And I may do research into seeing how I can get Folding@Home to play nice with BOINC, when I have no work from GPUGRID, though I don't expect it to work.

I have been doing Folding on GPUs while running BOINC on CPU projects for years, on everything from WinXP through Win10 and now Ubuntu 16.10. In general, it is not a problem. However, Folding often has problems on multiple GPU cards (in spite of their propaganda to the contrary), unless they are of identical GPU chip types. But if you have only one GPU, then I don't expect that you will have a problem running both Folding and BOINC.

On my main PC, I run Folding on a GPU card in the PCIe slot, while using the internal Intel adapter to run the monitor. That allows you to Fold continuously and avoid any lag on the desktop you might get from the Folding projects. (That works on some motherboard/GPU combinations, but not all as I have found.) But I don't do games, so don't need the GPU for that. If you do need the GPU for games, you can run Folding in screensaver mode, so that it shuts down while you are using the GPU for the desktop.

About the only trouble I have seen is when running some VirtualBox projects on BOINC. That sometimes causes some crashes on Folding, but even though were short duration and did not interrupt the work units much. I always caution against overclocking the GPUs though; that can cause problems on either BOINC GPU projects or Folding.

Thanks. What is the easiest way to make the GPUs work on BOINC/GPUGrid when GPUGrid work is available, but work on Folding@Home when GPUGrid work is inavailable? If the only way to do that is by manually suspending/resuming both, then that's a showstopper for me.

Thanks. What is the easiest way to make the GPUs work on BOINC/GPUGrid when GPUGrid work is available, but work on Folding@Home when GPUGrid work is inavailable? If the only way to do that is by manually suspending/resuming both, then that's a showstopper for me.

That is the hitch. They don't talk to each other, so you have to do it manually. I usually devote separate machines for each GPU type. But if there are long outages on a given project (not to mention any names), doing it manually may be better than nothing.

Thanks. What is the easiest way to make the GPUs work on BOINC/GPUGrid when GPUGrid work is available, but work on Folding@Home when GPUGrid work is inavailable? If the only way to do that is by manually suspending/resuming both, then that's a showstopper for me.

I've folded with a 570+980Ti in one system and a 1070+970 in another system just fine. Different generations work fine.

Running BOINC and FAH at once just requires limiting BOINC tasks to certain GPUs. FAH stays on certain slots/cards but BOINC will run on whatever is available unless configured. But there are options to limit BOINC to certain devices via config files. I've ran FAH on the 1070 and Asteroids@Home on the 970 at once and they were fine with each other.

My 1070 is the 1st NVIDIA GPU in the system so it is Device 0 and Asteroids would not run on it.

<cc_config>
 <log_flags>
 </log_flags>
 <options>
  <use_all_gpus>1</use_all_gpus>
  <exclude_gpu>
   <url>http://asteroidsathome.net/boinc/</url>
   <type>NVIDIA</type>
   <device_num>0</device_num>
  </exclude_gpu>
 </options>
</cc_config>

I'm focusing my "backup GPU project" efforts on both Einstein@Home and SETI@Home. And I may do research into seeing how I can get Folding@Home to play nice with BOINC, when I have no work from GPUGRID, though I don't expect it to work.

I was off to Folding@Home with my GTX 1060 since POEM quit about 2 weeks ago.

I still had the GTX 970 running GPU-Grid. Since 2 GPUs are taxing my case cooling quite a bit, the 970 is throttled to ~ 110 W. This makes it run at 1.00 V, the lowest supported active voltage. For GPU-Grid I highly underclocked its memory, yielding about 1.2 GHz core clock and 30 - 40% memory controller load. With stock memory clock I'd only get ~1.00 GHz, i.e. a lot slower. POEM as my backup project was fine with that and needed even less bandwith (<10%).

Now the problem arose: without POEM my remaining choice would be Einstein, which needs all the bandwidth it can get. I couldn't figure a mechanism to switch OC profiles based on which project was running. And given the recent shortage of GPU-Grid work I'm switching this card to Folding as well, because I like neither idle nor inefficient setups. Or baby-sitting the machine.

And to answer your question: I have not even tried to configure Folding as backup. A few years ago they tried to create a BOINC project, but got lost in all the changes they'd have to make...

MrS

---

Scanning for our furry friends since Jan 2002

My 1070 is the 1st NVIDIA GPU in the system so it is Device 0 and Asteroids would not run on it.

You can disable a card from boinc in this alternate way.
This option applies to all projects.

<cc_config>
...
<options>
 <ignore_nvidia_dev>1</ignore_nvidia_dev>
 <ignore_intel_dev>0</ignore_intel_dev>
...rest of config file...

My 1070 is the 1st NVIDIA GPU in the system so it is Device 0 and Asteroids would not run on it.

You can disable a card from boinc in this alternate way.
This option applies to all projects.

<cc_config>
...
<options>
 <ignore_nvidia_dev>1</ignore_nvidia_dev>
 <ignore_intel_dev>0</ignore_intel_dev>
...rest of config file...

Ah, thank you. I haven't had to setup more than Asteroids (which now has Pascal support) but this would make it easier. I have plans to put 1 card on FAH and 1 on BOINC in the same machine so I'll have to remember this.

And given the recent shortage of GPU-Grid work I'm switching this card to Folding as well, because I like neither idle nor inefficient setups. Or baby-sitting the machine.

That is more or less my situation too. I don't mind intermittent GPUGrid work so much on my GTX 960s, since they are not great (only good) on Folding, due to the Quick Return Bonus. However, when the weather gets colder, I will need a continuous supply of work. Maybe GPUGrid will supply it by then and I can leave two cards on it all winter. Otherwise, ...

I have been doing Folding on GPUs while running BOINC on CPU projects for years, on everything from WinXP through Win10 and now Ubuntu 16.10. In general, it is not a problem.

That's interesting; because when I tried to run Folding on WinXP (a few months ago, when there were no GPUGRID task available for a few days), it did not work. So I asked in the Folding Forum, and I was told that Folding won't work on XP.

That's interesting; because when I tried to run Folding on WinXP (a few months ago, when there were no GPUGRID task available for a few days), it did not work. So I asked in the Folding Forum, and I was told that Folding won't work on XP.

There was a previous version of the Folding client that worked on XP. It might still be around if you look hard enough, though they are trying to wean people off of it. It required more manual intervention to set up, but worked in a lot of cases where the later versions, with their more automated setup, often would not work. Two different GPU cards was such a case (sometimes), but there were others.

EDIT: Here it is, but I think they changed how they communicate with the servers in the later versions, so the earlier one might not work, though you could try. Also, that was back in the days of Core 15 and earlier, and may not work with the current ones at all. It has been years since I used it.
https://folding.stanford.edu/home/faq/faq-v6/