Open modeswitch opened 10 years ago
Allow jobs to run until target confidence interval is reached rather than after a set number of replicates. For some benchmarks, we can use a smaller number of replicates this way.
we run a benchmark for x times (tasks), so we can view and analyze the results from the first tasks to see if they are stable enough.
Allow jobs to run until target confidence interval is reached rather than after a set number of replicates. For some benchmarks, we can use a smaller number of replicates this way.