• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

GPU Folding Question

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

PaciFIST69

Member
Joined
Oct 14, 2006
Location
Clemson, SC
I'd like to fold my x1900 but I'm a little worried about heat. At full gaming load, I have to use 100% fan to keep the card at 70C. Problem is 100% fan sounds like a cross between a fire engine and a banshee and I don't think my roommate could tolerate that 24/7 (who am I kidding, I couldn't either). So, x1900 folders, how much heat does your card put out while folding?
 
Headphones... Really good headphones.
OH, a room mate issue? Get two.


You could i suppose try and find a better air cooling solution for the card that offers slightly better temps for a few less dB.
 
Sleepy_Steve said:
Headphones... Really good headphones.
OH, a room mate issue? Get two.

So I would sleep in the headphones, not hear the alarm in the morning and flunk out of college. And it would be all your fault ;)

I'd also rather not dump any more money into the comp before I get a dx10 card. Ebayers don't understand the value of a good aftermarket heatsink as I found with my 7800gtx :mad:

I was wondering, though, might folding produce less heat than a game so I wouldn't have to use 100% fan?
 
Why do you have to keep it at 70 degrees. ATi cards run without problems into the high 90s. My x1950 XTX runs at about 78 degrees (21 ambient, 695/1003) at 50% fan and is barely audible. However, the real claim to fame of the x1950 is its excellent stock cooling system. Anyway, folding doesn't heat the cards as much as some games.
 
Buy a better cooler. I got an Excellero for $15 from SVC, dropped my temps 20C on my X1950 Pro. GPU folding tops at 48C or so, games at 51C. This is with a slight OC. 100% fan speed is quieter than the stock fans at idle.
 
ChasR said:
Why do you have to keep it at 70 degrees.

Hmm... I guess I didn't really have a reason for 70C, just felt it was a good temp for a card. It would be nice if Ati would give us a reasonable maximum temperature. I think the cutoff point is like 120C or something ridiculous like that. I'm running atitool 3d view stock clocks at 50% fan (bearable) and it seems to level out at 94C after 15 minutes and it seems pretty stable. I'll give it a shot!

One more question: what is a good way to permanently stop one folding instance (service install) so I can free up a core for the gpu?
 
Permanently , as in delete it?

First stop the service by start->run->services.msc. Stop which ever service you want to delete.
then start->run->regedit->HKey_Local_Machine->System->CurrentControlSet->Services->Name of FAH service (i.e.FAH1). Simply delete all the entries under the service your delete.
then
go to the fold of your FAH service. (i.e. c:\fah\fah1\) and delete everything. Now that service edition of FAH is bye bye. Enjoy GPU folding.
 
before you delete the instance, let the current wu complete

Change the folding flags with regedit add the "-oneunit" flag so it stops when it finishes the current wu.

Then you can delete the instance as shel described.

Just deleting a wu is bad for the science ... that wu is part of a series and it cannot be reassigned until the preferred deadline is reached which could be days or weeks depending on the project. The next wu's in the series can't be assigned till the reaasigned wu gets turned in. So deleting a wu could cost the project weeks. If others delete wu's in the series same thing ... so project delays could quickly become months depending on the project.

The open betas for gpu and smp have very short deadlines (a few days) but most cpu wu's are measured in weeks at least.
 
imho, try to keep it under 90c. depends on your rig's airflow, I think around 50-70% stock fan will keeps the temp under 90c when folding. all my GPUs are just under 80c (78c or so) when folding, and those w/ stock fans, I think I've set them at around 55-65%. or better yet, get a aftermarket hsf like Zalman (I'm using one), really helps on both temp and noise!!
 
Thanks for the help everyone!

Ati tool seems to heat up cards more than anything else, so I hope folding will be <90. I have decent airflow with a slot blower right below the card and a 120mm exhaust right above. I also have an 80mm tornado that can blow across the card toward the back of the case, but I don't have it running right now (noise issue again :-/) As for the aftermarket heatsink, I don't have the money right now for a good one (accelero is $28 shipped at svc btw). I put a vf900 on my old 7800gtx which didn't add anything to the resell value - should have saved crappy stock heatsink :bang head . I'm going to sell the 1900 in a few months but it will be folding till then. I'll be sure to use -oneunit before stopping the client and I'll let you all know how the temps turn out.
 
i leave my card at 2d speeds for folding, (8 minutes a frame) as it is not running on my personal workstation and the user has not moaned about the heat/noise yet!

i think running it at 3d speeds would be a bit different if it was 24 hours a day

lee
 
My observed difference between OC'd 3D speed (78) and 2D speed (54) is 24 degrees.

The -oneunit flag will not work with a stanford service install or any install that uses -svcstart to start the service. If you have a Stanford service install and wish to stop folding on an instance, you really only need to stop the service for that instance using services.msc. Then run the FAH executable with the -config and -oneunit flags. Answer yes to discontinue service and leave everything else as is. The FAH service entries in the regisrty will be uninstalled for that instance and when the current WU is completed FAH will terminate and close the terminal window. You really don't have to uninstall the instance, just let it sit dormant. You never know what may come up that might make 2 instances attractive again.
 
Ok... I've used a hybrid of ideas here. I added the -oneunit flag to core 1 which shut down the client after it was finished. But in a brainless moment, I rebooted the computer, so it downloaded another unit (ribo! :santa: ) . So now I've taken the -oneunit flag off core 1 and added it to the nearly completed core 2. Then I went to services.msc and changed core 2 startup from 'automatic' to 'disabled.' So now it will stop after it finishes its unit and it won't start up again even if I reboot the computer. How does that sound?
 
That will work for a Wedo one click install which starts with -srvany.

On Stanford service installs, -oneunit appears to work, the client stops writing in the log, but svchost will notice the service stopping and automatically restart it. That causes a new WU to download and start processing but nothing is written into the log. This will happen with the service set to manual or automatic.
 
For all the people with a GPU client running, is the clock in the client becoming inaccurate? I'm not sure if the clock is running fast or slow, but it is more than 6 hours off real time. And it's not a time zone differential either, I think it's running fast!
 
OK, hit a little snag here. After installing the GPU client, it gives me an error message: "The application has failed to start because d3dx9_30.dll was not found. Re-installing the application may fix this problem." So I've tried reinstalling, I've upgraded to the latest display driver, I've ditched CCC, I've tried increasing the priority of GPU client in case my other instance was interfering, and I've stopped my other instance. I always get the same message about 10 seconds after pressing 'OK' on the GPU gui install. Any suggestions?
 
Back