Never found an actual hard answer on why this happens to some, the sum of the environment we guess. For instance if I had a HPF2 looping, there was [almost always] also an FAAH, AutoDock running.

Setting an AV scanning exception for windows on usually C:\ProgramData\BOINC + Subdirs is advised [picture a job writing hundreds of times and your Trend scanning this time and again, file lock / race condition]. BOINC is sandboxed and keeps track of what's there, as do the projects, hashing, MD5 checking, inventory of task files. Particularly the science apps are verified to make sure they're not tampered with and run in an isolated shared memory segment. Save to do, practically the very first thing I do after a new installation.

[ot]Interesting you have 20+ years on HPF aka HPF1. Think this never ran on BOINC, or just briefly, i.e. 1 thread per computer. Lots of hardware to get that far... goodoos.[/ot]

Hmm,

I have set up some exceptions as you suggested and I also tried turning off the Trend AV software. Neither of these actions seems to make any difference. (It's now all turned back on). I find it interesting that it only seems to be this one project that gives me this issue.

Riight now I'm looking at one CPU core that is running this task and it shows running time 07:04:15 and remaining time 01:16:02. I'm going to re-start BOINC now, and I now have a running time of 01:51:55 and a remaining time of 01:27:13 and CPU is working hard again.

[ot]Yes most of my early crunching was done on the pre-BOINC agent which I got on with really well. I did at one point have over 30 machines crunching - old single-core machines but the electricity bill became too much of a burden so they had to go. I now have just two single-core, one twin-core and one quad-core devices running.[ot]