I'm not sure if this is an appropriate question for Chemistry.SE, but I'm taking a stab. I'm running ab initio MD calculations in VASP and I'm trying to optimize different input tags. On the VASP website, it says that the MAXMIX tag should be set to at least 3 times as large as the number of electronic iterations required to converge the first step. I tried doing this, but I didn't see any speed up and instead a significant slow down. I'm wondering if anyone else has encountered this issue.
The system that I'm simulating is a silicon supercell in contact with water.