You are welcome. Usually, you benefit by enabling hyper threading, but there are exceptions. On the CPDN project (on my Intel Haswell chips under Windows), I don't see an increase in output when enabling hyper-threading. That is, turning it on doubles the cores, but each work unit takes twice as long, so there is no net gain. It is probably due to the old code they have to use as the basis for their studies.

However, here on WCG, I have always seen a benefit by enabling hyper-threading. But you do need more memory, since each running work unit consumes the same amount whether it is running on a real or virtual core. The WCG work units take up so little memory that it is not a problem on my machines, which have 32 GB, but if you had only 8 GB or so, you might want to disable HT and run on only real cores.

If hyperthreads have to share a floating point unit, you will see a definite performance hit on some types of application! That's why there's no real benefit for CPDN (lots of FP!) and you will also see a hit on MIP1 (also does some FP arithmetic, I believe.) However, the VINA jobs and MCM1 only seem to do integer arithmetic so the only issues left then will be memory access (which can be an issue with MIP1 if your processor has smaller caches.)

Until recently, a MIP WU on my Ubuntu-based laptop took 3-4 hours, sometimes only 2 hours. For about a week, a task has taken between 6-10 hours. I haven't changed anything about the setting, has anyone had similar experiences?

The latest batch I got today are running straight through without writing a checkpoint, which is a pain if you switch your computer off sometimes, as I do.

They also have very large data files compared to the usual, like 60MB as opposed to 22MB or less. Run times are variable, some only 45 minutes and some two and a half hours.

Tasks used to take under 2 hours for me (typically 1 hour 30 minutes), but now are 4 hours to 6 hours. I have not seen the memory warnings yet, but that could be because I have 32 GB on all machines.