Gpu having power issues.

[Jfairbanks94]

Ok so im at a total loss here. I'm having some kind of intermittent issue with my gpu. 

System specs:

Asrock b460 pro4 motherboard

Intel i5-10400

Asus geforce gtx 1660 super oc 6gb pheonix fan edition

4 x 8 gb hyperx ddr4 ram

Thermaltake smart 600w psu

Samsung 870 Evo 1 TB SSD

Seagate barracuda 500 gb HDD

 

Things I have tried:

Separate psu.

System reset

Bios default setting reset.

Separate pcie slot

Not raging on my pc.

 

This issue started maybe about a month ago. Out of nowhere I started having issues maintaining a steady framerate. My card was not overheated and  I haven't modified it at all. Some days it handles itself as it should. Others it can barely get to 10 fps on ungine heaven basic settings.

When I load up gpu z it tells me the perfcap reason is power. When the card is having issues it states this constantly when functioning normally it reads idle when nothing runs. I submitted a repair request to asus in hopes it's the card. I'm just so confused what could be causing this? Attached is a photo of gpuz open when my pc was at full idle.

 

 

 

 

 

[Bombastinator]

I take it 143-191w is not normal idle draw?  Dudley draw on my 6800xt is 30w.  I did something or other that opened the lowest pic in another window and allowed me to read the type.  Which is where I got the 191 number.  Power use is shifting and it’s really high.  That could actually be full load for some cards. I’m wondering if maybe you’ve got malware mining crypto on your gpu.

Edited November 24, 2022 by Bombastinator

[Jfairbanks94]

I was thinking that too haven't found anything through any scans though. I just don't know how I'd find it

[Jfairbanks94]

6 hours ago, Bombastinator said:

I take it 143-191w is not normal idle draw?  Dudley draw on my 6800xt is 30w.  I did something or other that opened the lowest pic in another window and allowed me to read the type.  Which is where I got the 191 number.  Power use is shifting and it’s really high.  That could actually be full load for some cards. I’m wondering if maybe you’ve got malware mining crypto on your gpu.

I don't know if it helps but I did do a few things to try and check for this. First I ran a full scan through windows security. I then ran an offline scan. Nothing was found. I ran a deepscan through malwatebytes also to no avail. After that I did another full reset and formatted both drives. That seemed to help for about 30 minutes. It started bouncing between idle and pwr until it settled back into the same trend. Then I booted into safe mode and scoured through my files looking for anything obviously abnormal (Programs, Downloads, etc.). I deleted all system restore points. I double checked my bios version and updated it to the newest beta. Then did another full reset and wipe. Still nothing. I'm currently at 143.7 W of gpu chip power draw.

[PDifolco]

Maybe there's just some bad contact, reseat the card and replug the power connector

As it's a 1660 that may be 2 3 years old time for cleaning and repaste may be now as well

[LeapFrogMasterRace]

That sensor is just wrong I do not think you are drawing 200 watts that is just ridiculous and besides it says you TDP is actually 10%. 

 

You already tried a different psu so..

 

Update the GPU BIOS find them by digging around on the ASUS website

Uninstall the Nvidia drivers using DDU and reboot then reinstall them

Uninstall all GPU monitoring software except for one program (I like MSI Afterburner personally) multiple programs can eff up the sensors readings sometimes. 

 

If all those fail a power chip on the gpu board itself has probably just failed due to age. 

[Bombastinator]

9 hours ago, Jfairbanks94 said:

I was thinking that too haven't found anything through any scans though. I just don't know how I'd find it

Might not be there.  Do you have an iGP?

[WoodenMarker]

Have you tried a clean GPU driver install?

Are you using any GPU software like Msi Afterburner?

[Jfairbanks94]

21 hours ago, PDifolco said:

Maybe there's just some bad contact, reseat the card and replug the power connector

As it's a 1660 that may be 2 3 years old time for cleaning and repaste may be now as well

I've done a full pull apart and clean on my step twice I've also tried various power cables and power supplies. I'm sending it to asus for a repaste.

[Jfairbanks94]

15 hours ago, WoodenMarker said:

Have you tried a clean GPU driver install?

Are you using any GPU software like Msi Afterburner?

Yes I have reset my gpu driver. Nope no software though I did try raising the voltage with afterburner to no avail.

[Jfairbanks94]

20 hours ago, LeapFrogMasterRace said:

That sensor is just wrong I do not think you are drawing 200 watts that is just ridiculous and besides it says you TDP is actually 10%. 

 

You already tried a different psu so..

 

Update the GPU BIOS find them by digging around on the ASUS website

Uninstall the Nvidia drivers using DDU and reboot then reinstall them

Uninstall all GPU monitoring software except for one program (I like MSI Afterburner personally) multiple programs can eff up the sensors readings sometimes. 

 

If all those fail a power chip on the gpu board itself has probably just failed due to age. 

