afaict, compile is already using all CPU now (parallelization at target level), so I don’t think parallelization itself will improve the performance. That is not to say it won’t benefit from other optimizations in hydra