Weird GPU Problems

[ButterLettuth]

Hi all!

 

First time poster here, usually just a lurker but I have run in to a problem and exhausted my own ideas, so I come to you for help!  

 

For the past few months my GTX 960 has been giving me problems booting, bsods (occasionally), and incredibly low benchmark performance (1800 ish on performancemark).  The problem is all of these issues seem to be intermittent.  For instance two weeks ago i did a performancemark test and scored over 6000, today i did another one and it was 1668.  I have done everything imaginable to try and figure out why this is happening, but i have no idea.  I have tried the following:

 

- tested the card at a local shop (booted fine, wasn't benchmarked)

- removed card and re-inserted

- flashed CMOS and updated BIOS

- different PCiE slots (and pcie 8 pin cables)

- updating drivers

- updating windows

- overclocking

- underclocking

- furmark, catzilla, ksi kombuster, performancemark

 

I'm not sure what else i can do, and I'm not quite ready to blow money on new parts yet as my partner and i just bought a house so money is kind of tight.

 

System Specs:

Asus Z97-A LGA 1150

Intel Core i5 4690K

MSI GTX 960 4gb 100 ME

8 GB DDR3 Corsair Ram

120gb ssd

2tb hdd

**forgot to include this originally, but the PSU is a Corsair TX750W, i tested all 4 pcie 12v cables and had the same issues with each.**

 

As a reference, i was playing Monster Hunter World yesterday averaging 50-55 fps, i go to try it today and im averaging 14-17fps.

 

Any tips would be super valuable so thank you in advance.  Please let me know if you have any questions.

 

Thanks!

 

[Turtle Rig]

19 minutes ago, ButterLettuth said:

Hi all!

 

First time poster here, usually just a lurker but I have run in to a problem and exhausted my own ideas, so I come to you for help!  

 

For the past few months my GTX 960 has been giving me problems booting, bsods (occasionally), and incredibly low benchmark performance (1800 ish on performancemark).  The problem is all of these issues seem to be intermittent.  For instance two weeks ago i did a performancemark test and scored over 6000, today i did another one and it was 1668.  I have done everything imaginable to try and figure out why this is happening, but i have no idea.  I have tried the following:

 

- tested the card at a local shop (booted fine, wasn't benchmarked)

- removed card and re-inserted

- flashed CMOS and updated BIOS

- different PCiE slots (and pcie 8 pin cables)

- updating drivers

- updating windows

- overclocking

- underclocking

- furmark, catzilla, ksi kombuster, performancemark

 

I'm not sure what else i can do, and I'm not quite ready to blow money on new parts yet as my partner and i just bought a house so money is kind of tight.

 

System Specs:

Asus Z97-A LGA 1150

Intel Core i5 4690K

MSI GTX 960 4gb 100 ME

8 GB DDR3 Corsair Ram

120gb ssd

2tb hdd

 

As a reference, i was playing Monster Hunter World yesterday averaging 50-55 fps, i go to try it today and im averaging 14-17fps.

 

Any tips would be super valuable so thank you in advance.  Please let me know if you have any questions.

 

Thanks!

 

You seem to have a nice system there.  After running all those tests and still have issue, then it could just be a faulty motherboard that needs RMA and what not.  Are you still under warranty ?

[ThisIsCheez]

Download GPU-Z and see if your core/memory clocks are showing up correctly.

[WebPlayz]

50 minutes ago, ButterLettuth said:

Hi all!

 

First time poster here, usually just a lurker but I have run in to a problem and exhausted my own ideas, so I come to you for help!  

 

For the past few months my GTX 960 has been giving me problems booting, bsods (occasionally), and incredibly low benchmark performance (1800 ish on performancemark).  The problem is all of these issues seem to be intermittent.  For instance two weeks ago i did a performancemark test and scored over 6000, today i did another one and it was 1668.  I have done everything imaginable to try and figure out why this is happening, but i have no idea.  I have tried the following:

 

- tested the card at a local shop (booted fine, wasn't benchmarked)

- removed card and re-inserted

- flashed CMOS and updated BIOS

- different PCiE slots (and pcie 8 pin cables)

- updating drivers

- updating windows

- overclocking

- underclocking

- furmark, catzilla, ksi kombuster, performancemark

 

I'm not sure what else i can do, and I'm not quite ready to blow money on new parts yet as my partner and i just bought a house so money is kind of tight.

