Replacement for Geforce GTX 970 HSF assembly?

[serraphine]

I bought an MSI Geforce GTX 970 graphic card at the beginning of 2016, and it has served me well the past two years in a Windows 7 setup.

Lately in the past two months though I've been experiencing occasional video framerate and audio stuttering. The game being played might crash, or even at Windows desktop there will be audio stuttering when watching a video, be it file-based or Youtube. It may eventually crash the entire system with an unseen blue screen (not sure why it doesn't show up) and reboot.

The latest BSOD had a summary log like

Problem signature:
Problem Event Name: BlueScreen
OS Version: 6.1.7601.2.1.0.768.3
Locale ID: 18441

Additional information about the problem:
BCCode: 116
BCP1: FFFFFA8039D084E0
BCP2: FFFFF8801014605C
BCP3: FFFFFFFFC000009A
BCP4: 0000000000000004
OS Version: 6_1_7601
Service Pack: 1_0
Product: 768_1

Files that help describe the problem:
C:\Windows\Minidump\043018-14726-01.dmp
C:\Users\USERID\AppData\Local\Temp\WER-26956-0.sysdata.xml

Read our privacy statement online:
http://go.microsoft.com/fwlink/?linkid=104288&clcid=0x0409

If the online privacy statement is not available, please read our privacy statement offline:
C:\Windows\system32\en-US\erofflps.txt

If I use BlueScreenView program to inspect the recent mini-dump files, it suggests a fault with dxgkrnl.sys, dxgmms1.sys, nvlddmkm.sys, or including ntoskrnl.exe. Various graphic driver updates through the months did not seem to stabilise this.

With HWMonitor, I can observe my GPU maxing at 82*°C with the fans spinning 2800 rpm + In reality, this was common in the past and did not present problems with gameplay.

However today after giving my graphic card a wiping and vacuuming I observed the (double) fans and it somewhat appears the left fan might be spinning slower than the right fan. I wonder if there might be a lopsided overheat problem? The HSF assembly looks to be rather monolithic and unique in design, and does not appear the parts are conveniently as replaceable like regular casing fans.

Are there shops that sell/change entire HSF assemblies? Or those that know how to dismantle and service the individual components?

 

[Outramkia]

The only third party gpu hsf i know is the arctic cooling accelero series. Can buy from amazon. Else can try use liquid aio and mount using nzxt g12. Or you can try to seek out any model number from the hsf. And ebay them out.

But first and foremost ensure compatibility

 

[serraphine]

But first and foremost ensure compatibility

Yes that is the big question; Is there a standard set of specifications for dimensions that these HSF designs adhere to, so that they are interchangeable? Personally when I look at my older graphic cards they appear different from each other.

 

[singupingu]

Gpu mounting holes around the gpu core itself is mostly standardized these days. Using NZXT Kraken G12 and buy a 120mm aio to cool it is feasible but slightly costly.

Which model of msi gtx 970 are you using? ARMOR (black white) or Gaming (red black) series?

Sent from OnePlus ONEPLUS A3003 using GAGT

 

[royfrosty]

Less than 3 years old why dont rma it? Still got chance to get the later gtx10 series equivalent card.

 

[wwenze]

Remove fan, tie a 120mm fan

 

[clon22]

You can download afterburner to test if the fans are running properly too.
http://download.msi.com/uti_exe//vga/MSIAfterburnerSetup.zip
If still in warranty, just send it down to our service center for RMA.
(You can check the warranty date of our CT sticker at the back of the card to see if it's still in warranty.)

 

[Krabs.]

my graphic card fan spoil , unable to spin unless I jumpstart it with 90% pwm , I put an old 80mm fan 3cm below to blow at it for healthier idle temperature , DIY a stand using juice bottle scrap plastic


otherwise , a replacement fan cost about 5 to 10 dollars off taobao IIRC

 

[watzup_ken]

If it's under warranty, send it back for service. If its out of warranty, I suggest you do some DIY and have a 120mm fan blowing at the heatsink instead. I have 2 after market GPU coolers though none of them fit the existing GPU. The mounting holes don't fit. So buy at your own risk.

