The way I got BOINC and Folding to run fairly well together involved several steps:

1. While BOINC is shut down, tell Folding to create a GPU slot. Note that you MUST find and click inside the rather faint circle to the left of the GPU information for this to work, and must have a suitable GPU.

2. Run Folding long enough for any current CPU task to finish. Now before the next CPU task starts, delete Folding's CPU slot.

3. Restart BOINC, set it to not use the GPU, and not to try to get tasks from any GPU-only BOINC project.

4. Decrease the number of CPUs BOINC is allowed to use until BOINC and Folding do not interfere with each other or with programs you run from the console.

This lets Folding use your GPU for COVID-19 work, while having BOINC projects using the CPU and also doing COVID-19 work..

Running Boinc and Folding together can cause resource conflicts. Boinc can't see that Folding is using 1 (light), 3 (medium) or all 4 (full) cores so Boinc will, itself, try to use those cores as well causing Folding, Boinc and anything else you're trying to do, to slow down. You could set Folding to light, to use only 1 core and Boinc to use 3 of 4, 75%, or medium, 3 core and limit Boinc to 1 of 4, 25%. Or maybe set Folding to light, Boinc to 50% or 25%, leaving 1, or 2 cores free.

You don't try to run Folding CPU work at the same time as BOINC CPU work.

But you can easily run Folding GPU work while running BOINC CPU work.
You just reserve a core in BOINC for use by the GPU, typically by setting the "Use at most xxx% of the processors" to the appropriate value.
For an 8 core machine, setting it to 90% leaves one core free for use by Folding.
For a 16 core machine, 95% works.
And for a 32 core machine, 97% will do.
You can of course reserve more than one core if you want.

That is the same thing you would ordinarily do (or at least I would do) for reserving a core if running a BOINC GPU project, so it is really no different for a Folding GPU project.

---

Running Boinc and Folding together can cause resource conflicts. Boinc can't see that Folding is using 1 (light), 3 (medium) or all 4 (full) cores so Boinc will, itself, try to use those cores as well causing Folding, Boinc and anything else you're trying to do, to slow down. You could set Folding to light, to use only 1 core and Boinc to use 3 of 4, 75%, or medium, 3 core and limit Boinc to 1 of 4, 25%. Or maybe set Folding to light, Boinc to 50% or 25%, leaving 1, or 2 cores free.

You don't try to run Folding CPU work at the same time as BOINC CPU work.

But you can easily run Folding GPU work while running BOINC CPU work.
You just reserve a core in BOINC for use by the GPU, typically by setting the "Use at most xxx% of the processors" to the appropriate value.
For an 8 core machine, setting it to 90% leaves one core free for use by Folding.
For a 16 core machine, 95% works.
And for a 32 core machine, 97% will do.
You can of course reserve more than one core if you want.

That is the same thing you would ordinarily do (or at least I would do) for reserving a core if running a BOINC GPU project, so it is really no different for a Folding GPU project.

Simpler way, don't use folding. If they want me to use my computers on their project, they need to make it work in Boinc like everyone else has.

---

Simpler way, don't use folding. If they want me to use my computers on their project, they need to make it work in Boinc like everyone else has.

That reminded me to check back with Folding. They have had such a heavy server load recently (for the CPU work) that they haven't been able to handle it.
It looks like they can now. I can do an all-Folding machine again.

---

Simpler way, don't use folding. If they want me to use my computers on their project, they need to make it work in Boinc like everyone else has.

That reminded me to check back with Folding. They have had such a heavy server load recently (for the CPU work) that they haven't been able to handle it.
It looks like they can now. I can do an all-Folding machine again.

Heavy load for CPU? Surely GPU does a lot more?

---

Heavy load for CPU? Surely GPU does a lot more?

It is the server load I am talking about. They have different ones for different purposes.

---

Simpler way, don't use folding. If they want me to use my computers on their project, they need to make it work in Boinc like everyone else has.

That reminded me to check back with Folding. They have had such a heavy server load recently (for the CPU work) that they haven't been able to handle it.
It looks like they can now. I can do an all-Folding machine again.

Folding is planning to offer a BOINC version. Probably not until after they finish their COVID-19 work, though.

Folding is planning to offer a BOINC version. Probably not until after they finish their COVID-19 work, though.

I thing "planning" is a bit ahead of where they are at the moment. I have been in those discussions for some time.
They will probably look into it at some point. It could be done, but it takes some work.

PS - By the way, they are beta testing a new CPU core a8. It is said to offer a 40 to 50% increase in output, and allow more advanced science. It is based on the latest GROMACS 2020.
I don't know much beyond that, but want to check it out for a while. (I don't do the betas, but you can look on their forums at the beta section if you have a login).
If you set the "advanced" flag in their Control app, you can get the first ones right after the beta finishes.

---

Quite a wonderful bunch of replies! I'm having a brain hiccup, though:

How many cores do you have?

...remind me where I look to answer this, please?

Computer's always on; I remember to take BOINC off pause about half the time before I go to bed [/sheepish].

I might also try that GPU-slot idea mentioned as well.

How many cores do you have?

...remind me where I look to answer this, please?

Computer details: Number of processors: 4

Running multiple BOINC projects you might need to review the Resource share setting for each one, and the associated client setting Switch between tasks, to make sure Rosetta tasks have sufficient chance to run. (I only run Rosetta, so others can advise better here.)

