I just fired up a bigadv unit at -smp 8. Now, my GPU client is not getting enough CPU cycles and has gone from 40 second frame times to 4 minute frame times. What is the easiest way to give it priority over the CPU client? My CPU client priority is idle and my GPU client client is low. I hate to back down to -smp 7, but I really don't want to cripple this video card so badly. Thanks
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Linux - give higher priority to GPU client?
- Thread starter musky
- Start date
Tigerbiten
[H]ard|DCer of the Month - February 2007/January 2
- Joined
- Sep 24, 2002
- Messages
- 5,028
I use WinAFC to set my GPU's to normal and my bigadv VM's to low.
If WinAFC does not work on the VM, just set it to low with Windoze Task Manager.
That way I loose around 500-700 PpD from the bigadv for each gpu client I run, but each gpu client does ~10x that PpD drop.
Luck ........
If WinAFC does not work on the VM, just set it to low with Windoze Task Manager.
That way I loose around 500-700 PpD from the bigadv for each gpu client I run, but each gpu client does ~10x that PpD drop.
Luck ........
musky, did you remember to start the GPU with the nice command?
nice wine Folding@ome-Win32-GPU.exe
I fold with -bigadv and 3 GPUS on native Linux and have not had to make any other adjustments.
nice wine Folding@ome-Win32-GPU.exe
I fold with -bigadv and 3 GPUS on native Linux and have not had to make any other adjustments.
Started with this:
nice wine Folding@home-Win32-GPU.exe -verbosity 9 -forcegpu nvidia_g80
I tried it without nice as well, same results.
nice wine Folding@home-Win32-GPU.exe -verbosity 9 -forcegpu nvidia_g80
I tried it without nice as well, same results.
Very odd musky. Please run the top command for me please and tell me what nice levels the fahcores are running at please. You'll have 8 A2 cores and then 1 for whatever GPU is running. For reference, my bigadv runs at a nice level of 19 and the GPUs run at level 10 and everything works well.
The higher the number, the lower the priority. So, in my case, 19 is lower than 10.
The higher the number, the lower the priority. So, in my case, 19 is lower than 10.
3021 dave 39 19 702m 392m 7964 S 100 6.6 42:11.79 FahCore_a2.exe
3022 dave 39 19 700m 394m 9032 S 100 6.6 41:42.71 FahCore_a2.exe
3023 dave 39 19 703m 402m 10m S 100 6.7 42:07.33 FahCore_a2.exe
3024 dave 39 19 707m 405m 9876 S 100 6.8 42:16.38 FahCore_a2.exe
3025 dave 39 19 700m 399m 10m S 100 6.7 42:23.26 FahCore_a2.exe
3027 dave 39 19 701m 396m 9280 S 100 6.6 42:19.59 FahCore_a2.exe
3020 dave 39 19 794m 475m 8808 S 97 7.9 41:59.07 FahCore_a2.exe
3026 dave 39 19 702m 402m 9m S 95 6.7 41:59.93 FahCore_a2.exe
2877 dave 30 10 3686m 61m 19m S 3 1.0 1:57.76 FahCore_11.exe
Looks like the GPU core is at nice 10 and the a2's are at nice 19, but the a2's are grabbing most of the available CPU cycles.
3022 dave 39 19 700m 394m 9032 S 100 6.6 41:42.71 FahCore_a2.exe
3023 dave 39 19 703m 402m 10m S 100 6.7 42:07.33 FahCore_a2.exe
3024 dave 39 19 707m 405m 9876 S 100 6.8 42:16.38 FahCore_a2.exe
3025 dave 39 19 700m 399m 10m S 100 6.7 42:23.26 FahCore_a2.exe
3027 dave 39 19 701m 396m 9280 S 100 6.6 42:19.59 FahCore_a2.exe
3020 dave 39 19 794m 475m 8808 S 97 7.9 41:59.07 FahCore_a2.exe
3026 dave 39 19 702m 402m 9m S 95 6.7 41:59.93 FahCore_a2.exe
2877 dave 30 10 3686m 61m 19m S 3 1.0 1:57.76 FahCore_11.exe
Looks like the GPU core is at nice 10 and the a2's are at nice 19, but the a2's are grabbing most of the available CPU cycles.
Not sure what is going on. From that, I see the GPU using about 3% of the CPU They are already at nice level 19. Mine is using about 4 to 10% depending on which WU the GPUs are crunching. It could be something about the kernel that comes with 9.10 Ubuntu. I keep hearing mixed reports that 9.10 isn't too good for folding and this could be part of the reason why.
I run a very custom kernel with the BFS scheduler instead of the one that comes with the standard kernels from Linus. I've been trying to think of a way to make a custom drop in kernel for folks but haven't come up with a way to make it "idiot" proof and, as much as I like to provide as much help as i can to folks here, providing support on a custom kernel is so far beyond how I can spend my free time right now. I am still mulling over this idea though. There are far too many distributions out there to make this easy for folks who don't have a good working knowledege of Linux.
I run a very custom kernel with the BFS scheduler instead of the one that comes with the standard kernels from Linus. I've been trying to think of a way to make a custom drop in kernel for folks but haven't come up with a way to make it "idiot" proof and, as much as I like to provide as much help as i can to folks here, providing support on a custom kernel is so far beyond how I can spend my free time right now. I am still mulling over this idea though. There are far too many distributions out there to make this easy for folks who don't have a good working knowledege of Linux.
I heard the same thing. I think version 9.10 of Ubuntu is best used only for -bigadv and no other client. Even for standard SMP it seems to be inferior. V9.04 might not be as good as v9.10 for -bigadv, not sure about this, but if it's true, that would be the only situation where v9.10 may have the advantage. I'm going to have to try both these versions sometime soon because I'm very curious about the kernel differences myself.