• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Rtx 3080 Shutdown Issues

ChugginHopps

New Member
Joined
Feb 9, 2022
Messages
12 (0.01/day)
I have a very odd problem that I cannot figure out. Bought a new PSU and EVGA RTX 3080 FTW3 ultra graphics card. All other components previously worked in my previous build with a GTX 1080 without issue.

Started with an 850 watt PSU. Had frames drop down to like 2FPS and would never come back. Would have to shut down to get a game to play again. RMA'D the psu and got the same model 850 watt. Had the same issue and then the PSU started tripping. Only resetting the power switch on the PSU would allow it to work again. When this happens the RAM led's stayed illuminated.

Ordered a different brand PSU that is 1,000 watts. Installed and played a few hours and the PSU tripped. Decided to undervolt the 3080 and the issue seems to have gone away. Undervolt is set to around 850mv at 1830mhz

My question is why won't this card work when set to stock settings? Is it possible that the GPU draws that much power to trip a 1,000 watt PSU? Prior to installing the card I ran ddu so that no drivers from the 1080 were left. I feel like I am leaving performance on the table and not getting what I paid for by undervolting.

System Specs:

MB - msi x570 unify
CPU - Ryzen 7 3700x
Ram - 32gb Corsair vengeance RGB pro
GPU - EVGA Rtx 3080 FTW3 Ultra
PSU - Seasonic gx 1000
Case - O11 Dynamics
Cooling - Bitspower distribution plate with pump and 360 radiator. 9 fans total. 3 on radiator, 3 on the back side and 3 on the bottom
Drives - 1tb M.2 Ssd, 1 - 2TB 2.5" Hdd 1 - external 3tb backup drive
Windows 11

All GPU drivers are updated and am running 3 separate pcie cables from the PSU. Temps for GPU max out around 65C (when not undervolted) CPU temps hang around 50-55C

Thanks in advance
 
Is the power your PC gets from the wall "clean"? This could very well be the true issue.
 
evga and FE 3080/3090s like to trip seasonic psus, thats like a known feature
It did it to this PSU as well as 2 fr
Is the power your PC gets from the wall "clean"? This could very well be the true issue.
It's plugged into a surge protector. I don't think you should have to have a power conditioner to get these to work properly.
 
It's plugged into a surge protector. I don't think you should have to have a power conditioner to get these to work properly.
Depends if the wall power is clean from the get go or not. We already tried different PSUs and it doesn't make much sense otherwise. Unless GPU itself is faulty.
 
I have a very odd problem that I cannot figure out. Bought a new PSU and EVGA RTX 3080 FTW3 ultra graphics card. All other components previously worked in my previous build with a GTX 1080 without issue.

Started with an 850 watt PSU. Had frames drop down to like 2FPS and would never come back. Would have to shut down to get a game to play again. RMA'D the psu and got the same model 850 watt. Had the same issue and then the PSU started tripping. Only resetting the power switch on the PSU would allow it to work again. When this happens the RAM led's stayed illuminated.

Ordered a different brand PSU that is 1,000 watts. Installed and played a few hours and the PSU tripped. Decided to undervolt the 3080 and the issue seems to have gone away. Undervolt is set to around 850mv at 1830mhz

My question is why won't this card work when set to stock settings? Is it possible that the GPU draws that much power to trip a 1,000 watt PSU? Prior to installing the card I ran ddu so that no drivers from the 1080 were left. I feel like I am leaving performance on the table and not getting what I paid for by undervolting.

System Specs:

MB - msi x570 unify
CPU - Ryzen 7 3700x
Ram - 32gb Corsair vengeance RGB pro
GPU - EVGA Rtx 3080 FTW3 Ultra
PSU - Seasonic gx 1000
Case - O11 Dynamics
Cooling - Bitspower distribution plate with pump and 360 radiator. 9 fans total. 3 on radiator, 3 on the back side and 3 on the bottom
Drives - 1tb M.2 Ssd, 1 - 2TB 2.5" Hdd 1 - external 3tb backup drive
Windows 11

All GPU drivers are updated and am running 3 separate pcie cables from the PSU. Temps for GPU max out around 65C (when not undervolted) CPU temps hang around 50-55C

Thanks in advance
Are you using a pigtail from the PSU, or seperate PSU cables for each connector on the GPU?
Are you using PCI-E power extensions?
Are you using a PCI-E riser?
 
My guess is that the GPU is faulty.

If the wall power was bad, it would have manifested itself with the OP's previous hardware.

My primary gaming PC (the one described in my system specs) draws more power (Ryzen 5900X + 3080 Ti OC) with a smaller PSU (800W SFX-L unit) and I've never experienced a shutdown. This gaming build is not on a UPS.

The Ryzen 3700X is a 65W TDP part (I have one in another build); it sips power for an 8-core/16-thread desktop CPU.

