Thanks for investigating this, SekeRob. We'll keep our eyes open on this.

If you lose power ALL of your work goes with it (including checkpoints).

.

I implemented the suggestions on a couple of Windows boxes, but frankly it was more in the vein of "Hey, this sounds logical..." than with the thought that I would immediately see measurable performance improvements. I don't see how you could document the improvement with any accuracy unless you ran the same task with the timestamping on and then with it off. Or in the case of multicore/hyperthreading processors, with an identical set of tasks. None of which is feasible.

Throw in the disk I/O and processor usage of the other software I have running (Windows Update, antivirus, Secunia to maintain updates on everything, etc.) obfuscating the big picture, and I truly don't see how I, at least, could verify what is likely to be a minuscule increase "per job" in performance. But like I said, it makes sense; each I/O involving mechanical movement saved multiplied by the number of disk accesses that CEP2 makes is likely to yield an improvement "over the long term". Needless to say, any disk I/O avoided is also likely to extend the lives of hard drives.

(O.T., but excluding the BOINC processes and data directory from their antivirus' monitoring is also a good thing.)