dominic
dominic
the right solution to this is to run until we reduce the confidence interval for the iterations within a run to a particular threshold (or max time to avoid infinite...
yes that is correct, we start from scratch. i suppose we could change it to do an incremental run and check the running total of iterations and time instead. it...
> Can we, though? I was under distinct impression that the current way it's done is intentional, and it is a design guarantee that the benchmark-under-test will run for at...