 

System Specs:

Asus Z97-A LGA 1150

Intel Core i5 4690K

MSI GTX 960 4gb 100 ME

8 GB DDR3 Corsair Ram

120gb ssd

2tb hdd

 

As a reference, i was playing Monster Hunter World yesterday averaging 50-55 fps, i go to try it today and im averaging 14-17fps.

 

Any tips would be super valuable so thank you in advance.  Please let me know if you have any questions.

 

Thanks!

 

Hi!

 

I used to have a GTX 960 and had a similar problem. The way I corrected it was installing DDU ( https://www.guru3d.com/files-details/display-driver-uninstaller-download.html ) 

 

Then booting into Safe mode and uninstalling the whole entire driver. Then Reboot the computer and Windows will load its Basic VGA Driver.

 

Proceed to install the latest Nvidia Driver ( https://www.geforce.com/drivers/results/142736 ) Sometimes the beta versions can help 

 

Then installing MSI Afterburner ( https://www.msi.com/page/afterburner ) To check the tempatures and see if the card is thermal throttling 

 

Let me know if that helps

[ButterLettuth]

On 2/15/2019 at 8:57 PM, ThisIsCheez said:

Download GPU-Z and see if your core/memory clocks are showing up correctly.

I did this now and it looks like they're off... by a fair bit.  GPU-Z says my core clock should be 1253Mhz, but it's benchmarking at 1380.  The memory clock is accurate (1753Mhz vs 1752.8 actual).

 

Would benchmarking at a higher clock rate give me frame rate issues?

 

On 2/15/2019 at 8:50 PM, Turtle Rig said:

You seem to have a nice system there.  After running all those tests and still have issue, then it could just be a faulty motherboard that needs RMA and what not.  Are you still under warranty ?

Thank you, it's a few years old but it's traditionally done me very well   Unfortunately I don't think the mobo is still under warranty, but it may very well need replacement.  My issue is that my ram and intel chip aren't that new so there aren't many new boards for me to move to, yknow?  I'm concerned if i upgrade the board i would have to upgrade other things as well

 

On 2/15/2019 at 9:23 PM, Killstreak said:

-snip-

Hello!

 

Thanks for your advice.  I was able to do all of this but unfortunately no luck.  One of the weirdest things that i neglected to mention above was that even when the gpu is being benched and under 90-95% load consistently, i rarely see temps over 50c.  I've heard operating temps of 70 are normal, so 50 seems really low.

 

Thanks again

[Turtle Rig]

57 minutes ago, ButterLettuth said:

Thank you, it's a few years old but it's traditionally done me very well   Unfortunately I don't think the mobo is still under warranty, but it may very well need replacement.  My issue is that my ram and intel chip aren't that new so there aren't many new boards for me to move to, yknow?  I'm concerned if i upgrade the board i would have to upgrade other things as well

You can keep the RAM you have.  But you don't neccesarily need a new cpu.  Just buy the same motherboard used or new if you can hopefully find it easy.

[ButterLettuth]

26 minutes ago, Turtle Rig said:

You can keep the RAM you have.  But you don't neccesarily need a new cpu.  Just buy the same motherboard used or new if you can hopefully find it easy.

so you think it's most likely a motherboard issue?

[Tristerin]

@ButterLettuth this sounds like a hardware issue.  However the only way to eliminate software issues you should do a fresh install of operating system.  If the issue persists it is then hardware.  

 

I wouldn't purchase anything until I did that.  Problem persists then we need to run that problem to ground with monitoring software for voltages, etc and replicate to isolate.

 

For me, I like HWinfo (sometimes to much information to users) and HWmonitor (cleaner look, clear cut layout).  You would then need to start benchmarking while monitoring and see what happens when you get those dips in performance, and where from.  At least that would be my strategy.

[ButterLettuth]

2 hours ago, Tristerin said:

@ButterLettuth this sounds like a hardware issue.  However the only way to eliminate software issues you should do a fresh install of operating system.  If the issue persists it is then hardware.  

