Folding Forum

toTOW wrote: ↑Mon Mar 30, 2026 10:42 pm See comments in the beta thread about p18281 : viewtopic.php?t=43321

Good thing you posted that. I was thinking about 18261 since I get lots of them, and not the 18281 the OP asked about.

Complete user error on my end!

I haven't had any failures on that project myself. Whenever you get a failed work unit it's best to try and see if it failed for the next user as well. Just use the below link. https://apps.foldingathome.org/wu If it fails for the next user it is probably a bad work unit. You could post on the forum...

You can also increase the memory clock, which is useful for very large WUs (many atoms) and most GPUs handle it far better. The actual values for both shader and memory clock offsets varies not just for individual models, but for actual individual cards themselves. That is why the current values ar...

I just fold 23/7 and have downtime when needed to blow out the system. With a stable system there is no need to quit folding for video calls, watching video, even light gaming, along with other tasks. So I don't need to "keep up" with any scheduling, the system just keeps running. LAR syst...

Just to add to the thread: I had a similar issue yesterday when I installed a new GPU on 7.6.21 and it detected but showed as disabled. Tried slot options, a full uninstall and reinstall of 7.6.21 and several reboots, the new driver confirmations, etc. Nothing was making it work. Installed the 8.5.6...

Lem, As stated above by ToTow, frame time and/or PPD values are the only real way to measure improvements on F@H. Though it can take more time to test, I'd suggest you do so, testing with multiple methods (CPU and GPU, CPU only, GPU only, Boinc included wherever you do it, etc) and find hard results...

Which work unit IS running? I suspect that your CPU is folding, and the issue lies with detection of the GPU and/or iGPU.

Other than the error showing up, it worked fine for me on V7. I just haven't had any of them since November.

Forgot about this thread, and have confirmed the same error. I finally caught the last one and watched when it tossed the error, and there was plenty of memory left to use if it was needed. I noticed the trend of it being 18261 for me back before this thread. I didn't get many of those projects but ...

I surely see your view, but that is where I have to make the point that folding is an act of giving and also, nothing is truly disconnected. For example, I can run a fold, heat my home and give a researcher what they need. Folding could not be described as a total loss. I wouldn't be willing to pay...

That's the way the system works. At this point the researcher should get a notification of the failures, and it should quit being assigned.

Then the researcher can take a look and try to figure out why it failed.

It happens at times. Once it fails 5 times the researcher is notified and it should quit going out on the assignment servers. They can then use the information to try to figure out why it failed at the point it did.

The testing I did was on Turing hardware, with many fewer cores. I would imagine at some point the memory bandwidth vs total core count comes into the picture, along with power limits and where a little extra power helps and doesn't. I've found that most changes will work on all projects, but at tim...

Just food for thought.... I have used both the EVGA (Precision X1) and MSI Afterburner software without any issues, but keep in mind in some cases you can set it to boot to that setting and might have missed it. But in either case, both seem to work fine. BUT I did have one project when I first got ...

All those projects are still in beta testing. If you are on the beta team, you should have posted in the beta section. If you are not on the beta team, you now know one of the reasons why beta testing can be less stable and result in failed work units. There may be issues that cause instability with...