• 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.

GPUs keep downclocking

T4rd

T4rd

Fully [H]
2FA
Joined
Apr 8, 2009
Messages
21,148
For some reason here lately one or both of my GTX280s are downclocking after almost every WU. I posted a guide in the "EUE on GPU client"thread a page or two back (I'm on my phone now or I'd paste it in here). I tried that out and it didn't seem to help at all. I even did the driver emulation for older drivers in Rivatuner to force 3D clocks (as the guide says to do if you look it up). The only thing that's changed recently is I got a new 2nd monitor. I've always had two monitors though. It seems the only way to get the clocks back up to "performance 3D mode" is to either restart the whole PC or to enable, then disable SLI again. Has anyone else had this issue and figured out how to fix it?

Also, did you know you can fold with SLI enabled on both GPUs? I tried this for a while too but they both kept downclocking still. I thought you couldn't fold on multiple GPUs with SLI enabled? If I can do this and keep the clocks up, it's preferable as I like to game on this rig too.

Thanks in advance for any help you can provide!
 
You can fold with SLI enabled (at least on the 9800GX2).

Can you set an OC profile in the nTune, and force it to keep (while folding)?

Also, do you have any power saving options in the windows power settings (specifically the PCIe power management)?
 
I think it was the 185 version and later that made folding while SLI was enabled possible.
I still keep SLI disabled while folding since it's a simple change in CP.

I haven't seen the downclocking problem you have though on my 285s.
I use EVGA's Precision for both OC and fan control.
 
I'm using Rivatuner and yes I can create and select an OC profile that it's supposed to apply upon startup, but the OC profiles only apply to what rendering mode it's in (i.e. 2D, low power 3D, and performance 3D) and I can't force it to use a specific rendering mode all the time. This is the guide I was talking about and it didn't work for me.

I'm using Win7 and I have the power settings to never turn anything off except my monitors after 15 mins. I know that's not causing it because I've always had these settings. I don't see any settings for PCIe power management..?

I'm using the 191.07 drivers.

Thanks guys.
 
I know it's not english, but:


that is what I was talking about (and it takes forever to create/upload these things, lol).

Also, are you daring enough to BIOS edit away your 2D downclocks?
 
Yeah, I went to that menu it shows "PCI Express > Link State Power Management > Setting: Off".

I'm no stranger to my BIOS, I play in mine occasionally to see what OC I can get and try to get it stable. So yeah, I'm definitely "daring" enough. Having an A+ certification doesn't hurt either :D. However, I haven't seen anything to change my card settings in the BIOS other than voltages to it.

Edit: I also found out that even when I start a game up (I just got L4D2 and played it), it even sticks to low power 3D clocks too! WTF? A game can't even get it to go to performance 3D clocks! The game ran pretty choppy too. of course.
 
Last edited:
Alright, last night I tried totally uninstalling Rivatuner since I thought it might be because I have the newest card drivers (195.62) and Rivatuner kept complaining it hasn't been tested with those drivers so it might not be compatible. I couldn't find a newer version than what I had though, which was version 2.24. I've trying using nTune to force clocks, and it still downclocks.

I think I may have found a clue though, it seems every so often my screen flickers and then I get a tray message that says the "GPU had to do a recovery" or something to that effect. Would an OC on the cards be causing this? I have my shaders clocked at 1500 MHz from 1300 MHz on my GTX 280s. Does anyone else here fold with 280s and tell me if that's a safe OC? I've been running them at those speeds for months before I started having this issue though. Maybe I should up the voltage a little to the cards?

Thanks again for all the help!
 
Yes, OC will cause the recovery error. Downclock the shaders.
 
Yea I didnt realize you had an OC.

Put it at stock, and then go up slowly.
If the GPU is getting too hot, then it will put it in 2D mode regardless of what is running.
What do you have the fan set at?

I set my fans to 70% on computers that are in public areas and 100% for anything in my server room.
 
I keep the fans on both cards at 100% all the time, they're not really that loud. It keeps the temps at around 75°C - 80°C on both cards. I'll try again from stock speeds and see if it still happens.

Thanks again.
 
T4rd said:
I keep the fans on both cards at 100% all the time, they're not really that loud. It keeps the temps at around 75°C - 80°C on both cards. I'll try again from stock speeds and see if it still happens.

Thanks again.
Click to expand...

80c is too hot especially if thats the temps when in 2D mode.
 
No, that's the temps while folding with the shaders @ 1500 MHz (3D performance mode). Right now one of mine is stuck in low power 3D mode (800 MHz shader) and it's at 63°C. Now that I think about it, I think that's the bottom card and it's prolly getting all the heat dumped on it from the top cards fan. So I'm sure it's a heat issue at least with that card. I'm about to enable/disable SLI to reset the GPUs and try a safer clock closer to stock.

Thanks!
 