It's also worth pointing out that under normal usage, it is rare to max out both the CPU and GPU at the same time so the maximum theoretical power draw between OP's CPU and GPU will hover around 50% of the 1000W rated capacity of the PSU.

I don't do any fancy overclock with my 3700X and it maxes out at 114 W (PBO is enabled). And that's during a Handbrake encode or Cinebench R23 run, not during gaming.
 
My guess is that the GPU is faulty.

If the wall power was bad, it would have manifested itself with the OP's previous hardware.

My primary gaming PC (the one described in my system specs) draws more power (Ryzen 5900X + 3080 Ti OC) with a smaller PSU (800W SFX-L unit) and I've never experienced a shutdown. This gaming build is not on a UPS.

The Ryzen 3700X is a 65W TDP part (I have one in another build); it sips power for an 8-core/16-thread desktop CPU.

It's also worth pointing out that under normal usage, it is rare to max out both the CPU and GPU at the same time so the maximum theoretical power draw between OP's CPU and GPU will hover around 50% of the 1000W rating of the PSU.
Yep, unless Mussels is right and the connection is bad. But this is unlikely, he already switched things too often for that to happen.
 
If the wall power was bad, it would have manifested itself with the OP's previous hardware.
Nah. i had all sorts of issues from known good hardware when i moved from a 1080 to a 3080.
Extensions caught fire, riser died and wouldnt do PCI-E 4.0 any longer, black screen crashes, reboots ,driver not responding errors - these cards are fussier.
 
Nah. i had all sorts of issues from known good hardware when i moved from a 1080 to a 3080.
Extensions caught fire, riser died and wouldnt do PCI-E 4.0 any longer, black screen crashes, reboots ,driver not responding errors - these cards are fussier.
That's a shame.

I don't see my 3080 Ti being any fussier than another build across the room with an RX 550 or the Mac mini 2018 right next to it.

As far as I can tell Ampere cards generate more heat and that's the main difference.

And I've lived in this particular building for 12 years. It's not like these are my first computers; the wiring in my place certainly is not pristine. The construction is 50 years old, the conductors are original Romex.

In fact, almost all of the places I've lived in feature wiring of the same age and quality; it's all from the big California housing boom in the late 60s/early 70s. Every dwelling in my price point has the same build quality. The conductors are all the same. The electrical panels came from a couple of companies (one of which went bankrupt after it was sued for defective panel design). Same with the plumbing, same with water valves.
 
Last edited:
As far as I can tell Ampere cards generate more heat and that's the main difference.
No he is absolutely right. As long as you don't have a absolutely normally working card, the higher the output of the card the higher the failure rate, in general. This is easily observable with rma rates of smaller GPUs vs highend GPUs.
 
Extensions caught fire, riser died and wouldnt do PCI-E 4.0 any longer, black screen crashes, reboots ,driver not responding errors - these cards are fussier.
@Mussels, whose other peoples' wrongs are his fussy!
3977067.jpg
 
No he is absolutely right. As long as you don't have a absolutely normally working card, the higher the output of the card the higher the failure rate, in general. This is easily observable with rma rates of smaller GPUs vs highend GPUs.
Well then that's the fault of the GPU rather than house power.

It's worth pointing out that there are a ton of AMD and Nvidia engineers living in construction of the same age and quality as the unit I live in.

Whether an Nvidia engineer is living in a $2.5 million Eichler in Saratoga or a $650,000 condo in Mountain View built by Thrust IV doesn't change the fact that both builders probably used the same conductors.

For what it's worth, I own three Ampere GPUs and they all work fine. They have all been in at least two builds each with wimpier GPU predecessors. Right now I have five PC cases in my house and four of them have had multiple builds. Only the high end gaming system (system specs) has stayed stable. But I assure you that all of the primary components lived elsewhere in another build prior to showing up in the Lian-Li O11D Mini case.

Looking at other Windows PC Q&A forums, I don't see a massive tsunami of GPU failures when switch from an old one to a new one. They are the exception not the rule.

And now that we are aware of it, things like PCIe power cable extensions are a more likely failure point. With the amount of current these things carry, there's more heat-related expansion and contraction which can cause shorts.

Even in a well-maintained house the best practice is to periodically check the screws holding down conductors.
 
Last edited:
Well then test it one time without the extension.
I'll try that at some point. It's a little more difficult in the O11 being that the PSU is in back. I have an extra 650 watt I may hook up to just the GPU and jump the 24 pin to make it easier. Should isolate the issue if it's the card
 
Well then that's the fault of the GPU rather than house power.
Of course, but I still think it was worth mentioning.

I'll try that at some point. It's a little more difficult in the O11 being that the PSU is in back. I have an extra 650 watt I may hook up to just the GPU and jump the 24 pin to make it easier. Should isolate the issue if it's the card
Yep I kinda expected that. Good luck
 
That's a shame.

I don't see my 3080 Ti being any fussier than another build across the room with an RX 550 or the Mac mini 2018 right next to it.

As far as I can tell Ampere cards generate more heat and that's the main difference.

