Today I noticed there are two instances of "minirosetta_3.48_windows_intelx86.exe" in Windows Task Manager. In the BOINC Manager windows I clicked on the "Tasks" drop-down box and noticed there are two different Rosetta Mini tasks running, which makes sense since there are two instances of minirosetta running.

I don't recall seeing two instances previously. Is that normal and I just didn't notice it previously? If it is not normal, how did it happen and how do I fix it?

Thanks

Yep, normal since you have two CPUs. If you are running other BOINC projects as well, perhaps it is more common for you to see one Rosetta, and one from another project.

You can control the number CPUs BOINC is allowed to use in the configuration of the compute preferences. So if you only wanted BOINC to use 1 of your CPUs, or only run 75% of CPU rather than up to 100%, you can define these preferences. If you have questions on how to do this, please describe your objectives and there are lots of message board readers that can help.

---

Rosetta Moderator: Mod.Sense

Mod.Sense,

Thanks for replying.

The only BOINC project I'm running is Rosetta@home.

Correct me if I am wrong, but whether I continue running two Rosetta@home tasks simultaneously or limit BOINC to one CPU, thereby running only one Rosetta@home task at a time, won't my computer - in the "long run" - still end up *completing* the same number of Rosetta@home tasks in a given amount of time?

Incidentally, your comment that I have "two CPUs" was surprising to me because I didn't know the Intel P4 in my old GX280 desktop was a dual-core processor. I had to read about hyper-threading to understand that the CPU "appears" to the OS to be dual-core, even though it technically is not.

Thanks

Ah, ok, so the hyperthreading causes BOINC Manager to report 2 CPUs. And because you physically only have one that is why you are questioning if it is any faster to run two tasks.

I believe the jury (of user discussion) is out on that point. Much of Rosetta's work is floating point arithmetic, and for that reason, I believe you are on the right track in your thinking. But there are non-float operations that a second thread may be able to perform while the other thread does a float. So, I believe running two hyperthreaded CPUs does, slightly win over running a single CPU. It's been a long time since I've seen it discussed, but seem to recall the gain was on the order of 10% by running two.

If you'd like to discuss this point further, I'd suggest posting on the Number Crunching board. It attracts people that have tested such things for days and across multiple BOINC projects that could offer a more precise answer.

---

Rosetta Moderator: Mod.Sense