Last edited:
Alright, it seems stable with a 100 MHz OC so far on both cards. Can I run both these cards with different clocks with SLI enabled and still game/fold on them? It lets me do it, but I'm not sure if that's recommended as I thought with SLI enabled both cards should run at same speeds.

Also, does anyone know what port(s) the GPU client uses to upload WUs? I'm setting up QoS on my router and I want to assign F@H traffic to lower priority so it doesn't affect my VOIP phone and games.

Thanks again!

Edit: GRR, It just did another GPU recovery and downclocked. This is the exact message in the system event log: "Display driver nvlddmkm stopped responding and has successfully recovered." I'm not sure it's temp related now as the cards were still only around 80°C when it happened. Does anyone know at what point the GPU forces the downclock due to heat? Or have any other suggestions?
 
Last edited:
I also didn't know you were running OC'd.
One of my 285s, (five of them :p) doesn't like the shader set @ 1620 and would downclock during a wu.
It's been running for months stable @ 1566. The rest are set @ 1620, so it was just one that didn't like the higher OC.
 
Kowan said:
I also didn't know you were running OC'd.
One of my 285s, (five of them :p) doesn't like the shader set @ 1620 and would downclock during a wu.
It's been running for months stable @ 1566. The rest are set @ 1620, so it was just one that didn't like the higher OC.
Click to expand...

Yeah, I guess I forgot to mention that minor detail, heh. It's just that I've been running them OC'd to that for the last couple months and they just started doing this the last week or two. I have them both running stock clocks for everything right now just to see if it still happens. If it doesn't then I'm chalking it up to a software issue of some sort, which I think it is, because it still did the recovery with just a 100 MHz OC on the shaders. Any card should be able to achieve an ~6% OC with no issues. One thing about this though; should I OC the shaders separately from the core clock or keep them linked? I did have them linked for a while, but can't remember if I unlinked them around the time I started having this issue.

I know it's normal for these cards to top out at 80°-85°C which is what my bottom card is hitting even at stock speeds with the fat at 100%. I'll see in the morning though when I check on it.

Does anyone still know if the clocks should be the same in SLI? One of my cards has a BIOS OC, so I dunno if I should match it with the reference card.

Also, what exactly is the hard limit temp for these cards before they force downclock?

Thanks again, guys.
 
About 90-100C, iirc.

How about switching around the order of the cards (physical slot locations), or removing one?

See if it works then.
 
Im not sure what the temp limit is.

What is the setup? ie Case, mobo, cpu, ram. What kinds/how many fans?
What is the ambient temp?

Ambient temps are really the larger issue at least for my setups.
Once I get those under control, the the GPUs are fine.
 
nitrobass24 said:
Im not sure what the temp limit is.

What is the setup? ie Case, mobo, cpu, ram. What kinds/how many fans?
What is the ambient temp?

Ambient temps are really the larger issue at least for my setups.
Once I get those under control, the the GPUs are fine.
Click to expand...

It's my primary rig in my sig except with a 3.8 GHz OC as of last night. My case is a Thermaltake Armor+ with 3 case fans, one of which is a 230mm fan on the side panel. Rivatuner read ambient tems at 45°C in my case.

Shouldn't be too bad, right?
 
I'm running with the shaders unlocked. It wasn't temps that caused my card to downclock.
It ran 1620 for months then decided it needed a break. :p

My cards would run near 80C in the summer, but never got above it.
 
I had a downclocking problem on my BFG GTX 275 OCs and what I ended up doing was removing the drivers, run Driver Sweeper and then reinstalling the drivers. I disabled SLI and enabled PhysX to work around the dummy plug thing for now.

Both cards are currently running at a core of 648 and 1440 for the shaders but had the shaders up to 1566 when I was running just one card.
 
Cool, thanks a lot guys.

I already uninstalled, ran driver sweeper and installed the new 195 drivers (I originally had the 191.07s). They seem to be doing a little better now that they're running at near stock speeds. I think my bottom card is just getting roasted by my top cards fan running at 100%. I wish I had a longer SLI connector or I'd move it down to my next PCI-E x16 slot and give it a bit more breathing room. I just want to rule out the heat issue before I start messing with software anymore.

Thanks again, guys. Still open for suggestions!
 
I guess it might be a heating issue now. It weird because both of my 280s have a reference cooler, but my bottom card runs about 6° to 12°C hotter even when both of my cards fans are at 100%. I guess my top one is getting more air circulation from my side panel fan though which is keeping it cooler.

I kept my basement door cracked last night to let the cold air in and my GPUs were running at a nice 60°C at full load and so was my i7 @ 3.8, lol. I think I'll start doing that more often now, I just gotta get a screen or something for my door. Anyways, no GPU driver recoveries over the night, so I assume that's it now. I did disable SLI though, so maybe that had something to do with it? I'm getting that Nexus fan off Swoody in his DC FS thread and will keep that on my 2nd card, so hopefully that'll help.

Mucho thanks to everyone.
 
Last edited:
You must log in or register to reply here.
Back
Top