[SOLVED] Troubleshooting GPU crashes and performance issues after upgrading to a 4060 Ti ?

[Calmcloud]

Hi,

It has been almost a decade since I last built a machine, and I’m apparently so outdated on that front that I feel completely hopeless while trying to troubleshoot this GPU issue.

I tried to upgrade my 1060 6GB to a 4060 Ti in hopes of playing some of the new games, but all I got from the upgrade was constant crashes and huge lag both during simple use and even worse when I tried to start a game, including simple games like WoW and LoL. When I switched back to the old 1060, everything worked fine and smooth, but I can’t run newer games or better graphics without lagging. I thought it might be some kind of “bottleneck” causing system imbalance, forcing my CPU to work overtime and causing shutdowns.

I have tried the following:

Reinstalled with DDU (didn't help)
Reinstalled/updated all drivers, BIOS, and chipsets (didn't help)
Rechecked all cables, connections, and thermal paste (didn't help)
Clean install of Windows and everything else (didn't help)
Adjusted all the graphic settings (reduced frequency of shutdowns/restarts)
Changed monitor from 1080p to 1440p (reduced frequency of shutdowns/restarts). CPU usage went down from 100%+ to around 20-30%, but GPU is still at 100%.
As I understand it, GPU usage at 100% isn’t bad, but it shouldn’t cause crashes/restarts after a fresh install, right?

Since the 4060 Ti is completely new, I haven’t ruled it out as faulty yet and tried to offset the CPU load to the GPU, but it’s still lagging/crashing in all attempts. It crashes less on 1440p than on 1080p.

Everything is freshly installed. I completely reinstalled Windows and all the drivers, but the problem persists. There is no visual damage to the GPU, and Device Manager doesn’t show any errors.

I have been considering an interim upgrade from the 3900X to the 5700X3D to solve the issues and use the AM4 setup for a few more years before changing the entire setup. Crashes/restarts stop simply by switching back to the 1060 on the same setup without any software changes.

Will upgrading the CPU solve the problem for now? Is this caused by a bottleneck leading to an overheating problem, or are the restarts caused by a lack of wattage on the PSU? I tried some wattage calculators, and it should be around 400-550W. Is there any other problem that is clear to you that I can’t see with this setup?

There are a lot of suggestions about a faulty GPU. If that’s the case, shouldn’t increasing the resolution from 1080p to 1440p cause more shutdowns/restarts? How do I troubleshoot this further?

Any advice is appreciated. Thanks.

Setup:
Ryzen 9 3900x
ROG STRIX B550-I-Gaming
Patriot Viper Steel DDR4 RAM 32GB (2X16GB) 3200MHz CL16
RTX 4060 ti
650W gold PSU.

Utilization of GPU/CPU simply just browsing on chrome.

 

[Eximo]

100% CPU usage in games? That seems highly unlikely. What do the under load temperatures and clock speeds look like?

What CPU cooler do you have on there?

Idle measurements aren't that useful. Fire up a benchmark or something and screenshot it then.

Which specific PSU? The 4060 Ti is 160W vs the 120W of the old card, but the transient spikes from 40 series cards are known to be much higher. An older PSU may be tripping over current protection when the card does a sudden increase in power usage.

Are you using a single daisy chained cable, or multiple cables from your PSU. You may be overloading a rail, again if the PSU is an older multi-rail design.

 

[Calmcloud]

100% CPU usage in games? That seems highly unlikely. What do the under load temperatures and clock speeds look like?

What CPU cooler do you have on there?

Idle measurements aren't that useful. Fire up a benchmark or something and screenshot it then.

Which specific PSU? The 4060 Ti is 160W vs the 120W of the old card, but the transient spikes from 40 series cards are known to be much higher. An older PSU may be tripping over current protection when the card does a sudden increase in power usage.

Are you using a single daisy chained cable, or multiple cables from your PSU. You may be overloading a rail, again if the PSU is an older multi-rail design.

I have not checked the PSU cables yet, that is a good point.

Looks something like this when I try to turn on WoW.

 

[Eximo]

Those max numbers are gibberish for the CPU and the GPU 3D measurement. May need to update your monitoring tool, or switch to a different one.

Pulled 101W from the CPU, but the GPU peaked at 58W, which is basically nothing. I

Launching a game is not exactly a benchmark. Something standardized so you can make a comparison. 3DMark has some free tests, available via Steam.

Only thing I can think of that might be of some use is to force PCIe 3.0 mode in the BIOS. 4.0 might be causing instability, particularly if you are using a riser cable.

Complete specifications would certainly help.

 

[Calmcloud]

Those max numbers are gibberish for the CPU and the GPU 3D measurement. May need to update your monitoring tool, or switch to a different one.

Pulled 101W from the CPU, but the GPU peaked at 58W, which is basically nothing. I

Launching a game is not exactly a benchmark. Something standardized so you can make a comparison. 3DMark has some free tests, available via Steam.

Only thing I can think of that might be of some use is to force PCIe 3.0 mode in the BIOS. 4.0 might be causing instability, particularly if you are using a riser cable.

Complete specifications would certainly help.

Thanks so much! I have been banging my head for 2 weeks finding a cause, it seems like its the riser cable. Got distracted by all the other suggestions and didnt even think of the riser cable. I forced it to 3.0 and no more lag, no more crash. You are a hero!