[SOLVED] Problems with new 3080 and PSU

[platonicpotato]

Hi, I’ve recently posted on linustechtips forums regarding an issue with my pc after upgrading my GPU from a 1080 ti to an msi rtx 3080 gaming z and my psu from an old 850w Corsair psu to hx1200 Corsair psu. Unfortunately, the folks on the Linus forums were unable to help me so I’m trying my luck here (will post it the forum link below for more context). In summery, I’ve been getting performance problems along with occasional stuttering and crashing in games. Someone on my forum post suggested for me to use GPUz and post a screenshot of the sensor readings while playing a demanding game. They found that the card was running a clock speed higher than the manufacturer boost clocks while using a lower than expected voltage for that clock speed (I’ll post a screenshot of the sensor readings below). I’ve tried using 3 separate 8 pin PCIe cables for the card and it made no difference. I’m assuming that the reason why the GPU is running higher boost speed is because of GPU boost 3.0 and maybe the voltage is off because it might have something. to do with msi afterburner settings or Evga precision x app I uninstalled after installing the GPU (pic of the afterburner voltage curve below). Please let me know what you think I should do. Someone suggested that I should flash the bios but I’m really not a tech savvy guy and I’m afraid of creating a new and even bigger problem than the one I currently have by flashing the bios ( something I’ve done only once many years ago on a cheap laptop). I need to exhaust every other option first and leave a bios flash as a dead last resort. Please let me know if you have any idea how I can solve this issue. Thank you.



ADDITIONAL INFO:



Temps are all fine (GPU and CPU typically stay at around 60-65c at load). In order to try to get this stuttering to go away, I've tried to use that OC button on my motherboard because I know nothing about overclocking and afraid to ruin something if I do it manually myself (I'm really good at breaking stuff unintentionally) but to my sadness that OC button doesn't light up when i press it while the pc is turned off and it doesn't work. Am i suppose to press it while the pc is turned on? can i enable it from the bios? please let me know if you can figure out the reason why it isn't working.

Anyway, I've also tried:
1- enabling/disabling hardware accelerated gpu scheduling on windows
2- turning game mode in windows on and off
3- set the power management mode in Nvidia control panel to set maximum performance and normal along with windows power management setting to “max performance”
5- Turning ray tracing/ DLSS settings on and off
And yes, before upgrading my GPU i went into safe mode and used DDU to remove my old driver then I installed the GPU and installed the new driver.

 

[platonicpotato]

Here is an image of the Afterburner voltage curve

 

[Phaaze88]

Where's that forum link?

There's going to be at least one Perfcap Reason most of the time. With Gpu Boost 3.0, cards boost/overclock themselves, but not to infinity. And yes, the manufacturer advertised boost can pretty much be ignored - it's more of a guaranteed clock in a worst case scenario.
In the case of your gpu, there's mostly the blue Vrel(it does not boost any further because it will not be stable at the current voltage), with a very small blip of orange Vop(the voltage cap defined by the voltage controller on the card was reached).


Since vbios updating is off the table for now, it's time to look at something else. It could well be that the new gpu exposed a problem that was already there.
But yeah, where's that link, so others don't ask/suggest what was already brought up over there.

 

[platonicpotato]

Where's that forum link?

There's going to be at least one Perfcap Reason most of the time. With Gpu Boost 3.0, cards boost/overclock themselves, but not to infinity. And yes, the manufacturer advertised boost can pretty much be ignored - it's more of a guaranteed clock in a worst case scenario.
In the case of your gpu, there's mostly the blue Vrel(it does not boost any further because it will not be stable at the current voltage), with a very small blip of orange Vop(the voltage cap defined by the voltage controller on the card was reached).


Since vbios updating is off the table for now, it's time to look at something else. It could well be that the new gpu exposed a problem that was already there.
But yeah, where's that link, so others don't ask/suggest what was already brought up over there.

