Cannot use GPU for Folding@Home on Linux

[ROBOCRIPPLE]

The computer that I'm using for Folding@Home is running Linux Mint and has a GTX 960. I want to use that GPU for Folding@Home, but for some reason, it's not letting me. I'm using the latest Nvidia driver compatible for the GPU (440). I tried adding the GPU in FAHControl by doing it this way:

 

Quote

Configure > Slots > Add > GPU > gpu-index: -1 > OK > Save

but I end up with this error message

Quote

On Client "local" 127.0.0.1:36330: No available GPUs

All I can contribute is my CPU. But I want to contribute my GTX 960, because I've heard that projects get done much faster on GPUs than they do on CPUs.

 

I do not understand why this must be. Is there a proper way to use my GPU for Folding@Home on Linux? If so, how?

[ely105]

I found FAH doesn't correctly identify the GPU on install even with all the drivers installed.  You have to manually edit the config xml file to say GPUs exist and then you can add them to a slot.  I had to do this on 2 different systems.  -m

[Quoda27]

All you need to do is add a few lines to the config.xml file. Once you've located it add the following:

<gpu v='true'/>

And then, underneath your CPU slot add:

<slot id='1' type='GPU'/>

(Assuming that your CPU slot is index 0)

Then restart F@H and you should find it's folding on both CPU and GPU. It might splutter a bit over updating your GPUs.txt file but once that's done you're good.

[ROBOCRIPPLE]

8 hours ago, Quoda27 said:

All you need to do is add a few lines to the config.xml file. Once you've located it add the following:

<gpu v='true'/>

And then, underneath your CPU slot add:

<slot id='1' type='GPU'/>

(Assuming that your CPU slot is index 0)

Then restart F@H and you should find it's folding on both CPU and GPU. It might splutter a bit over updating your GPUs.txt file but once that's done you're good.

I did just that (in /etc/fahclient/config.xml), except replaced <gpu v='false'/> with <gpu v='true'/>. Was that right? FAH does detect the GPU, I haven't got it to work on a project (status = ready), and it's been hours since I turned it on.

[MSMSMSM]

Debug log? That'll show you what's exactly going on. If it's a work unit problem (no allocation), then it's just a matter of waiting, if it is anything else, it's a configuration problem. Sometimes running FAHClient with sudo helps (not sure about Linux Mint or NVIDIA in particular)

[Quoda27]

8 hours ago, ROBOCRIPPLE said:

I did just that (in /etc/fahclient/config.xml), except replaced <gpu v='false'/> with <gpu v='true'/>. Was that right? FAH does detect the GPU, I haven't got it to work on a project (status = ready), and it's been hours since I turned it on.

Yes that was correct. If it's detecting the GPU but isn't doing any work it's more likely that it's just not getting given anything from the assign servers, which are being pretty overloaded at the moment. I have a few rigs not really doing much because they can't get a workload. Leave it running and it'll pick some work up when it's available.

Of course, if you send me / post up your logs I can confirm that for you.

[ROBOCRIPPLE]

10 hours ago, Quoda27 said:

Yes that was correct. If it's detecting the GPU but isn't doing any work it's more likely that it's just not getting given anything from the assign servers, which are being pretty overloaded at the moment. I have a few rigs not really doing much because they can't get a workload. Leave it running and it'll pick some work up when it's available.

Of course, if you send me / post up your logs I can confirm that for you.

That makes sense, especially during this time of the coronavirus pandemic. Most of the projects that my CPU (FX-8350) has been working on are 14329 (Coronavirus Project: Coronavirus SARS-CoV-2 (COVID-19-causing virus) protease - potential drug target). I'm not very familiar with Linux, so how do I post a log?

 

Thank you, LTT community, for helping me with this.