• 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.

Nvidia GTX 470 - heat sensor freaking out

rebirth

New Member
Joined
Aug 5, 2013
Messages
12 (0.00/day)
System Name LOL
Processor Intel I7 2600k @ 4.5ghz
Motherboard Asus P67 Pro
Cooling Corsair closedloop water cooler
Memory 16 GB (4x4) Corsair 1600mhz
Video Card(s) GTX 470
Storage SSD's and standard HDDs (RAID)
Display(s) 2 x 24 inch + 42 inch HDTV (only use one monitor for games though)
Case Akasa Mirage (a good few years old)
Audio Device(s) Asus Xonar D1
Power Supply Corsair 850 watt
Software Win 7 64bit Ultimate
Benchmark Scores It boots up faster than the time it takes to even put the kettle on. :(
Hi. Today after turning on my PC I started to crash after about 5 minutes even when idle. I tracked it down to my GFX suddenly overheating to 139 degrees with no GPU load!

I took the card apart, cleaned the fan (including fins), took off the heatsink, cleaned off all the dried up and brittle compound and replaced it with a generous helping of Arctic Silver.

When I power it up all has worked fine and haven't crashed since. After I powered up i was getting a reading of between 38-45 degrees, which is good for this particular card (it has always been hot).

I have now noticed that my temperature is fluctuating between 176 and 48 degrees. This reading appears to be changing between these temps on every update, so cannot possibly be valid.

I just had a moment when the card's fan kicked in full speed for a second and then go back to normal.

I have tried three programs to monitor the temp and they all report this fluctuation.

Is the sensor malfunctioning? Is it time to invest in something new?

Screenshots taken less than a second apart:
2807i3s.gif

30cc2ew.gif


Note the GPU temp readings.
 
Last edited:
Really hard to tell you what to do here but, I would run the card as normal and feel if the PCB is getting super hot.
If so I would take it back apart and check to see if the paste is making good contact.
Also have you updated drivers lately around the time this started happening?
If so go back to the one you was using and see if it still happens.
 
looks like the sensor is malfunctioning. may just be thats its soldering has come off.

most temps are measured in NTC/PTC thermistors. i.e. higher temps less/more resistance.

so if your sensor is a ptc, and one of the contacts go loose, it will show weird results. if on chip sensor, out of luck. only baking can help.
 
I don't know where the sensor is, but it is the Zotec GTX 470 reference card.

This one I believe: http://www.zotacusa.com/geforce-gtx-470-zt-40201-10p.html

It's running this bios: 70.00.21.00.03

I just tried to update to the newest bios found here: http://www.techpowerup.com/vgabios/index.php?manufacturer=Zotac&model=GTX+470

But I got this from nvflash:

WARNING: Firmware image PCI Subsystem ID (19DA.1153)
does not match adapter PCI Subsystem ID (10DE.079F).
ERROR: PCI subsystem ID mismatch

Scrap the last bit: I obviously used the wrong file, thinking they were all for the same card.
 
Last edited:
The bios is not the problem.

Can you answer my first post?
 
The bios is not the problem.

Can you answer my first post?

It isn't a driver problem. Nothing was altered, but the PC wasn't on for a couple of days before this problem appeared. I was actually running the 3.14 and did (since this problem showed) installed the 3.20, but this hasn't altered the problem.

I also installed nvtools, but it isn't running. I'll uninstall it (I got revo) just in-case there is a conflict.

BRB
 
Asylum - It would seem nvtools messed something up even when it wasn't running. I also uninstalled the performance module (overclocking utility) and something quite strange has happened.

2gxqmc3.jpg


I didn't even alter the clocking settings and simply used the monitoring program, which appeared to mess something up.

For now I am relieved. I shall have to put this all down to the card needing a good clean and a change of grease, as well a strange line of events. I hope this is now stable.

Note: That screenie was taken just after reboot. All the clock settings went back to idle straight after and the temps are reading as they should be. I'll have to go into BF3 for a good test run to make sure.
 
Last edited:
Sadly, the last post may have simply been a coincidence.

The fan just went to full speed, and just before I black-screened it was bouncing between 4 and some ridiculously high temp.

I'm guessing something is wrong with my card. :(

All is reporting fine in GPU-Z since another (forced) reboot.

So guys, I'm gaming on a single HD monitor. Does the extra cores and power of a 770 over a 760 benefit someone like me, considering the price difference? Taking life expectancy into consideration too.
 
Last edited:
Just to let you know I think I have fixed this.

Uninstalled everything Nvidia through Revo and removed all traces. Rebooted into safe mode and cleaned nvidia drivers and physx with Driver Fusion (just to be sure). Installed 320 drivers with clean enabled (minus Experience), and have run geek3d physics test and furmark on multiple occasions and the card is running as expected. Been a couple of hours now.

I'll probably buy a 770 before BF4, but hopefully this 470 can be given to a friend as a working card.

Cheers for the tips. Happy benchmarking...

P.S. Avoid nvtools too.
 
Thanks for the update.
 
may have used too much arctic silver. Its very capacitiiive.
 
may have used too much arctic silver. Its very capacitiiive.

If this was the case then I would imagine something would have burned out, and it wouldn't explain why the problem was solved without any further physical handling of the card. I have also used the stuff for years without damaging anything, so tend to measure it quite well. Much of the surrounding circuitry was also protected by a plastic cover and only the die was visible and protruding. I did actually look for overspill when the heat-sink was firmly reattached and even went around with the back-end of a matchstick.

I believe the card was genuinely overheating and that I fixed this problem. I mentioned earlier that it was reporting a valid low temp. Very soon after I updated the GFX drivers and installed NVTools, and something (not entirely sure what) caused this fluctuation bug, which caused a similar crash as before. This is what threw me off and pushed me towards potential hardware failure. I didn't notice the temp fluctuating the first time, I just took the procedure to clean the card after it black-screened a couple of times.

Since my last post everything has worked fine, but who knows? Maybe it was something else entirely, but right now I am just happy to get my solid rig back. :)
 
Back
Top