 

I wouldn't purchase anything until I did that.  Problem persists then we need to run that problem to ground with monitoring software for voltages, etc and replicate to isolate.

Hello!  I actually did a re-install earlier, and the problem persisted.  I downloaded HWInfo like you said, and ran the Unigine Heaven bench.  Again GPU power remained under 50%, and this time i could see GPU power averaging about 44w, maxing out at 65w.  Im pretty sure the 960 has a suggested power demand of 120w, and in real testing i saw it reported to draw closer to 180.  Does that confirm a hardware issue?  Wouldn't that throttle clock or memory speed?

 

Thanks for your help

[Tristerin]

8 minutes ago, ButterLettuth said:

Hello!  I actually did a re-install earlier, and the problem persisted.  I downloaded HWInfo like you said, and ran the Unigine Heaven bench.  Again GPU power remained under 50%, and this time i could see GPU power averaging about 44w, maxing out at 65w.  Im pretty sure the 960 has a suggested power demand of 120w, and in real testing i saw it reported to draw closer to 180.  Does that confirm a hardware issue?  Wouldn't that throttle clock or memory speed?

 

Thanks for your help

Its getting power from two sources, the PCIe x16 slot and the PSU cable.  Since you've tested the other cables/ports and nothing is shutting down I don't think its the PSU atm.  Have you tried the other PCIe x8/x16 (if you have them) slots on the motherboard with the GPU? 

[ButterLettuth]

44 minutes ago, Tristerin said:

Its getting power from two sources, the PCIe x16 slot and the PSU cable.  Since you've tested the other cables/ports and nothing is shutting down I don't think its the PSU atm.  Have you tried the other PCIe x8/x16 (if you have them) slots on the motherboard with the GPU? 

I have two pcie 3.0 x16 slots, so i tried plugging in to the other one and i think it's not functioning correctly.  I had to update bios to force it to run at Gen 3, but it still shows up in HWinfo as: PCIe v3.0 x16 (8.0 GT/s) @ x8 (2.5 GT/s).

 

Additionally it now only shows GPU temp in sensor status, and shows link speed at 2.5 GT/s.

 

Any idea how i can make sure it's using Gen 3 besides setting it up that way in bios or is this a hw issue?

 

Edit: I went back in to the bios and confirmed when the card is in slot 2 (labeled pciex16_2), it's locked at x8.  I am now assuming this is a mobo issue.

[Tristerin]

10 minutes ago, ButterLettuth said:

I have two pcie 3.0 x16 slots, so i tried plugging in to the other one and i think it's not functioning correctly.  I had to update bios to force it to run at Gen 3, but it still shows up in HWinfo as: PCIe v3.0 x16 (8.0 GT/s) @ x8 (2.5 GT/s).

 

Additionally it now only shows GPU temp in sensor status, and shows link speed at 2.5 GT/s.

 

Any idea how i can make sure it's using Gen 3 besides setting it up that way in bios or is this a hw issue?

The pic attached is from the motherboard manual - I think its saying that when you use the second PCIe slot its autos into x8/x8?  

 

When you bench whats the power doing now?  Your card wont be bottlenecked by x8 so really nothing to be worried about.  Need to see if the PCIe slot is giving 75w of juice or not (since the board is rated for xfire/sli) - if the card still under performs Id lean towards saying something is wrong with the card not the mobo or PSU (since cables/PSU ports and perhaps both slots are still showing the GPU failure)

[ButterLettuth]

15 minutes ago, Tristerin said:

The pic attached is from the motherboard manual - I think its saying that when you use the second PCIe slot its autos into x8/x8?  

 

When you bench whats the power doing now?  Your card wont be bottlenecked by x8 so really nothing to be worried about.  Need to see if the PCIe slot is giving 75w of juice or not (since the board is rated for xfire/sli) - if the card still under performs Id lean towards saying something is wrong with the card not the mobo or PSU (since cables/PSU ports and perhaps both slots are still showing the GPU failure)

So i actually was not able to run any benchmark software while the card was in slot 2, and i am not sure if i know how to force it to work.  It was also not showing any sensors in hwinfo when connected through slot 2.

