing the erl args.
Wouldn’t https://github.com/bencheeorg/benchee#saving-loading-and-comparing-previous-runs this cover your case of wanting to compare runs? Independent of erl options.
I should have RTFM, accepted that as a solution even though it answered a different question than the original one. Thanks!