And I've lived in this particular building for 12 years. It's not like these are my first computers; the wiring in my place certainly is not pristine. The construction is 50 years old, the conductors are original Romex.

In fact, almost all of the places I've lived in feature wiring of the same age and quality; it's all from the big California housing boom in the late 60s/early 70s. Every dwelling in my price point has the same build quality. The conductors are all the same. The electrical panels came from a couple of companies (one of which went bankrupt after it was sued for defective panel design). Same with the plumbing, same with water valves.
They draw more power.
Simple as that.

Some devices are perfectly fine with a 150W card, then fail miserably when a card can draw 350W - that's not hard to understand

This why setups using extensions, pigtails and so on can be "fine" until a more demanding GPU is added
 
My 3080ti shuts down every psu I've tested it with under 1000w.

psu I've tried that shut down. Seasonic GX 750, RM 750, RMx 850, Prime Tx 850, G2 850.

PSU that have worked just fine with extensive testing. Prime GX-1000, Prime TX-1000.

I currently have a build I've done with a ftw 3080 10G and a Evga G3 750 that works fine though. Zero OC stock power limits.

Also oddly enough I tested all those lower wattage psu with a 3090 Zotac that has dual 8 pin instead of 3x and they all didn't shut down.

In my testing what was most easily repeatable shutdown was rapidly going from a low load 20-40% gpu usage instantly to 100% load. If I ran a 2-3 hour loop on any of the psu at 100% load none of them shut down.
 
Some devices are perfectly fine with a 150W card, then fail miserably when a card can draw 350W - that's not hard to understand

This why setups using extensions, pigtails and so on can be "fine" until a more demanding GPU is added
Which is why I wrote that the extension is a more likely failure point.

I don't own the Seasonic 1000W SFX unit that the OP does (although I own two Seasonic Focus Gold ATX units) so I'm not sure how long the cables are.

OP did not disclose in their original post that extensions were being used.

Some of the cables included with the Silverstone SX800 PSU are too short for the O11D Mini so I ordered two custom ones. In the meantime I used extension cables that worked fine. I don't like extension cables -- even if they work fine -- mostly because they are way too long and create the problem of finding a place to stuff the extra cable length hence the custom length cables.

Again, going back to the OP's original post, I wouldn't point a finger at the PSU being inadequate. More likely this is a GPU problem or -- with the tardy disclosure -- a problem with one of the PCIe power extension cables.
 
Last edited:
They draw more power.
Simple as that.

Some devices are perfectly fine with a 150W card, then fail miserably when a card can draw 350W - that's not hard to understand

This why setups using extensions, pigtails and so on can be "fine" until a more demanding GPU is added
I'll test this probably later this week. I bought the extensions at microcenter and instead of using my better judgement and getting cablemod, I went with their inland brand to save $10. Shame on me

Which is why I wrote that the extension is a more likely failure point.

I don't own the Seasonic 1000W SFX unit that the OP does (although I own two Seasonic Focus Gold ATX units) so I'm not sure how long the cables are.

OP did not disclose in their original post that extensions were being used.

Some of the cables included with the Silverstone SX800 PSU are too short for the O11D Mini so I ordered two custom ones. In the meantime I used extension cables that worked fine. I don't like extension cables -- even if they work fine -- mostly because they are way too long and create the problem of finding a place to stuff the extra cable length hence the custom length cables.
Never thought extension cables would be any kind of issue since they are just a direct line extension. I'll know more later. May try to wire it up tomorrow and test over the next week or so.
 
Never thought extension cables would be any kind of issue since they are just a direct line extension. I'll know more later. May try to wire it up tomorrow and test over the next week or so.
It's not a problem whether or not they were wired wrong. It has to do with build quality and durability.

Remember that GPU PCIe power cables are carrying a sizable amount of current, especially with a high-end GPU like your 3080 and my 3080 Ti.

I've been using computers long enough where I've seen pretty much every single type of cable fail: power, Ethernet, IDE, SATA, SCSI, PS/2, ADB, USB, video, serial, parallel, Toslink, coax, POTS, FibreChannel, RCA, S-Video, HDMI, DisplayPort, VGA, DB9, DB15, whatever, you name it.
 
It's not a problem whether or not they were wired wrong. It has to do with build quality and durability.

Remember that GPU PCIe power cables are carrying a sizable amount of current, especially with a high-end GPU like your 3080 and my 3080 Ti.

I've been using computers long enough where I've seen pretty much every single type of cable fail: power, Ethernet, IDE, SATA, SCSI, PS/2, ADB, USB, video, serial, parallel, Toslink, coax, POTS, FibreChannel, whatever, you name it.
This is only my 4th build so I'm fairly new. This time I went custom hard tube so was really happy when it all leak tested fine and started off running great. Then this came up and kind of took some of the joy away because of the frustration with it. Never had an issue before but these cards are an entirely different animal.
 
Low quality post by Kanan
Back
Top