Folding Forum

Shortly after changing my name and team affiliation, I accidentally put the system into an almost-unrecoverable state (doing things unrelated to FAH) and had to use sysrq-E to kill a non-responsive Xorg that wasn't letting me change VTs. To my surprise, fah-client was still running but at 100% CPU (not the cores, the client!). I stopped the service and then started it up again and fah-client began to report that I was getting over 250M PPD, which obviously is not possible since my iGPU maxes out at around 250k PPD. The estimated PPD has been slowly going down since.

I apologize that I don't have more information about this event.

While PPD issues is being worked on, I would suggest using FAH on the machines which are 100% stable, please. Leave your experiments on the machine which has no FAH on it

I've rectified that problem. I'm now doing my tests with old WUs without giving the client access to the internet, so my experiments, FAH-related and not, won't cause issues anymore.

arisu wrote: ↑Wed Mar 05, 2025 8:00 am I've rectified that problem. I'm now doing my tests with old WUs without giving the client access to the internet, so my experiments, FAH-related and not, won't cause issues anymore.

If you are using the command line I gave you with the WU o sent you, you can run it on any PC with internet. Fahcore on its own does not send any work to any servers.

muziqaz wrote: ↑Wed Mar 05, 2025 10:11 am

arisu wrote: ↑Wed Mar 05, 2025 8:00 am I've rectified that problem. I'm now doing my tests with old WUs without giving the client access to the internet, so my experiments, FAH-related and not, won't cause issues anymore.

If you are using the command line I gave you with the WU o sent you, you can run it on any PC with internet. Fahcore on its own does not send any work to any servers.

I'm also testing the client and how it responds to the core's behavior but when I'm testing only the core, I use that WU you sent me (and an expired WU for a8 that I already finished and saved) by running the core directly.

I also observed the huge inflated PPD after a Manjaro Linux system crash. WU was approximately 38% complete at crash.
As the reboot resumed folding, the PPD steadily decreased as folding progressed until it normalized before finish.
I figure some clock must have reset FAH start time internally, so upon reboot it appear my machine had miraculously completed 38% of the WU *instantly*.

I was not worried about it at any rate.

That was my thought too. The TPF was normal so it wasn't actually skipping any steps. I'll make regular backups of client.db and do tests if it happens again to see if I can isolate the problem. It's purely a UI issue and doesn't impact the core's work, but even UI bugs need to be fixed.

The UI just shows what the client says.

I think the eta overestimate is from the checkpoint being close to the next integer percentage. When the percentage bumps up quickly, the client becomes confused about the TPF.

The whole PPD reporting needs a rework, as currently it is not that accurate, and it starts measuring the TPF from the moment WU is downloaded instead of when 0% is printed out

I am just very glad that this 8.4.9 Linux version retains the partially complete WU state upon abnormal shutdown!
I used to lose the completed folding upon reboot, but now it picks right up where it left off.