Folding Forum

Arisu's mention above reminded me... If you fire up a Docker host running CUDA MPS, with the GPU in Exclusive mode, and discover your containers are having problems leveraging MPS, or the GPU at all... Make sure to bind mount the relevant host-level MPS directories to your individual containers. For...

I restarted all of my clients and the WebUI machines list is all fixed now. Maybe this helps someone else solve the same problem in their setup.

Appears to be working now. Kinda. I have one Fedora 43 rig which somehow can see all the machines via the 8.5.5 web console. My Windows 11 machine can see all my folding rigs except one Fedora 43 rig. My phone can see my Windows 11 box and my Ubuntu (docker folding) virtual machine but not either o...

I have four Folding clients in my house: my physical Windows-based gaming rig, two physical Fedora 43 rigs, and a virtual machine with a GPU passed-thru running Ubuntu 22.04 LTS. Normally I keep tabs on them all using my phone to log into and view the native client: https://v8-5.foldingathome.org/ma...

I'd suggest posting this in the "Issues with a specific server" subforum here: viewforum.php?f=18&sid=0fbe777e32c44562b414fe78cfc53a67

GROMACS 2026 now explicitly lists Apple M-series OpenCL as the recommended backend for macOS GPU acceleration. If the very engine F@H relies on is now confident enough to officially support it, perhaps the stability concerns of the previous decade are finally behind us. Why do you keep posting this...

I have two dual-GPU configs: an RTX 4090 paired up with an RTX Pro 6000 Blackwell Workstation, and a 3080Ti paired up with a 4070 Super. The former crushes about 80MPPD and the latter produces about 20MPPD. These figures are well in line with their expected output, as all cards appear to be pulling ...

If you're curious as to why muziqaz linked you to the ATI/AMD driver forum, it's because your GPU is flagged as unsupported (either because it really IS unsupported or because the driver is having a problem.) Your log file directly says so right here: 07:26:51:I3: "gpu:197:00:00": {"v...

For anyone else running Fedora OS who sees this thread and runs into this problem, and who isn't running Bazzite or another locked-down variant, I can help you solve this error. This no longer works because of dynamic linking to libbz2.so.1.0, which is named libbz2.so.1 on redhat. This is the true r...

I think you can be forgiven for not remembering the minutae from an instruction set that came out like 20+ years ago

There's no way a docker host would instantiate Exclusive Mode for the hardware GPU; that completely breaks the concept (and base reasoning) for using a container ecosystem to begin with. Containerized workloads were designed and built as a very lightweight means to run multiple workloads on the same...

I thought x86 AVX was 128, AVX2 bumped it to 256 That was Intel SSE with the 128-bit registers; MMX was 64-bit and AVX introduced 256-bit registers. AVX is also when x86 picked up dedicated FMA instructions. As for the original question, keep in mind you'll lose a measurable chunk of performance du...

Thought i'd come back and give some thoughts on the tuning thread linked above. First, HFM (Harlam's Folding Monitor) doesn't support the v8 client, which means there's no good way to ascertain "points per frame". It looks like Harlam has abandoned the project somewhere around the beginnin...

That thread is a gold mine, thank you for posting it Bob! I know what I'm gonna be doing this weekend :) Overheating your GPU? :D :D :D I mean, in theory, the result should be even less power draw than the roughly ~1500W (GPU-only) I currently dedicate to the fold. My office and "wiring closet...

is it wrong that I want to upvote your post?

Thank you for the work you and the team do.