 

[serraphine]

Which model of msi gtx 970 are you using? ARMOR (black white) or Gaming (red black) series?

It's the MSI ARMOR model.

 

[serraphine]

Less than 3 years old why dont rma it? Still got chance to get the later gtx10 series equivalent card.

It's not entirely broken? Besides it only seems like it when I peek at it upside-down in opreation; I cannot confirm for sure because HWMonitor only registers a single fan speed - average value? It would be nice if it realises the GPU is cooled by two fans, and presented individual ratings.

 

[serraphine]

You can download afterburner to test if the fans are running properly too.

I do have Afterburner installed in the system. But it does not have granular monitor for individual fans? There's only a control setting for fan speed ramp-up against temperature.

For now I am setting a faster speed on lower temperature thresholds.

 

[watzup_ken]

It's not entirely broken? Besides it only seems like it when I peek at it upside-down in opreation; I cannot confirm for sure because HWMonitor only registers a single fan speed - average value? It would be nice if it realises the GPU is cooled by two fans, and presented individual ratings.

There is no harm in bringing it in to the distro to check. Anyway 82 degs for a GTX 970 is on the high side considering that the cooler used on the Armor series is considered to be quite beefy.

 

[serraphine]

There is no harm in bringing it in to the distro to check. Anyway 82 degs for a GTX 970 is on the high side considering that the cooler used on the Armor series is considered to be quite beefy.

Ok I'll bring it to Corbell for a checkup.

 

[watzup_ken]

Ok I'll bring it to Corbell for a checkup.

I think it is a good choice. Anyway, it may not be that convenient getting there, but sometimes it is better to let them test since we may not have the experience nor parts/tools to properly test it.

 

[serraphine]

I think it is a good choice. Anyway, it may not be that convenient getting there, but sometimes it is better to let them test since we may not have the experience nor parts/tools to properly test it.

Well I stay at Bishan and their facility's within scooter and even walking distance.

 

[singupingu]

Lucky still in warranty. If it is not, do hoot from taobao/aliexpress for a pair of 95mm gpu fans that's compatible with your gpu. Solved in about 20-30 bucks usually. Also take the chance to blow all the dust away and reapply thermal paste

Sent from OnePlus ONEPLUS A3003 using GAGT

 

[watzup_ken]

Well I stay at Bishan and their facility's within scooter and even walking distance.

Then this is good. Make use of this service since you paid for it.

 

[watzup_ken]

Lucky still in warranty. If it is not, do hoot from taobao/aliexpress for a pair of 95mm gpu fans that's compatible with your gpu. Solved in about 20-30 bucks usually. Also take the chance to blow all the dust away and reapply thermal paste

Sent from OnePlus ONEPLUS A3003 using GAGT

I think just strap a 120mm fan with cable ties and will keep the GPU cool enough. 30 bucks can get a good 120mm fan that can be reused. Haha.

 

[serraphine]

Just came back from Corbell service centre and re-fitted my video card back in. According to the tech support staff, the fans are operating just fine, although he couldn't specifically answer for sure what exactly that single fan rpm value represents (be it HWMonitor or MSI Afterburner) - is it average of both fans? or sum? Is that a guessed value by the software? Or a direct value output from the video card?

Anyway 82°C is still a tolerable operating temperature and he did not find any problems running through all the test suite programs.

That leaves a big question what is the origin of the crashes involving dxgkrnl.sys, dxgmms1.sys, nvlddmkm.sys, and even ntoskrnl.exe.

Since I update video drivers via Geforce Experience automated process, he suggested a full wipe of the drivers and a clean re-installation. I've never had to do these in all my years of computing/gaming (perhaps because in much earlier versions of Windows I'd occasionally re-format system drives and install Windows entirely afresh). I'm gonna have to read through the procedure somewhere.