Graphics clocks WONT GO OVER 549MHZ even though my card's default base speed is 980MHZ

omonoiatis9 · Aug 10, 2015

CPU: i5 3570 (no issues)
GPU: Geforce GTX 660 (stuck at extremely low clock speed no matter the load and therefore overheats quickly under load)
OS: Windows 7

My issue is that even though I have a rig that should support Ultra settings easily for the game in question (they were the default suggested settings too), the pc crashes after 3 minutes of being in game at ultra settings due to what I imagine is overheating (GPU temperature goes over 100 degrees Celsius which is not normal).

The reason for this I imagine is that my clock speed for some reason is extremely underclocked, stuck at 549MHz (according to HWMonitor - also tested in MSI afterburner) under the heaviest loads, and won't ever go above it even though the base clock speed is almost double that (980MHz).

Any ideas why this is happening? The only way I can play the game without overheat crashing is at low settings, when the clock speed falls to 324MHz and temperatures are normal.

I did clean reinstall of drivers and everything. Before we go through the painstaking process of determining the exact problem, can we first assume that the problem is that the card is running very underclocked, and does anyone know any possible fixes for that?

Been trying to fix this issue for over 5 hours now, I'm desperate and I'd appreciate any help.

migronesien · Aug 10, 2015

Are the fans of your GPU working correctly? Monitor the fan speeds with a programm like EVGA Precision X or MSI Afterburner.

detrius · Aug 10, 2015

RMA your card if under warranty. Check that the fan is spinning. It should always be spinning and should be very audible when its hot. If the fan is working its probably that the cooler is not properly mounted on the GPU die. If your card is no longer under warranty, remove the heatsink, reapply themal paste then reseat the heatsink.fan.

omonoiatis9 · Aug 10, 2015

migronesien :

The fans are working perfectly fine. Like I said, I monitored every component of my computer in both HWMonitor and MSI afterburner. The only thing that acts weird is my card's clock speed which is extremely low even while under heavy load (gaming). And that causes overheating.

omonoiatis9 · Aug 10, 2015

detrius :

The fan is in fact spinning and audible. To make sure I even turned off "auto" mode in MSI afterburner and put the fan to work on max. It works fine. What is a GPU die? And how do I remove the heatsink and do all that stuff you mentioned?

migronesien · Aug 10, 2015

omonoiatis9 :

So the fans are giving 100% (hearable) and the card is still overheating even when it running low clocks?
Whats the idle temp of your GPU?

omonoiatis9 · Aug 10, 2015

Exactly. The idle temp of my GPU is 53-55 celsius degrees, running dual monitors. If that seems a tad bit higher than normal, it might also be relevant that I live in a pretty hot place. (35 degrees at the moment)

migronesien · Aug 10, 2015

omonoiatis9 :

That sounds pretty normal for idle temps with 35c outside. Since when do you have the problem?
Maybe try to delete your GPU drivers and get older ones. I know that there was a new GPU driver update from Nvidia recently there might be something wrong.

Otherwise if you still have warranty on your card the best thing is to get a replacement...maybe the chip broke in some kind of way.

Edit: Have you made sure that your card really gets that how and thats it's no a broken heat sensor? Because a broken sensor would lower the clocks the these levels, when it's reading 100celsius at the core.

omonoiatis9 · Aug 10, 2015

migronesien :

To be honest I don't know how long I've had this problem for, because the games I played until now were extremely lightweight so the card never had to even work at higher than 324MHz clock speeds. I tried a new game today that is slightly more demanding (but my rig should still be able to run it perfectly at max settings) and the clock speed gets stuck at 549MHz and overheats. For all I know the problem could be there from months ago and I never noticed. So my only option is technician/new graphics card?

migronesien · Aug 10, 2015

Maybe you dont saw my edit in my last post it came a bit late.

Have you made sure that your card really gets that how and thats it's no a broken heat sensor? Because a broken sensor would lower the clocks the these levels, when it's reading 100celsius at the core. Dont really know how to test it...maybe touch it to feel if it's really hot as fuck while under load or still cool.

omonoiatis9 · Aug 10, 2015

migronesien :

Yeah, I did check with my hand and it feels pretty hot under load. Don't think it's a broken heat sensor. Also the bluescreen I get after a crash from overheating is bccode 116, which relates to GPU problem. So that confirms it has to be the GPU. Memory readings look fine, highly doubt it's the power supply.

migronesien · Aug 10, 2015

Last option would be to get some screwdrivers and get the fans and the heatsink of the card itself and apply some new thermal paste. Maybe the hot climate eats through thermal paste pretty fast.
If that also doesnt work i really dont know what to do and would get a new card. You should get it replaced for free when it's still under warranty. Otherwise you have to buy another bad luck.

omonoiatis9 · Aug 10, 2015

migronesien :

Is this easy to do and where can I find thermal paste?

