I've had this happen several times now, and I don't know what to make of it.

Completely randomly the fan on my GTX 470 will run full bore at what sounds like 110%. Some times it will do it for half a second, other times it will do it for 5-10 seconds. If I check the fan speed in MSI Afterburner it does NOT report this increase in fan speed, which I find totally mystifying. For example, it'll show that its still running at 70% while the fan ramps up to deafening levels and then goes back down after a few seconds.

When I manually set the fan to 100% it still isn't as loud as when it ramps up on its own, so it seems like its running at at least 110% normal speed.

I've had this happen at the windows desktop and I've had it happen in games. I have no idea what could be causing it because my temperatures are stable and there are no spikes in temperature OR fan speed being reported by MSI Afterburner.

Its extremely irritating because its just SO loud. I'm currently using the 257.15 drivers with the latest version of MSI Afterburner with a custom fan profile. I've turned off speedfan just in case it was causing some sort of conflict, and the problem still occurs. I also do not have any PWM fans in my case, so no other fans should be changing speed. I'm 99% certain that it is the GTX 470's fan that is making all the noise because it has a pretty distinct sound.

I can't find any mention of this issue anywhere online.

I posted this problem here as well, so there is more information in that thread. Its very strange. I've tried disabling Afterburner, going back to stock voltage and stock clocks, setting the GPU power saving mode to "performance" rather than "adaptive". I have no other hardware monitoring programs or services running. It will happen totally randomly for like half a second, usually when I'm just sitting around browsing the internet, and its so loud that it makes me jump out of my chair. It'd be funny if it wasn't so irritating. ;)

Also, here are my specs (copied from my signature on the other forum).

CoolerMaster 690 II Advance - Gigabyte GA-EP45-UD3P - Intel Core 2 Quad Q9550 @ 4.0Ghz + Thermalright Ultra 120 Extreme
6GB DDR2 @ 942Mhz 5-5-5-20 1.9v (2x1Gb Wintec AMPX PC2-8500 & 2x2Gb G.Skill PC2-6400) - EVGA Geforce GTX 470 @ 750/1500/1850 (1.050v)
Sparkle Geforce GTS 250 1Gb Low-Profile (Physx) - Seagate 7200.12 500Gb SATA - Seagate 7200.10 320Gb SATA - Seagate 7200.8 200Gb SATA
ASUS VW266H 25.5" LCD - OCZ GameXStream 700W PSU - ASUS Xonar DX - Logitech Z-5500 5.1 Surround - Windows 7 Professional x64

may want to try re-installing the drivers?

Once you re-install the drivers I would also recomend using the OC SCanner software to scan for artifacts and ensure no hardware problems with the card too.

This challange your having is the same thing happening to mine.  One or both of my fans on the 470's goes all the way up to at least 100% cause it just gets so loud out of noware but on precision the fans are consistent.. It only happens when I game for more than 10 minutes then the fan will crank up all the way without warning but again precision says that fan speed has not changed?  How can this be?  I dont know what the problem is.

hi, i had the exact same thing happen, 1 of my 2 -470 cards fan would ramp up to full and could not be turned down by any utility.   i had to rma the card and evga sent me a new one . it's the only way to fix  the issue

After reading a suggestion somewhere online I decided to pop the cover off of my 470 to see if the fan wasn't plugged in properly.

First of all, I was amazed at how easily the cover can be removed. It pops right off with no fuss and it doesn't even have any screws in it, yet its still tight. Very convenient for cleaning.

Second, I was surprised to see that the fan's plug actually wasn't seated properly. It was "in" but I doubt all four pins were making perfect contact all the time. With the card installed and heating up and cooling down all the time its definitely possible that it could lose a connection at times. Maybe the fan control pin was losing connection but the power was still there, which would basically let it run full bore.

I'm really hoping that this fixes the problem. I really don't want to RMA my card. :(

Thanks for all the suggestions guys!

My card just did it for the first time that I know of, after three weeks.  This is very very odd.  I have read about this but never experienced it.

Keep us posted on your finding...

Bob

I think its the drivers because I think it happened to me this morning with a 260.

I had the exact same problem with my 470.  It would randomly ramp up to 100% for like 2-5 seconds and then go back to normal.

Weird issues for sure.  It hasn't happened since, so I cant really determine what the problem was.

It only happened once to me.  Can't seem to get it to happen...granted...no one seems to be able to.  All things aside...my card has, prior to the one episode, and since...been acting and reacting normally.  Very perplexing.  Not sure it's the card...