Computer's always on; I remember to take BOINC off pause about half the time before I go to bed [/sheepish].

Do what I do, tell Boinc what the programs you run that need it to pause are, then it does it itself and always remembers to turn back on.

If you're using Boinc Manager, it's in "Options", "Exclusive Applications". Then just add the path to the program in either the top box for stopping Boinc altogether (ignore what it says, it's wrong, it stops your GPU aswell), or the bottom box to only stop the GPU and leave the CPU running.

If you're using Boinctasks, it's in "Extra", "Boinc Preference", "Exclusive Applications", making sure you have the correct computer selected. Then add the path to the program and tick "GPU only" if applicable, otherwise it pauses everything.

---

Problems and Technical Issues, eh? How about 41GB of RAM for ONE task? Name: ygG5REMC******1009391_1307_0

Problems and Technical Issues, eh? How about 41GB of RAM for ONE task? Name: ygG5REMC******1009391_1307_0

Time to upgrade your computer ;-)
Two of mine have 36GB RAM, but they'll take 128GB :-)

By the way, your image requires an Anandtech login.

---

I'm having a similar issue. Seems like one or two tasks keeps using as much memory as it can. It typically starts with normal behavior, then suddenly memory usage starts climbing fast at a linear rate of about 10GB per minute. If no memory limit is set for boinc, memory usage increases it hits the 32GB limit in my system (and quickly burns through swap). Then memory usage crashes back down to normal (about 12GB). If I set a memory limit, it would generally just suspend the task as "waiting for memory" (which leaves a core and a chunk of memory unavailable to other boinc tasks)

Sometimes it will behave normally a few minutes and sometimes it immediately starts another cycle of fast increase in usage, then immediately falling back to normal once it fills memory.

In my case, the problem workunits that have popped up while writing this are some variation on q1RftdTf_fold_and_dock. (This wasn't the only problem workunit, just the one that happened to be causing problems at this moment)
Here are links to one of the tasks that seemed to be causing the problem (which I aborted)
https://boinc.bakerlab.org/rosetta/result.php?resultid=1255773464

I'm currently running Rosetta on two computers, but only one of them seems to be running into this issue (at least at the current moment):
https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=5287905

The other computer seems to be running a non-overlapping set of tasks, so I don't know whether or not it would experience the problem given the same tasks.

This isn't really a major issue for me, but I'm assuming it's not intended behavior.

Problems and Technical Issues, eh? How about 41GB of RAM for ONE task? Name: ygG5REMC******1009391_1307_0

Looks like another batch of dodgy Work Units.

BTW- I'd suggest reducing the size of your cache, to 0.
You only need to cache Tasks if there's a chance of running out of work before you next get new work. Being signed up to a dozen active projects, that will never occur, so no need for a cache.
So no more missed deadlines- no point processing work if you're not going to get Credit for it.

Your account, computing preferences,

Other	
           Store at least 0.01 days of work
Store up to an additional 0.01 days of work

---

Grant
Darwin NT

Thanks for letting me know the image can't be accessed, but clearly you are BOINC'ing around with the wrong team, if you don't have an AnandTech forum account. :P The pic is just a screenshot of the properties of the offending task. As for the RAM upgrade, I am already maxed out at 512GB. I am not quite ready to buy 8 sticks of 128GB just yet....not sure the MB can support it anyway. ( I am teasing....X570 MB with 64GB of RAM, Ryzen 3950X. So you can imagine my surprise at seeing tasks "waiting for memory," especially since I am only letting 8 Rosetta tasks run for the moment )

Even if I had an account, I doubt I'd be asked to log in for an inline image. Or does it stay logged in forever on your browser?

---

Problems and Technical Issues, eh? How about 41GB of RAM for ONE task? Name: ygG5REMC******1009391_1307_0

Looks like another batch of dodgy Work Units.

BTW- I'd suggest reducing the size of your cache, to 0.
You only need to cache Tasks if there's a chance of running out of work before you next get new work. Being signed up to a dozen active projects, that will never occur, so no need for a cache.
So no more missed deadlines- no point processing work if you're not going to get Credit for it.

Your account, computing preferences,

Other	
           Store at least 0.01 days of work
Store up to an additional 0.01 days of work

Unless you run Milkyway on a GPU. Those have tasks that can take 30 seconds. And they refuse to fix the server (I've asked two successive project leaders and nothing gets fixed) - you cannot download new tasks if you're reporting completed tasks, so you need a big buffer (well 3 hours anyway).

---

Even if I had an account, I doubt I'd be asked to log in for an inline image. Or does it stay logged in forever on your browser?

My Firefox browser will save logon information indefinitely as long as you use it every so often.

Unless you run Milkyway on a GPU. Those have tasks that can take 30 seconds. And they refuse to fix the server (I've asked two successive project leaders and nothing gets fixed) - you cannot download new tasks if you're reporting completed tasks, so you need a big buffer (well 3 hours anyway).

If it were your only project, yes. If you're running more than one project, it's still not necessary even if one of the projects has issues with work allocation. Your other project will pick up work, and then BOINC will do extra for the first project when it can get work to balance out the debt between projects.

---

Grant
Darwin NT