QSV · Aug 10, 2015

What brand of GTX 660 is it? MSI? EVGA? Gigabyte? etc. Exact model please.
You could try to go to the Nvidia control panel and set Power Management Mode to maximum, which normally fixes stuck low clocks.
But I think the low clock is a cause of your card overheating that easily.

migronesien · Aug 10, 2015

omonoiatis9 :

Not to hard id say. Just remove every screw you can find and dont loose them.
Thermal paste can be obtained in any hardware store or online.
It usually is like 5€ here per tube which lasts like a lifetime.

omonoiatis9 · Aug 10, 2015

QSV :

Not sure which brand it is exactly (not home at the moment). I didn't know I could control power management there, but I pray to god that it will fix it when I get home. However chances are that it needs maintenance/replacement. I will reply again once I've tried your suggestion

omonoiatis9 · Aug 10, 2015

QSV :

Tried that, didn't work. I also tried a different monitoring tool than HWMonitor, EVGA Precision and on it it says the clock speed is 979 (normal clock speed - maybe HWMOnitor was glitched and showed only up to 549?). It just overheats extremely quickly under even just medium load.

Any ideas what could be causing this given the fans are working fine?

Reaper_7799 · Aug 10, 2015

660 came out in 2012, unless you just bought it not too long ago...it's probably like someone else said earlier and has something to do with your thermal paste. I would try re applying thermal paste and see if that helps drop your temps and then you can get your clock speeds up. If temps go to high, the card will throttle and lower clocks so it probably has to do with heat more than anything else. Thermal paste is only like $8 or something and easiest and cheapest thing to do to trouble shoot it.

omonoiatis9 · Aug 11, 2015

Thing is guys... I did read a lot on thermal paste and stuff, and in every occasion where people re-do their thermal paste it simply drops the temperature by something like 10 degrees. Which does make sense of course, but the problem of my card is not that it has a stable high temperature. It overheats **constantly** under load and never stops climbing, as if cooling doesnt take place at all (and the fan does work). As far as I understand what the thermal paste does is it conducts heat better between the heatsink and the GPU chip, so unless there's absolutely 0 thermal paste left on my card.... The problem is more likely a damaged heat sink?

QSV · Aug 11, 2015

But if there is no or only little, it will cause exactly that. Or if the heatsink isnt seated properly (missing or loose screw?).
Use MSI Afterburner to check clocks and temps.

Jak_Sparra · Aug 11, 2015

Is it getting enough power?

omonoiatis9 · Aug 11, 2015

Jak_Sparra :

Most definitely. I had this PC built for me by a friend and the power supply we picked could handle all of the components and have some spare energy too.

omonoiatis9 · Aug 11, 2015

QSV :

Like I already said, I checked temps and clocks already in HWMonitor, in MSI Afterburner and in EVGA Precision. Clocks stuck low when under medium load because it overheats very fast.

Reaper_7799 · Aug 11, 2015

omonoiatis9 :

Two things....What you need to do is make sure that it doesn't have something to do with the card or the clock speed itself from the bios...you say the clocks are stuck low but will it ever go high? Before you start hitting high temps do you hit normal clock speeds? Do you have an on screen display that shows you real time GPU clock speeds and temps while you are playing or gaming? If not, use msi afterburner on screen display and let us know if your clock speeds go up before the heat causes them to go back down.

I know the temps go up quick...which again, if there is no thermal paste or it is all dried up, will cause it to shoot to 90 or above in a heartbeat. Do the temps raise up within the first minute you start playing? Most people will replace thermal paste before it is completely gone and therefore it could only make a 10 or 15 degree difference.

When I installed my hybrid, I had to completely take the card apart and remove the regular heatsink and put on the water cooling heatsink and that went fine but the power to my pump was not fully connected and the first time I ran it, as soon as firestrike turned on it started climbing fast...within a minute it was at like 95 degrees and it would have kept climbing if I hadn't shut it off. Took a min to cool down too. That's what will happen if no thermal paste or no cooling. Your clocks will get reset and drop way down if that happens too and stay there until you reset them and fix it...sounds like what you're experiencing.

EDIT: You're right....it could be the heatsink itself but it could be the thermal paste too.

Graphics clocks WONT GO OVER 549MHZ even though my card's default base speed is 980MHZ

Distinguished

Honorable

Honorable

Distinguished

Distinguished

Honorable

Distinguished

Honorable

Distinguished

Honorable

Distinguished

Honorable

Distinguished

Honorable

Honorable

Distinguished

Distinguished

Distinguished

Distinguished

Honorable

Distinguished

Distinguished

Distinguished

Distinguished

Share this page