Does AutoTVM support parallel candidate trials?

I don’t mean that XGBoost algorithm to execute in parallel which is currently supported. How about the parallelism support of multiple trials from the search space?