Hi, everyone. I realize that this isn’t really important, but, I was wondering if anyone knows that answer to this question? I recently downloaded 2 Hadcm3 models onto a dual core machine. I noticed that they are running at slightly different rates. They have been running the same amount of time. CPU time is just about the same, but, one is 34 days (at 4.5% vs. 4.7%) ahead of the other, and the gap is widening. At this rate the difference will be something like 720 days by the time they finish.

I was wondering if this difference in processing rate is caused by some differences in the WU’s themselves or is it something in my computer. Maybe the computer draws asynchronously on 1 core to run the OS (Vista) and other programs? I know that it is not really important, but, I was wondering if others have noticed this.

---

This happens all of the time.

There are at least 2 causes:
1) a difference in the number and type of starting variables, and
2) one core will be used for \'other things\'. e.g writing your message, Windows house keeping, etc.

[Les Bayliss wrote]...
2) one core will be used for \'other things\'. e.g writing your message, Windows house keeping, etc. ...

Insofar as I\'ve been able to figure out processor preferences at all, it seems that CPU-0 will be preferred by some processes. This includes HADAM3 models which will use some CPU-0 time even if scheduled by processor affinity settings to run on another processor. I guess CPU-0 is a default value buried in Windows or some run-time library. However, there\'s a great deal of variability and, as Les says, WUs themselves differ.

Thanks Les and Iain for the info.

---