 

That being said, i ran another benchmark in unigine heaven with the card inserted in to slot 1.  This time the card drew an average of about 105w (way more than the 60 from an hour ago) and benchmarked at 1129 compared to 325 an hour ago.  I honestly don't know what changed cause i didnt change any settings.  This is what usually happens and it's why im trying not to replace things.

 

 - did another benchmark (catzilla 720p) and got 14.8k, compared to 3.6k earlier today.  Also gpu power went up to 130w, compared to a previous max of 60.

[Tristerin]

12 hours ago, ButterLettuth said:

So i actually was not able to run any benchmark software while the card was in slot 2, and i am not sure if i know how to force it to work.  It was also not showing any sensors in hwinfo when connected through slot 2.

 

That being said, i ran another benchmark in unigine heaven with the card inserted in to slot 1.  This time the card drew an average of about 105w (way more than the 60 from an hour ago) and benchmarked at 1129 compared to 325 an hour ago.  I honestly don't know what changed cause i didnt change any settings.  This is what usually happens and it's why im trying not to replace things.

 

 - did another benchmark (catzilla 720p) and got 14.8k, compared to 3.6k earlier today.  Also gpu power went up to 130w, compared to a previous max of 60.

That's odd you cant run the card in slot 2 it should be plug and play.  So you are getting a display from Slot 2, but software wont run? 

 

Slot 1 + PSU cable now delivering power - is there a potential that your GPU is somehow partially unseating itself or you are not seating it properly?  I only ask because Ive had my GPU unslot after moving case (very slightly) and it booted up and everything until it went to pull full power loading a game it just shut down.  So not indicative of whats happening with you but just asking to clear that up.

 

Have you run 3dmark firestrike to see how your GPU is running compared to other systems like yours? 

 

The only options I can think of is there is a power delivery or power receipt issue going on here and Im thinking its likely power receipt on the GPU side.  The only way to isolate further in my opinion is you need a secondary GPU to try next.

[ButterLettuth]

3 hours ago, Tristerin said:

That's odd you cant run the card in slot 2 it should be plug and play.  So you are getting a display from Slot 2, but software wont run? 

 

Slot 1 + PSU cable now delivering power - is there a potential that your GPU is somehow partially unseating itself or you are not seating it properly?  I only ask because Ive had my GPU unslot after moving case (very slightly) and it booted up and everything until it went to pull full power loading a game it just shut down.  So not indicative of whats happening with you but just asking to clear that up.

 

Have you run 3dmark firestrike to see how your GPU is running compared to other systems like yours? 

 

The only options I can think of is there is a power delivery or power receipt issue going on here and Im thinking its likely power receipt on the GPU side.  The only way to isolate further in my opinion is you need a secondary GPU to try next.

Its very odd! I was very confused.  I was able to start windows but couldn't run any benchmarks.  I was able to open browsers and apps but nothing intended to stress the system.  I was concerned about whether I had seated it properly or not as well so I was extremely careful when I put it back in slot 1, and it worked fine once placed, though this morning it's back to being trash.

 

I haven't run 3Dmark but I did run performancetest (gpubenchmark) and it tested above average (6100 ish compared to 5800 average), but when its having problems it will test closer to 1600.  I will run firestrike tonight to see.

 

I wish I had the money for a new gpu, so I guess the next step would be to drop it off to a shop eh?

[Tristerin]

4 minutes ago, ButterLettuth said:

Its very odd! I was very confused.  I was able to start windows but couldn't run any benchmarks.  I was able to open browsers and apps but nothing intended to stress the system.  I was concerned about whether I had seated it properly or not as well so I was extremely careful when I put it back in slot 1, and it worked fine once placed, though this morning it's back to being trash.

 

I haven't run 3Dmark but I did run performancetest (gpubenchmark) and it tested above average (6100 ish compared to 5800 average), but when its having problems it will test closer to 1600.  I will run firestrike tonight to see.

 

I wish I had the money for a new gpu, so I guess the next step would be to drop it off to a shop eh?

If you can google how to use a volt meter to check the PCIe slots and the PSU cables to ensure they are delivering correctly, they will eliminate the PSU and mobo from the equation and isolate to the GPU (would be what I Would do next) I have a voltmeter but haven't had to do this myself so the next step would be googling how