Yeah sorry i forgot to post the link, here it is.
https://linustechtips.com/topic/139...most-all-games-other-issues/#comment-15147932

so what do you think of the voltage that the card is getting? is it adequate? also in other games i get a whole mix of orange and blue so its not always blue like that

 

[Phaaze88]

Voltage is fine. There's no more room to push it, due to Vop. That's a hardware limit from the onboard voltage regulator.

 

[platonicpotato]

Voltage is fine. There's no more room to push it, due to Vop. That's a hardware limit from the onboard voltage regulator.

Well how about the 8-pin #1 drawing more power than 2 and 3? does that matter at all?

 

[Phaaze88]

I see in the other forum that a Windows reinstall is also on the backburner...

OK, how about storage drives: What are they, and have you checked their condition lately?

Monitors: Single, more than one? Native refresh?

Did you install anything else - hardware/software - around the same time as the gpu?

Well how about the 8-pin #1 drawing more power than 2 and 3? does that matter at all?

No, I don't think so, except if power draw is beyond spec for the cable, and I haven't seen anything suggesting that's happening.

 

[platonicpotato]

I see in the other forum that a Windows reinstall is also on the backburner...

OK, how about storage drives: What are they, and have you checked their condition lately?

Monitors: Single, more than one? Native refresh?

Did you install anything else - hardware/software - around the same time as the gpu?


No, I don't think so, except if power draw is beyond spec for the cable, and I haven't seen anything suggesting that's happening.

Yes I uninstalled EVGA percision x (what i used for my old GPU) and installed Afterburner.
Single monitor @1080p 144hz
Storage 1x 250gb Samsung 840 SSD and 1x 1TB Samsung EVO 860 SSD (I get a bunch of Disk# was suprise removed warnings in my event viewer a lot but other than that I don't know if any issue with the SSDs. I also don't think they're the problem because i was using the same SSDs with my old card and didn't have crashing like i do now.

with my 3080, i installed a brand new PSU HX1200w corsair and im currently using multi rail. You think switching to single rail will do anything?

 

[Phaaze88]

(I get a bunch of Disk# was suprise removed warnings in my event viewer a lot...

What? They got flagged for high number of bad sectors? That could cause stuttering.

with my 3080, i installed a brand new PSU HX1200w corsair and im currently using multi rail. You think switching to single rail will do anything?

I am not familiar with that, so I can't give any advice on it.

Do you still have the old psu? What is it?
I'm thinking maybe trying it out and see if it still crashes.

 

[platonicpotato]

What? They got flagged for high number of bad sectors? That could cause stuttering.


I am not familiar with that, so I can't give any advice on it.

Do you still have the old psu? What is it?
I'm thinking maybe trying it out and see if it still crashes.

Putting in the old psu back just to test it requires an intense amount of work and after the hell I went through installing the new psu I don’t want to do that. I’ve tried to test my 1tb ssd with Samsung magician and everything turned out fine, it won’t let me test the 250gb ssd that windows is installed on tho. The old psu is a 5 year old hx850 Corsair bronze

 

[Karadjgne]

No difference between single and multi rail at those wattages. Multi rail just means there's multiple circuits supplying a single output (in most cases) vs a single circuit. So a 60A output multi rail might be 2x 30A circuits or 3x 20A circuits etc which doesn't require the much larger capacity caps of a single rail.

But it all goes to a single buss link which is tied to all the various outputs.

It's only in the oddball super cheapo psus that the rails are seperate feeds directly to certain outputs, done to cut component costs and the additional work to add the buss.

The advantage of multi rails is stability with a massive spike draw, since no one rail takes the full brunt of the hit, it's split by the buss into all the rails. But that draw can exceed the rating of any one rail. The advantage of single rail is over-provisioning, a 60A claimed output psu might have a 90A capable rail, so no matter what the hit, it's covered temporarily.

Doubt the psu has anything whatsoever to do with any of the issues.

They found that the card was running a clock speed higher than the manufacturer boost clocks while using a lower than expected voltage for that clock speed

So? Welcome to Gpu Boost 3. That's a normal occurrence. That's not a psu problem as the card will only pull what it requires, has zero to do with the psu not supplying enough.