It sounds ridiculous but I do believe it to be accurate as performance goes down as well. Using basic ungine setti gs I'm maxing o I t my gpu load getting only 10 fps. I will definitely try the gpu bios. Temporarily I got another card from a buddy of mine. I plan on using that to see if the odd behavior continues. Possibly pointing at board issues.

[Jfairbanks94]

17 hours ago, Bombastinator said:

Might not be there.  Do you have an iGP?

I do. When I pull the card and use it alone the issue is no longer ther. Part of why I believe it has to be the gpu

 

[Jfairbanks94]

Update: I believe it is the card for sure. Possibly Thermal paste maybe a bad power chip. I cam to this final conclusion after testing a secondary gpu in the pc. I swapped it with a gtx 1080 xtreme gaming. Currently with this card these are my outputs. I plan on sending the gpu off to repair as asus claims thermal repasting voids the warranty. I will update if the issue is solved once the card returns and I am able to test it. Thanks for the help guys!

[LeapFrogMasterRace]

50 minutes ago, Jfairbanks94 said:

It sounds ridiculous but I do believe it to be accurate as performance goes down as well. Using basic ungine setti gs I'm maxing o I t my gpu load getting only 10 fps.

Your performance is going down because the GPU THINKS it is drawing max power so it goes into a sort of limp mode (this is why you are getting low performance)but it is actually drawing probably 30 watts. Imagine if your GPU had a temp sensor that read 90c your GPU would downclock like crazy but in reality it might only be 30C, the GPU does not know though as that sensor is the only thing it can go off of. That is what is happening now but with power instead. It is saying uh oh if I draw any more power I will blow up so I will limit myself to 300Mhz, my lowest rated clock speed). If you were at that many watts your temps would be much higher. 

 

Just seen your new comment.

 

3 minutes ago, Jfairbanks94 said:

Update: I believe it is the card for sure. Possibly Thermal paste maybe a bad power chip. I cam to this final conclusion after testing a secondary gpu in the pc. I swapped it with a gtx 1080 xtreme gaming. Currently with this card these are my outputs. I plan on sending the gpu off to repair as asus claims thermal repasting voids the warranty. I will update if the issue is solved once the card returns and I am able to test it. Thanks for the help guys!

 

 

 

Sounds good.. don't forget about removing any other monitoring programs. One in particular I had issues with was Corsair stuff it makes the sensors go haywire. Also when uninstalling the old drivers did you use DDU? 

 

After all these test fail I still would start thinking your power chip has gone bad and you will either need to send it off to ASUS if you still have warranty anyway or find a good shop that specializes in PC board repairs. 

 

[Bombastinator]

3 hours ago, Jfairbanks94 said:

I do. When I pull the card and use it alone the issue is no longer ther. Part of why I believe it has to be the gpu

 

Or the software for that gpu, or an interaction with that motherboard bios and that gpu.

[Jfairbanks94]

4 hours ago, LeapFrogMasterRace said:

Your performance is going down because the GPU THINKS it is drawing max power so it goes into a sort of limp mode (this is why you are getting low performance)but it is actually drawing probably 30 watts. Imagine if your GPU had a temp sensor that read 90c your GPU would downclock like crazy but in reality it might only be 30C, the GPU does not know though as that sensor is the only thing it can go off of. That is what is happening now but with power instead. It is saying uh oh if I draw any more power I will blow up so I will limit myself to 300Mhz, my lowest rated clock speed). If you were at that many watts your temps would be much higher. 

 

Just seen your new comment.

 

 

 

Sounds good.. don't forget about removing any other monitoring programs. One in particular I had issues with was Corsair stuff it makes the sensors go haywire. Also when uninstalling the old drivers did you use DDU? 

 

After all these test fail I still would start thinking your power chip has gone bad and you will either need to send it off to ASUS if you still have warranty anyway or find a good shop that specializes in PC board repairs. 

 

Ok that makes sense I thought you meant faulghty software reading my bad. Luckily I am still in warranty so I'm gonna go ahead and send it off.

[WoodenMarker]

On 11/25/2022 at 11:04 AM, Jfairbanks94 said:

Yes I have reset my gpu driver. Nope no software though I did try raising the voltage with afterburner to no avail.

Did you have Msi Afterburner installed before? 

Check and make sure the power limit is set to 100% in the software. 

[Jfairbanks94]

On 11/26/2022 at 4:57 PM, WoodenMarker said:

Did you have Msi Afterburner installed before? 

Check and make sure the power limit is set to 100% in the software. 

No the card is pre overclocked so I genuinely didn't see a need to. I only downloaded the software when I started having issues to help with possible troubleshooting. It was all set default and I reset it upon uninstall. I honestly thought it was board related and I was just missing something obvious. I've had a second card in for a few days with zero issues. I've just never had a card act that way. Everytime I've had issues it's pretty well just died on the spot or I had choppy video. I'm hoping when it gets repaired or if it needs replaced they will tell me what was wrong. I think this thread narrowed it down pretty well. Either bad sensor or power controller.