Were you guys running GPU-Z, Eleet Beta or Everest 5.5 in the background when this happened?  Pretty easy to replicate actually, these programs seem to interfere with NVAPI and its automatic fan control ability, I can open Eleet Beta and get 100% fan spikes within a minute.  Should look something like this:



Intermitten fan spikes aren't the worst thing that can happen though, some users (including myself) have had the fan stop completely.  Simply closing these programs out should eliminate any further conflicts.  If your fan stop it looks something like this (fan is showing 96% but its actually stopped and GPU temps steadily climb to ~93C):

Hey I've had the fan stop too, but with only precision running. It has done it in games (didn't hear it and thermal shutdown) and I actually caught it on video in OC scanner. It stopped the OC scan because I had the thermal protection active.
I also had the fan do the super fast thing once where coming out of precision I set the fan from manual 70% to auto and it went crazy, even though it was at about idle temp. And there was no measured spike as you mentioned. I had to reboot to get it to go back to normal.
I guess what I'm saying is precision might be one of the culprits as well.

I've had that happen a number of times. And the fan spikes don't show up in afterburner or precision.
I have a custom fan profile in afterburner and had precision running. In precision the fan was locked at 50%, so the 2 could have been conflicting. So far i don't think its happened when i close precision or set the fan to auto.

Come to think of it...I remember that I was responding to another thread on the board when it happened to me.  I run Afterburner 1.6 final in the background at all times.  In response to someone's question about freezing...I opened E-Leet and checked what my default bios speed was for PCIE on my E758 (99.2)...I distinctly remember looking at it and responding to the other thread when I heard the fan ramp up.  I quickly looked at Afterburner in my tray and fan speed said 86, while my GPU temp was 44.  Very likely there is something with Precision, Afterburner and the like...Unwinder writes all the programs.

Bob

I can remember when I first installed precision that I couldnt have precision and rivatuner installed on the same computer.I always had some kind of problems.You dont need 10 different overclocking programs on your computer.precision is enough by itself.

bobmitch
 
I quickly looked at Afterburner in my tray and fan speed said 86, while my GPU temp was 44.  Very likely there is something with Precision, Afterburner and the like...Unwinder writes all the programs.

 
No need to involve me into it please. I'm not developing E-LEET, GPU-Z and Everest causing known conflicts with NVIDIA driver fan control implementation due to low-level access to I2C bus instead of using NVAPI. Fan control layer of both PRecision and Afterburner use NVAPI only for maximum safety.

Unwinder...

Not looking at involving you.  Thought someone may have an answer to a possible .dll conflict that you may be aware of.  I appreciate your input...
 
In other threads...it seems that E-Leet may be the culprit

Unwinder

bobmitch
 
I quickly looked at Afterburner in my tray and fan speed said 86, while my GPU temp was 44.  Very likely there is something with Precision, Afterburner and the like...Unwinder writes all the programs.

 
No need to involve me into it please. I'm not developing E-LEET, GPU-Z and Everest causing known conflicts with NVIDIA driver fan control implementation due to low-level access to I2C bus instead of using NVAPI. Fan control layer of both PRecision and Afterburner use NVAPI only for maximum safety.

+1
 
I've run Precision and now AfterBurner for over 2 months with GTX 480s and the only time I had the fanspeed problem was after Eleet Beta was released and I was running it concurrently.  Closed Eleet out, only ran it while necessary and haven't seen any similar fanspeed problems since.  There's been numerous nearly identical reports with users keeping GPU-Z and Everest open while monitoring temps/voltages and Nvidia has directly acknowledged the problem to be exactly as Unwinder stated.  Precision and AfterBurner "follow the rules" by sticking to NVAPI and shouldn't run into these problems on their own.  Here's Nvidia's response to a Legit Reviews article about similar crashing/fanspeed problems:
 
http://www.legitreviews.com/article/1264/1/

"Upon low-level examination of the GPU-Z application, we have determined that it is not using our NVAPI for all temperature/fan control calls, but is instead banging registers directly. This is interfering with our fan control and causing boards to crash when GPU-Z is run in the background. We need GPU-Z to use NVAPI to avoid these conflicts..."- NVIDIA

For those experiencing these problems make sure you close out all monitoring software other than RT/Precision/AfterBurner and then see if the problem replicates itself.
 

Would CPUID Hardware monitor fall under those bad programs?

thetacowarrior


Would CPUID Hardware monitor fall under those bad programs?

Pretty sure HW Monitor uses the same SDK and is made by the same folks as CPU-Z (Eleet uses CPU-Z SDK for sure) and GPU-Z, so its very possible.  Its definitely reporting temps/fanspeed for GTX 480 but I'm not sure if its using NVAPI as it should or if its using low-level I2C bus as Unwinder mentioned. Try closing it out and see if your problems persist? 

chizow


Were you guys running GPU-Z, Eleet Beta or Everest 5.5 in the background when this happened?  Pretty easy to replicate actually, these programs seem to interfere with NVAPI and its automatic fan control ability, I can open Eleet Beta and get 100% fan spikes within a minute.  Should look something like this:



Intermitten fan spikes aren't the worst thing that can happen though, some users (including myself) have had the fan stop completely.  Simply closing these programs out should eliminate any further conflicts.  If your fan stop it looks something like this (fan is showing 96% but its actually stopped and GPU temps steadily climb to ~93C):

i see your gpu temps spike to like mine..iv had all these programs closed then i ran fur mark and temps would still spike like that..so odd

thetacowarrior


Would CPUID Hardware monitor fall under those bad programs?

