Solved Check voltage for gpu

user1010 said:

Hi, I'm troubleshooting flicker/blinking black window and artifacts and was thinking the PSU maybe causing it because it started to happen after I changed the PSU. I got artefacts one time at logon screen and sometimes there is black window very fast while working in Windows. No problem when playing games and on high load for gpu. It's a nvidia RTX 4060 Ti.
Looking in Hwinfo -> Sensors. GPU Core voltage is 0.865. There is also GPU Rail Voltage and PCIe 12v input voltage. What should I look for and what is a normal voltage? It seems stable with that. Soon I'll get my old Corsair PSU back with a new one via RMA. I can change to that and test to see if it helps. It seems that it's not happen often. Bought a new 1000w earlier, which I'm running with now, because the 850w was making noise when running and maybe it was some more problem with that because it delayed web browsing process and other things, that I can see now. At first the performance got better with the new psu but there is some instability sometimes, had to reset the computer two times and very slow reponse at logon screen. Tried chaning power cable to graphic card and reconnected the power cables to motherboard. Ran the CleanupTool and reinstalled with the new driver + hotfix (572.24). No artefacts yet but I think I saw some black window blink, if I'm not see in sight.

Thankful for any tips of testing hardware in this case. Maybe it's a driver problem, I recently installed 572.16 (before psu upgrade). I ran some gpu benchmark and nothing unusual there but I think I saw some black fast blink window over the hole screen but I don't know if it should be like that.

Click to expand...

CountMike said:

Voltage varies with usage/load. Maximum voltage (non-OC) should be 1250mV @ 160W., recommended PSU is only 460W.

Click to expand...

user1010 said:

Tested Unigine Heaven Benchmark 4.0.

GPU Core Voltage goes from 0.870 to 1.075. The voltage seems stable, if it's that to look for.
It's also GPU PCIe +12V Input Voltage and GPU 16-pin HVPWR Voltage. Don't know if them are interesting too.

Click to expand...

CountMike said:

Output voltages from PSU have +/-5% deviation.

Click to expand...

user1010 said:

Seems normal then, at least for Core Voltage.
GPU PCIe +12V Input Voltage and GPU 16-pin HVPWR Voltage doesn't seems to affect on load like core voltage.

Ah, now I see maximum gpu power 164w.

Click to expand...

Helmut said:

You are just guessing as to the cause.
Your PSU is more than adequate, GPU core voltages vary, they are supposed to, so measuring that gets you nowhere.

So, based on symptoms:

Artifacts - usually the result of loss of Sync between game generated frames not syncing with Monitor frame rate.
These days that is complicated by the use of Monitors with a frequency higher than 60 Hz. So you would look for game settings around the subject of Syncing, change settings and see if there is effect for better or worse.

A black fast blink screen is a lot more difficult to suss out. It could be the syncing again but there are other possibilities.
Those people forcing a 24H2 version upgrade can get that symptom with a few games. Ignoring the MS blocks.
Look up what games are affected.

"MSI GeForce RTX 4060 Ti VENTUS 3X 16GB OC"
That can be a problem running higher than rated speeds, so try reducing the OC to run at stock speed, quiet mode or whatever they describe it. I'm sure MSI must supply a utility to do that. Also record GPU temperatures with an appropriate App when running the game where you get those black screen glitches.

"Maybe it's a driver problem"
Very unlikely, I have been using Nvidia Graphics for at least 25 years. I intentionally use drivers around 1 year old, only changing that on a yearly basis. There is no way I'm going to install ~12 new drivers a year.

Click to expand...

neves said:

Probably the PSU as you suspect (if that's when you started to see issues). You could always lower the voltage - using: MSI Afterburner - and see if that helps.

Here's a simple guide on how to do that:

Click to expand...

user1010 said:

Ok, the first profile that he descibes? What is he actually doing? Setting the voltage to a fixed value?

Click to expand...

neves said:

Undervolting (lowering the Voltage). Your GPU runs at a min and max voltage (depending on GPU usage - if low or high = decrease or increase). He's lowering the Max Voltage Value. Most GPUs have a higher voltage than required from factory - a value at which they run stable. But they can run even at lower voltage. Nothing bad can happen to the card - if undervolting (worst case scenario - you've lower the voltage to lowered and just have to increase it). Only Overclocking (increasing voltage) - can damage your GPU.

And BTW, as he pointed out - lock the profile (only when you have/found a stable value). This is just software Undevolt - which is applied when Windows and Afterburner Starts (hardware undevolt is done - within the GPU Bios - but that's more complicated and there's some risk to that - tho, it can't be done with this tool - you need a different tool to modify the BIOS values). If the values are to low - running a GPU stress tool/benchmark (like Heaven Benchmark - it's what he's using) - will give you a BSOD. Which is normal (as in, you know the cause - nothing wrong with your system - you lowered the volatge lower than the GPU needs to run stable). Yet, if the system runs perfectly fine at a given voltage (no BSOD and such) - that's the one you keep by locking the profile. Again, software undervolting can't do any damage. Quite the opposite - it's usually beneficial since lower voltage means - less heat (lower temperatures).

Click to expand...

user1010 said:

Could it be "coil whine"?

Click to expand...

neves said:

It's possible. Manny GPUs have Coil Whine - but that doesn't affect the GPU in any way, it's purely annoying - especially for those with OCD.

Click to expand...

Helmut said:

You are painting a completely different picture in post #9

I would not have bothered posting if that was in your original post.

Such a minor "problem" is of zero consequence.

You will not have the test equipment available for a PSU to check for subtle points like noise and ripple being within the specs.
Coil whine is not clicks it is more continuous, the result of poor quality transformer design i.e. wire coils being loose and vibrating when current changes etc.
An audible irritating noise.

"I have not overcloked or anything if it that you saying."

The name of the Graphics Card includes OC which is OverClocking as standard.

With your extra info. You have vastly more power than you need to just run Windows. The PSU has very little strain on it.
Much more when running a game.
You are saying it is stable, that does not indicate a PSU fault, quite the opposite.

Click to expand...

neves said:

You could always lower the voltage - using: MSI Afterburner - and see if that helps.

Click to expand...

user1010 said:

Why do you think that can help?

Click to expand...