Linux freeze/stop working while FAH running

[MaxBarcon]

OS: Manjaro Linux x86_64
Kernel: 5.16.11-2-MANJARO
DE: Plasma 5.24.2
CPU: Intel i7-10700KF (16) @ 5.100GHz
GPU: NVIDIA GeForce RTX 3070
Memory: 4578MiB / 15887MiB


When I'm using FAH it suddenly freezes all computer and i need to hard restart becuse it does nothing.


(I would put the logs of the crash/freeze, but I don't know what is needed, i will post them when someone tells me what to put)

[toTOW]

Check your system stability (memtest, temperatures, PSU, ...) ...

Try to run only CPU and only GPU alternatively ...

[ajgringo619]

If running GPU tasks, what driver version are you using? I'm on the same kernel version as you (EndeavourOS) and use the v470 Nvidia drivers (downgraded for other reasons).

[Knish]

to post logs, at the top of every forum section there are some stickied threads, 7th item in "Welcome to the F@H Support Forum" viewtopic.php?f=61&t=26036 will guide you for linux

altho for Manjaro (if that's arch based?) if u don't see them in the path described, for arch they're located at /var/log/foldingathome/

[MeeLee]

Check for overheating and if your PSU is powerful enough.
If you have a cheaper CHinese PSU, you'll want to stay below 80% of the total rated power draw at the wall, because they're not reliable at 100% of their rating.

[rickoic]

I have been having the4 same problem on 2 different computers, one Windows 10 and the other Windows 11. I ran memtest with no problems found. Then I stopped running FAHClient on both and just monitored via web control. The one on Win 11 was freezing and rebooting every 3-6 hours while the one on Windows 10 every day or so. For the past 3+ days and NOT running FAHClient there have been ZERO freezes or reboots. The Windows 11 computer is running 3 GPU's (2080 Super, 2080 A and 2070 Super) on a eVga 2099 Dark MB with a Hercules 1lll600W PS and 32GB RAM. The Windows 10 computer 4 GPU's 2080 2070 and 2 1070's with 2 1000W PS sharing the load. Win 11 and Win 10 both running FAHClient 7.6.21 Nvidia Drivers are both current on both computers.

[Peter_Hucker]

Check for overheating and if your PSU is powerful enough.
If you have a cheaper CHinese PSU, you'll want to stay below 80% of the total rated power draw at the wall, because they're not reliable at 100% of their rating.

50% with ones I've had. CIT and Alpine (now bankrupt for some reason - those ones actually went bang above 50%).

[MeeLee]

I have been having the4 same problem on 2 different computers, one Windows 10 and the other Windows 11. I ran memtest with no problems found. Then I stopped running FAHClient on both and just monitored via web control. The one on Win 11 was freezing and rebooting every 3-6 hours while the one on Windows 10 every day or so. For the past 3+ days and NOT running FAHClient there have been ZERO freezes or reboots. The Windows 11 computer is running 3 GPU's (2080 Super, 2080 A and 2070 Super) on a eVga 2099 Dark MB with a Hercules 1lll600W PS and 32GB RAM. The Windows 10 computer 4 GPU's 2080 2070 and 2 1070's with 2 1000W PS sharing the load. Win 11 and Win 10 both running FAHClient 7.6.21 Nvidia Drivers are both current on both computers.

Just for a 2080 super, 2080 and 2070 super, you'll need roughly a 1000W PSU.
600W for the GPUs, 100W for a small quadcore PC, or 200W for a Ryzen with 105W TDP system; totals around 700-800W, which falls right in the efficiency curve of a 1000W PSU.

For the second PC, with 4 GPUs, it's best to run a single 1200 min to 1500W max PSU. Never run 2PSUs on one system, it's asking for trouble; unless it's an industrial server system with 2 PSUs built from factory.
I'd also be careful running a 1500W PSU, as many homes built in the 70s use aluminum wires, vs copper (due to the high copper prices), and these wires are rated for up to 15 AMPS (or 1800W peak), but you more than likely want to keep it to 1200W continuous (not to cause a fire).

[rickoic]

I have been having the4 same problem on 2 different computers, one Windows 10 and the other Windows 11. I ran memtest with no problems found. Then I stopped running FAHClient on both and just monitored via web control. The one on Win 11 was freezing and rebooting every 3-6 hours while the one on Windows 10 every day or so. For the past 3+ days and NOT running FAHClient there have been ZERO freezes or reboots. The Windows 11 computer is running 3 GPU's (2080 Super, 2080 A and 2070 Super) on a eVga 2099 Dark MB with a Hercules 1lll600W PS and 32GB RAM. The Windows 10 computer 4 GPU's 2080 2070 and 2 1070's with 2 1000W PS sharing the load. Win 11 and Win 10 both running FAHClient 7.6.21 Nvidia Drivers are both current on both computers.

Just for a 2080 super, 2080 and 2070 super, you'll need roughly a 1000W PSU.
600W for the GPUs, 100W for a small quadcore PC, or 200W for a Ryzen with 105W TDP system; totals around 700-800W, which falls right in the efficiency curve of a 1000W PSU.

For the second PC, with 4 GPUs, it's best to run a single 1200 min to 1500W max PSU. Never run 2PSUs on one system, it's asking for trouble; unless it's an industrial server system with 2 PSUs built from factory.
I'd also be careful running a 1500W PSU, as many homes built in the 70s use aluminum wires, vs copper (due to the high copper prices), and these wires are rated for up to 15 AMPS (or 1800W peak), but you more than likely want to keep it to 1200W continuous (not to cause a fire).

Home built in 2001 so wiring should be ok. Running Hercules 1600W for motherboard and gpu's so power should be alright there. Plan on trying FAHClient tomorrow to see if I start getting freezes and reboots again, will update with further information.

Second PC am running a Rosewell 1000W power supply for the motherboard and the 2080 gpu. Have it set up in a mining rig frame which allows me to plug in a second power supply which is turned on/off by connection to the main power supply. Mining power supply is running only the other 3 gpu's.
Will also try FAHClient on it tomorrow also. And will do a reboot of both computer prior to starting FAHClient.