Can anyone tell me why software calling mlockall on 1 linux box is much faster than calling the same call on another linux box. Both computers are running Red Hat Enterprise Server (one has 5.4, the other 5.0). They both have 8 processors, etc. I am guessing this could be a system configuration issue. So the computer running 5.4 has a mlockall performance that is roughly 10 times slower than the 5.0 computer. Any ideas?