Improve stability of results
mitar opened this issue · comments
Results have higher instability than what I would like. Both during multiple iterations inside one run, but also across multiple runs. Not sure how this could be improved. Maybe by running the benchmark on a dedicated machine without any other interference at the same time?