I use two laptops. One is an older, Dell G7 17. Other is newer, Alienware G15 20 cores and the source of earlier posted issues.

I have now observed that the newest batch of LHC jobs run well enough on the older Dell. In fact, heat isn't even generated to excess, and my earliest problem was solved by switching off EFmer TThrottle and dedicating that older model to LHC jobs; they run cool and run on time.

On the newer machine, I thought I had solved the problem by disabling TThrottle and removing any other job; I could dedicate 16 cores to LHC, two jobs at 8 cores apiece. It still ran like Zeno's Paradox, it would just keep slowing until it could never reach the end.

I'm giving new machine a shutdown, time out, lots of rest, but maybe it just isn't any good for LHC jobs. It can run Einstein, Prime Grid, Amicable, it can run 19 core Amicable jobs but then it has to be dedicated to that one job because it has 20 total cores. But I just have to steer it away from LHC and keep those on the older model I guess.

---

Dylan Madeley, Independent Author

It could be a lot of things
hardware , os , boinc , virtualbox , task (which I personally have encountered quite often with lhc), etc
The task itself can be for a very long period, like rosetta does (>10000h)
I suggest that you collect the data on all of the above and send it to the lhc forum
https://lhcathome.cern.ch/lhcathome/forum_index.php
They know more about their tasks and are more likely to be able to help
good luck!

Thanks. I have joined their forum. In the meantime, I have found ways to keep the LHC ATLAS simulation tasks steadily going to the one machine that processes them pretty well, and I find other things for the newer one to do.