GPU issues in F@H
GPU issues in F@H
The issue involves F@H turning off both GPU slots and failing to detect them in several applications. The cause appears to be related to the Coolbits configuration and the 20-nvidia.conf file added to etc/X11/xorg.conf.d. After rebooting, the system became unresponsive, crashed, and eventually stopped powering on, showing no life signs. After resetting CMOS, it started again but only the integrated GPU displayed output, while the discrete GPUs remained inactive. I removed the 20-nvidia.conf file and reinstalled F@H, testing various GPU settings, but the problem persisted. NVIDIA X-Server and PSensor didn’t recognize the GPUs at all. The BIOS still attempted to use GPU 1, so I had to set it as the primary display to see any output. A warning message about Nvidia-Persistenced-Service appeared briefly before shutdown. I’m using a proprietary Nvidia driver that previously worked. I also noticed a brief failure message when the system powered off. System specs include an i7-4790K CPU, 2x4GB DDR3 RAM, ASUS H87M-Pro GPU, RTX 2070 Super and GTX 1060, EVGA 750BR PSU, 256GB SSD, Debian 12 OS, and F@H client version 7.6.21. Help would be appreciated!
Have you attempted to fully delete and reinstall the Nvidia driver package?
The system doesn't seem to be working as expected. Even with faulty Xorg or drivers, it should have booted into text mode without problems. If it functions at all, the current desktop settings are: cat /etc/X11/xorg.conf.d/01-nvidia.conf Section "Monitor" # HorizSync source: edid, VertRefresh source: edid Identifier "Monitor0" VendorName "Unknown" ModelName "GBT M28U" HorizSync 246.0 - 246.0 VertRefresh 48.0 - 144.0 Option "DPMS" EndSection Section "Device" Identifier "Device0" Driver "nvidia" VendorName "NVIDIA Corporation" Option "NoLogo" "1" Option "Coolbits" "12" EndSection Section "Screen" Identifier "Screen0" Device "Device0" Monitor "Monitor0" DefaultDepth 24 Option "Stereo" "0" Option "nvidiaXineramaInfoOrder" "DFP-3" Option "metamodes" "3840x2160_120 +0+0 {ForceCompositionPipeline=On, AllowGSYNCCompatible=On}" Option "SLI" "Off" Option "MultiGPU" "Off" Option "BaseMosaic" "off" SubSection "Display" Depth 24 EndSubSection I wonder how to activate Coolbits on Wayland soon. Some say it's possible without it: https://github.com/nan0s7/nfancurve I'm skeptical since I usually relied on the nvidia-settings CLI and think Coolbits was necessary.
Yes, the CMOS battery might be deteriorating and causing issues with system settings or power management.
I swapped it last year since it was flat, and it keeps its settings without issues, even after being unplugged for extended periods (the computer starts up normally now, which hasn’t happened before). I’ll see how it’s holding up when I’m back home.
I attempted it now, but it remains inactive (still displays "[Failed] Nvidia-Persistenced-Service" during startup). It attempts to launch after boot fails and redirects you to "syslog," which isn't accessible. I used the removal command: apt purge "*nvidia*". For reinstallation, I ran: apt install nvidia-driver firmware-misc-nonfree. Edited April 16, 2024 by Average Nerd
You can adjust clock speeds via the CLI, which helps more than just capping TDP—by lowering speed until it reaches a better efficiency point on the voltage line. Once optimized, you can assess power usage by running nvidia-smi without extra flags or custom options like: nvidia-smi --query-gpu=gpu_bus_id,timestamp,driver_version,pcie.link.gen.current,pcie.link.width.current,temperature.gpu,fan.speed,utilization.gpu,utilization.memory,memory.used,memory.free,clocks.current.graphics,clocks.current.sm,clocks.current.memory,power.draw. This method works well on my Folding@Home setup, even though it’s not as precise as Windows tools like OC Scanner in Afterburner.