• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

*** F@H Tech Assistance Thread ***

You could, but they're not very efficient.

OK, so folding doesn't take advantage from the massive DP numbers of the Kepler Titan it seems ...
This spreadsheet only has 1 sample, and no Core22 performance numbers.
But yeah, efficiency won't be that great anymore, meaning more heat output.
Maybe I'll just have to try it out this coming winter :)

I found this in a Q&A dating from 2013:
"The folding cores in current use do single precision calculations. Some older cores tried out double precision calculations, but the additional overhead was found not to be needed. "

Now this article from 2015 really gets me interested: The NVIDIA GeForce GTX Titan X Review
The GPU's can compute with implicit and explicit single precision.
In implicit single precision mode, the performance increase for the Kepler Titan is roughly x7. It is measured in nanoseconds per day (as found in the F@h log).
It goes from 48,7 to 344,4 ns/day. Now this is definitely worth a try.

I did a run on the Kepler Titan, got assigned a 13429 WorkUnit with only 2100 base credit.
Finalized in about 1h23m, and performed at 167 ns/day. The GPU was not fully utilized, on average only 160W powerdraw
It netted 7922 points.

2nd run on a somewhat faster CPU:
13429 workunit with 2100 base credit completed in 52 minutes, performance of 244 ns/day.
70% GPU load and 155W powerdraw.
Net result 12876 points.
 
Last edited:
PPD performance is in line with a GTX 1070m (which oddly is slighly faster than a regular GTX 1070).
I would like to see these GTX Titans working on some big work units to get an apples to apples comparison.
 
Today I had 1 GPU stopped with error in device manager:
Device PCI\VEN_10DE&DEV_1B00&SUBSYS_119A10DE&REV_A1\4&3623647d&0&0008 was not migrated due to partial or ambiguous match.
Something about adding nvlddmkm service by device driver manager.

Anyone else experienced something similar?

Sigh...

Windows update again, wanted to update the device driver which in turn halted the FahCore_22.exe proccess..
How can I stop windows from updating the GPU device drivers?
 
How can I stop windows from updating the GPU device drivers?
I think you can do that from Local Group Policy (gpedit.msc), going into Computer Configuration, Administrative Templates, Windows Components, Windows Update. There should be an item about disabling hardware driver updates.
 
I think you can do that from Local Group Policy (gpedit.msc), going into Computer Configuration, Administrative Templates, Windows Components, Windows Update. There should be an item about disabling hardware driver updates.
Thanks.

This PC is running windows 10 pro, but I have other machines running windows 10 home edition.
Does it matter in windows 10 which version is running , this now works on windows10 home edition as well?
In windows 7 gpedit was available in the more advanced editions only I believe.
 
Thanks.

This PC is running windows 10 pro, but I have other machines running windows 10 home edition.
Does it matter in windows 10 which version is running , this now works on windows10 home edition as well?
In windows 7 gpedit was available in the more advanced editions only I believe.
Not sure if there even is gpedit on Home editions... I think you could try adding the entry in the Registry (which I attached) but it may not work, since apparently it has to be enforced through MMC due to it editing some other files.

On that topic, I think the 2004 update made driver updates optional. In my case, there is also a AMD driver update, which I don't know why it shows up, considering I always run the latest driver :confused:, but Windows doesn't install it unless I tell it to do so. And that's without using the GPO setting.
20201208-110244.png
 

Attachments

Last edited:
Yet again two machines stopped folding because of windows automatic updates.
I found this next method online to stop these updates:

In Registry editor, goto

HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows

Create a new "WindowsUpdate" key

In this new key, create another new key, "AU"

In "AU", create a "AUOptions" value (DWORD 32-bit), and set to 2, which is for "Notify for download and notify for install".

Reboot for changes to take effect.

Ofcourse, do this at own risk.
 
Last edited:
Yet again two machines stopped folding because of windows automatic updates.
I found this next method online to stop these updates:

In Registry editor, goto

HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows

Create a new "WindowsUpdate" key

In this new key, create another new key, "AU"

In "AU", create a "AUOptions" value (DWORD 32-bit), and set to 2, which is for "Notify for download and notify for install".

Reboot for changes to take effect.
WIN7 and Linux FTW ;)
 
Yet again two machines stopped folding because of windows automatic updates.
I found this next method online to stop these updates:

In Registry editor, goto

HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows

Create a new "WindowsUpdate" key

In this new key, create another new key, "AU"

In "AU", create a "AUOptions" value (DWORD 32-bit), and set to 2, which is for "Notify for download and notify for install".