To follow-up, it does look like it messes with temp reading but I didn't get the fan speed spikes with HW Monitor.  Just ran it for about 30 mins and here's the results:
 

Ok thanks, It has acted up without it running, I actually just had both my cards approved for RMA. I was just wondering if I should avoid future use of the program.

On a related note, what would you guys then recommend for CPU temp monitoring and recording? The whole reason I use Hardware Monitor is to watch my CPU temp, and more importantly, record its max if I'm doing testing or gaming or something, because I can't watch it all the time.

I've had the same issue, but it is different than the way you guys describe it. I usually set my computer to sleep mode rather than shutdown and I have found it running (woken up) sometimes on its own with the fan running full bore. I can turn it back down through precision (which I thought was the culprit). The first time was in the morning when I got up I heard it and the second time was after work I came home to find it running. Not sure what woke the machine up but I have tested putting it to sleep and waking it up and am unable to repeat the problem. I installed the 197.77 and haven't had the issue till now but I am going to test a few more things.

I have also seen that Precision is not reporting the fan speed at full bore it usually shows it at 30-40%. I can turn it back down by setting it to manual and then back to auto. :S baffling

thetacowarrior


Ok thanks, It has acted up without it running, I actually just had both my cards approved for RMA. I was just wondering if I should avoid future use of the program.

On a related note, what would you guys then recommend for CPU temp monitoring and recording? The whole reason I use Hardware Monitor is to watch my CPU temp, and more importantly, record its max if I'm doing testing or gaming or something, because I can't watch it all the time.

Ya I would wait a few days if you can before you RMA as it may just be a software monitoring issue.  Would save both you and EVGA some expense and also eliminate any downtime on your end. 
 
As for software CPU monitoring, you can try RealTemp or CoreTemp.  They should be better as they don't poll GPU temps or voltages and shouldn't interfere with NVAPI. 
 
http://www.techpowerup.com/realtemp/
http://www.alcpu.com/CoreTemp/
 
 

faudau


I've had the same issue, but it is different than the way you guys describe it. I usually set my computer to sleep mode rather than shutdown and I have found it running (woken up) sometimes on its own with the fan running full bore. I can turn it back down through precision (which I thought was the culprit). The first time was in the morning when I got up I heard it and the second time was after work I came home to find it running. Not sure what woke the machine up but I have tested putting it to sleep and waking it up and am unable to repeat the problem. I installed the 197.77 and haven't had the issue till now but I am going to test a few more things.

I have also seen that Precision is not reporting the fan speed at full bore it usually shows it at 30-40%. I can turn it back down by setting it to manual and then back to auto. :S baffling

Is your computer going into S3 Sleep or just Standby where the monitor and GPUs go into Standby mode?  If your computer is turning back on and fans are going full blast that sounds like it could be a bigger problem where for some reason your GPU temps are spiking up while the fans are powered down, causing a system event alert which would wake your PC from Sleep.  Once awake it would realize your GPU temps are way too high and blasts your fanspeed to compensate as it should.  Did you notice high temps in the 95C range when it woke up by itself?  Either way this sounds more like an S3 Sleep stability issue more than anything. 

Just an update guys. Since I reseated the GPU fan plug inside the heatsink the fan has not ramped up on its own, not even for half a second.

If anyone is having this issue (specifically the issue of monitoring software not showing the random fan spikes) I'm willing to bet that its related to the fan not being plugged in fully.

It takes less than a minute to pop the cover off of a GTX 470, so its definitely worth looking into.

I'm surprised this isn't a more common problem because the fan wires are really short and wrapped around in a strange way. They barely reach the power connector on my card.

Is your computer going into S3 Sleep or just Standby where the monitor and GPUs go into Standby mode?  If your computer is turning back on and fans are going full blast that sounds like it could be a bigger problem where for some reason your GPU temps are spiking up while the fans are powered down, causing a system event alert which would wake your PC from Sleep.  Once awake it would realize your GPU temps are way too high and blasts your fanspeed to compensate as it should.  Did you notice high temps in the 95C range when it woke up by itself?  Either way this sounds more like an S3 Sleep stability issue more than anything. 

S3 sleep mode is what I have been setting it in.. unfortunately the 2 times it happened I was away and so if there was a spike I couldn't tell due to too much time going by. But that may explain the problem. I have only seen it happen twice so I'll have to wait till it reoccurs to be able to tell. So far its been solid and working well.

my card does this also. it is getting old having it come on an do this. i don't want to rma the card but this really is  pretty bad an a fix for it needs to be made

Thank you chizow, I will look into those

chizow, I got real temp, and it still has a little box with GPU temp, is that going to mess up my video card you think?

I am just using coretemp right now, it seems to work, but I would like to try real temp If it doesn't mess with the video cards.