GPU at factory settings causing screen blackouts during game.

[watzup_ken]

The fact that the issue occurs less frequently when you undervolt it makes it seem like a power issue. Try with a few different driver versions and see if it helps. Otherwise it’s best you test it with another PSU to try and identify the issue. I suggest you test it fast since your Amazon MSI card has a limited time return window, just in case it really is a faulty card.

 

[Yongkit]

I DDU-ed my AMD drivers and shutdown before putting in the new card. Then I straightaway download the latest nvidia driver and installed that. To be safe, I DDU-ed the nvidia driver again last night and reinstalled.

Any chance for studio driver a try?

 

[Navillera]

I used 650W super flower Leadex III gold argb for 2 different models of rtx 3080. No crashes. The card will draw until 340W when monitoring with gpu Z.

Posted from PCWX using iPhone13,2

 

[Reavl]

try 3 cable into your gpu rather than 1 daisy chain + one extra cable?

 

[BlurKillerEX]

try 3 cable into your gpu rather than 1 daisy chain + one extra cable?

This model only require 2x 8pin

 

[Crooni]

alternatively run the 3700x at stock.

overclocked 3700x will spike at 150w. plus x570 and other peripherals, you'll need to allocate 180-200w. means you only got ~400w for gpu. most probably gpu spikes are above 100% liao

 

[Psycovirus]

Do note that although 3080 is rated to be 320 Watts average, there are very high Peak power draws for milliseconds sometimes.

An example test from Igor's Lab shows a 3080 FE can pull 489 Watt max peak for a split-second. If the Power Supply is sensitive, it might trip the Over current protection.

So a 650W power supply might not be enough and might cause the disruptive behaviours you're seeing.

Edit: Same idea as poster above. haha.

 

[noinimod]

Don't rule out driver issues here.

I've been having a lot of random black screens and crashes on the newer nvidia drivers. Also using RTX 3 series.

Went back to 457.51 and all seems good (for now). Fingers crossed.

If you're using afterburner that could cause system instability also. It did for me.

You should run two separate PCIE cables from your PSU if there is provision. Don't daisy chain it

 

[BlurKillerEX]

Do note that although 3080 is rated to be 320 Watts average, there are very high Peak power draws for milliseconds sometimes.

An example test from Igor's Lab shows a 3080 FE can pull 489 Watt max peak for a split-second. If the Power Supply is sensitive, it might trip the Over current protection.

So a 650W power supply might not be enough and might cause the disruptive behaviours you're seeing.

Edit: Same idea as poster above. haha.

When I was testing just before going to bed last night, I was monitoring the power via GPU-Z. Set the sensor to show the highest power draw and it peaks out at about 82% at .818v and 65% at .781v

82% is around 263w, nowhere even near 320 but there was still the blackout issue.

 

[Reavl]

try another dp port? or another cable? did you rule out those?

 

[Psycovirus]

When I was testing just before going to bed last night, I was monitoring the power via GPU-Z. Set the sensor to show the highest power draw and it peaks out at about 82% at .818v and 65% at .781v

82% is around 263w, nowhere even near 320 but there was still the blackout issue.

You have to understand that using GPU-Z sensor polling rate is default to 1 second, 1000 millisecond. Power spike happens for less than a millisecond. You have 1/1000 chance of actually detecting a power spike in the sensor. You only see that it is peaked at 263 Watts but the spikes are still there.

igor's Lab use special lab equipment to test the cards. Polling constantly to identify the spikes.

If PSU is sensitive, it will trip the PSU's overcurrent protection. However, it seems your PSU protection isn't tripped but it is highly the culprit causing the issues.

All I could suggest is to try a bigger power supply.

On a side note, there are reports of 6900 XT Ref causing Silverstone SFX 800W Platinum Power supply to trip due to current protection while a Corsair 750W SFX power supply runs fine.

 

[BlurKillerEX]

try another dp port? or another cable? did you rule out those?

Tried second dp port, same issue. Will go back and try using hdmi

 

[tsteo77]

You need to upgrade psu to at least 750w and above.

 

[BlurKillerEX]

So... Funny thing, I did a DDU again, installed one more fan in my PC, fiddled around with the cables a bit, uninstalled MSI afterburner.

I've been playing cyberpunk 2077 on RTX psycho settings for the past 2 hours without any blackouts. Didn't even undervolt anything this time.

 

[BlurKillerEX]

When I first installed the card, I did some superposition test runs.
Back then I was consistently getting around 78 avg fps, 98 max fps at 1080p extreme,

Now that the problem seems to be resolved, I've ran superposition a few more times, same settings, I'm getting a +4 avg fps and +10 max fps. Something was definitely wrong before and it wasn't the psu or card.

 

[Dark_Skye]

Just set power limit to 75% in afterburner. Works for me. I'm using sf750 BTW and before I did this, I was crashing like no tomorrow on all games. Now I can play for 6 hours straight without a single crash. Using latest drivers and msi 3080 ventus just like you. Don't need to undervolt or anything, just lower the power limit. Try it and let us know.

 

[Dark_Skye]

So what you can do is try to ask the psu manufacturer or distro for advice, its a known issue by now with rtx 3000s and see what they can do. There are reports where they replaced such users with updated models of psu without an oversensitive ocp. Or just get a new psu, preferably higher one so future upgrade u wont face the same issue again.

Some info:
https://www.reddit.com/r/nvidia/comments/ity1hb/rtx_3080_489w_transient_spikes/
https://www.reddit.com/r/nvidia/comments/k6axtc/fyi_seasonic_and_rtx_3000_users_w_shutdowns/

Bro, convergent so nice will replace the psu with a less sensitive ocp unit? If true, I should go rma my sf750 already. I suspected my 3080 crashes also due to transient spikes. But I keep thinking is cyberpunk ****ed up. Then it happened in every game for me, even black Mesa. Now I set power limit to 75% all is fine again. 80% mostly fine but still can crash 4 times before the gpu decide to behave and let me play games.

 

[BlurKillerEX]

Just set power limit to 75% in afterburner. Works for me. I'm using sf750 BTW and before I did this, I was crashing like no tomorrow on all games. Now I can play for 6 hours straight without a single crash. Using latest drivers and msi 3080 ventus just like you. Don't need to undervolt or anything, just lower the power limit. Try it and let us know.

It's not crashing or causing any black screen issue anymore though. I feel a bit reluctant to download msi afterburner again since it might be the one that was causing the issue after all.

 

[noinimod]

So... Funny thing, I did a DDU again, installed one more fan in my PC, fiddled around with the cables a bit, uninstalled MSI afterburner.

I've been playing cyberpunk 2077 on RTX psycho settings for the past 2 hours without any blackouts. Didn't even undervolt anything this time.

Had almost the exact same experience with afterburner being the culprit of crashes.

Wasted a lot of time merry go round trying other solutions but the crashes only stopped when i uninstalled afterburner

 

[CopySeeSewDye]

Bro, convergent so nice will replace the psu with a less sensitive ocp unit? If true, I should go rma my sf750 already. I suspected my 3080 crashes also due to transient spikes. But I keep thinking is cyberpunk ****ed up. Then it happened in every game for me, even black Mesa. Now I set power limit to 75% all is fine again. 80% mostly fine but still can crash 4 times before the gpu decide to behave and let me play games.

Ocp tripping causes psu to shut down, not only gpu. If its only gpu its probably something else. 750w should definitely be enough, maybe try uninstall afterburner. Also, do not daisy chain your gpu power connection if u r doing so, even if it save space in a sfx build.