Reboot for changes to take effect.

Ofcourse, do this at own risk.
Couldn't you just disable them from the Control Panel? :confused:
 
Couldn't you just disable them from the Control Panel? :confused:
I have unchecked "Restart this device as soon as possible ..." in advanced options, but still it forced an automatic restart.
I think Microsoft is now actively pushing the 20H2 upgrade to home users.
 
Last edited:
Here is a link to Microsoft documentation on restart behaviour:

Manage device restarts after updates

So, to delay rebooting if a user is logged on, AUOptions should be 4, while NoAutoRebootWithLoggedOnUsers is set to 1.

Whatever happened to the Keep It Simple-rule?
 
Last edited:
Here is a link to Microsoft documentation on restart behaviour:

Manage device restarts after updates

So, to delay rebooting if a user is logged on, AUOptions should be 4, while NoAutoRebootWithLoggedOnUsers is set to 1.

Whatever happened to the Keep It Simple-rule?
It's Microsoft??

I dunno... I just make sure I have the ability to check on any of my crunchers etc. just in case this/any stupid problem/s should happen... If I see one has updates, I make sure it restarts and doesn't stop half way because it's bugging me to sign in to something...

I wish Linux would be easier with the folding... Wouldn't use Windows for any of the crunchers then...
 
Whatever happened to the Keep It Simple-rule?
Long gone.

That aside, and until they fix this, maybe consider using the "pause updates" feature?
 
Long gone.

That aside, and until they fix this, maybe consider using the "pause updates" feature?
Yes, but one can only pause for a maximum of 35 days.
And then you are required to apply the updates, even if you don't want or need them.
I mean, I don't want "preview" updates, and I don't see the need for driver updates when my system is working well.
 
In attempting to get two GT 1030s running in an Optiplex 990 (which I couldn't, so any thoughts there are welcome), I noticed something else. The CPU is an i3 2120, and for a little while it was idle due to no assignments available for configuration. That's fine, it generates just north of bupkus anyway. BUT: the estimated PPD for the 1030 went up while the CPU was resting, higher than the total est.PPD when both are running together. So I removed the CPU slot and am only running the 1030. CPU usage is still around 60%, so my hypothesis is that there isn't enough juice in the i3 to do its own work and properly support the 1030's task. Does this sound right? If so, I may try to put both in the X58 rig I just brought down because FAHcore kept crashing. Of course, then I'll need to solve THAT problem.

[edit: spelling]
 
You need 1 CPU core per 1 folding GPU.
On low end CPU's, performance of GPU can drop if it hasn't got enough CPU performance to support it.

I also noticed there is a new version of FAH available, v7.6.21
 
Last edited:
"Quality" ahum Update KB4598242 since January 12, 2021 failed to install 10 times already.
I stopped the machine to manually download and install, but it just will not install.
I took the opportunity to update FAH to v7.6.21 and the machine is now up and running again. So far, so good.
 
You need 1 CPU core per 1 folding GPU.
On low end CPU's, performance of GPU can drop if it hasn't got enough CPU performance to support it.

I also noticed there is a new version of FAH available, v7.6.21

One entire core, or one CPU thread? I've got a GT 1030 attached to an X5670, and was wondering how much to dedicate, since the remaining CPU resources are on WCG. Relatedly, does a faster GPU require a commensurately faster CPU to feed it?
 
On my 4c/8t system I see the fahcore_22 process using 15% of CPU.
With one thread being always fully utilized , this means another thread is partially utilized (20% of a thread). I would count that as one core needed.

Second part of the question, intuitively I would answer yes.
 
yah, FAH been acting up on me last couple days. I should be getting 1.4 million per day. I gave up yesterday. but I will try again later today.

i click FAH, run, and the gpu just stalls, nothing, no folding of anything. really frustrating I can't figure it out. I tried reinstalling and everything. I figure w.e work assignment I got, just had an error and now I can't fold again until it resets... will try again soon, hopefully its reset by now... CPU is still folding fine, but those are like useless low amount of points.
Windows or Linux?
 
Last edited:
I see from a lot of posts everyone using this >> FAHControl - Folding@home Client Advanced Control app, can you tell me where i can access/download this app for my own use.
I've tried searching the Folding Forum - Index Page but cannot find anything there (maybe have to look harder).
Any help would be most welcome & appreciated.
 
Program files x86\fahclient\fahcontrol.exe
 
Back
Top