2.5.65-mm3,4 with contest
Anticipatory scheduler tested...
no significant difference from 2.5.65-mm2
CFQ scheduler shows some fluctuations in results. At first I thought it was a mem leak because they were getting longer, but then they decreased again. Results are not consistent enough to give meaningful benchmark results at the moment. Still trying to track down what happened.
The only thing I can say with certainty is read load is faster than the AS (108 seconds v 122), and mem load is a little faster (98 v 102).
Con