PC shutdown + fans going crazy + no temp problems

[Agall]

6 minutes ago, GuiltySpark_ said:

Every single time this scenario comes up. I mean EVERY time, its a dead/dying GPU. Specifically something in the power stage. Same symptoms, PC goes black but doesn't turn off and the GPU fans go full blast. Countless times I've seen this. Can happen at random but always under load and some games are much more susceptible. 

 

The PSU in this instance clearly isn't tripping a protection circuit and shutting the PC completely off. 

Just because the PSU isn't tripping on OCP doesn't mean its not potentially related to a power spiking issue. The PSU still has to supply stable 12V rails, which it can struggle to maintain. The whole issue of capacitance in PSUs extends to more than just OCP.

 

OP's 3080 could be more sensitive to dips in the 12V rail as example, caused by power spiking that is not enough to trip OCP. Where my suggestion of lower TDP is a $0 test and can do a great job at variable elimination in this kind of scenario.

 

HWinfo can track that sort of thing, if the 12V rail is dipping below like 11.6V as example.

[Xemosh]

1 minute ago, Agall said:

That would make a software issue even less likely, so at that point (assuming its not my initial theory) then I'd just suspect Diablo 4 itself just disagreeing with your hardware configuration. That's an unlikely scenario, but we'd be grasping at straws at that point given the variables already tested.

Then the only solution would be some sort of fix from blizzards end?

I'm currently monitoring gpu power draw in HWMonitor. If you've used that before, is the "Max" value the highest value I've encountered since I open the application?

[Agall]

Just now, Xemosh said:

Then the only solution would be some sort of fix from blizzards end?

I'm currently monitoring gpu power draw in HWMonitor. If you've used that before, is the "Max" value the highest value I've encountered since I open the application?

I'd recommend yeeting hwmonitor and installing hwinfo64 instead. Something I did recently myself, within the last couple months. HWmonitor doesn't track Ryzen's PBO boosting properly and I was getting reports of 6GHz or even 11GHz boosting on my 7950x3D.

[Xemosh]

7 minutes ago, Agall said:

I'd recommend yeeting hwmonitor and installing hwinfo64 instead. Something I did recently myself, within the last couple months. HWmonitor doesn't track Ryzen's PBO boosting properly and I was getting reports of 6GHz or even 11GHz boosting on my 7950x3D.

Is there a way to store the data from the monitoring, in order to see the values before the crash? Or would that be impossible?

 

I don't see the gpu power draw anywhere. How do I turn that on?

[Agall]

13 minutes ago, Xemosh said:

Is there a way to store the data from the monitoring, in order to see the values before the crash? Or would that be impossible?

 

I don't see the gpu power draw anywhere. How do I turn that on?

Oh on HWinfo64, I'd recommend running 'sensors-only' on launch.

[Xemosh]

Just now, Agall said:

Oh on HWinfo64, I'd recommend running 'sensors-only' on launch.

This is the highest I've gotten the last 10 minutes. It JUST did a jump from having the max being 140W to 190W. But without jinxing it, this could suggest it's something to do with the power spikes you were mentioning

[Agall]

Just now, Xemosh said:

This is the highest I've gotten the last 10 minutes. It JUST did a jump from having the max being 140W to 190W. But without jinxing it, this could suggest it's something to do with the power spikes you were mentioning

Are you running 75% TDP still?

[Xemosh]

1 minute ago, Agall said:

Are you running 75% TDP still?

Yes, still: 75% power, quality preset: low, and 60 fps limit and dlss off

[Xemosh]

17 minutes ago, Agall said:

Are you running 75% TDP still?

I have to leave for about 45 minutes. Should I just leave the pc and game on, and stay in town? If that hasn't crashed, it probably is the PSU, right? Would an extra 150W be enough to ensure it doesn't happen again?

And would I be able to keep 75% TDP but change the framerate and graphics settings until I buy a new PSU?

Can I reuse the cables already in my pc, or should I replace everything? I didn't build the pc myself and haven't done that before (only changed gpu and ram before) so I assume changing all the cables is going to be rather difficult

 

Edit: It has now spiked to 199.904W

[Agall]

29 minutes ago, Xemosh said:

I have to leave for about 45 minutes. Should I just leave the pc and game on, and stay in town? If that hasn't crashed, it probably is the PSU, right? Would an extra 150W be enough to ensure it doesn't happen again?

And would I be able to keep 75% TDP but change the framerate and graphics settings until I buy a new PSU?

Can I reuse the cables already in my pc, or should I replace everything? I didn't build the pc myself and haven't done that before (only changed gpu and ram before) so I assume changing all the cables is going to be rather difficult

 

Edit: It has now spiked to 199.904W

Using PSU cables from other power supplies is a big big big no no no. Only the component side connectors are standard and PSU manufacturers follow their own standards for the PSU side connectors, so its possible to fry components using cables from other PSUs.

 

Swapping a power supply isn't too difficult, I'd recommend just being careful around some of the other connectors when pulling the cables out to not damage them, especially USB 3.0 headers since those are quite fragile. I'd unplug them from the front and then carefully pull them out through the back, then remove the power supply physically.

 

 

You could just leave the game on, but if you're worried about it, best to defer testing until you're able to monitor the system.

[Xemosh]

4 minutes ago, Agall said:

Using PSU cables from other power supplies is a big big big no no no. Only the component side connectors are standard and PSU manufacturers follow their own standards for the PSU side connectors, so its possible to fry components using cables from other PSUs.

 

Swapping a power supply isn't too difficult, I'd recommend just being careful around some of the other connectors when pulling the cables out to not damage them, especially USB 3.0 headers since those are quite fragile. I'd unplug them from the front and then carefully pull them out through the back, then remove the power supply physically.

 

 

You could just leave the game on, but if you're worried about it, best to defer testing until you're able to monitor the system.

I have these in my system:

- Vision 24pin Extension (Sleeved)
- Vision 6+2pin Extension (Sleeved)

 

I assume these are okay to reuse. They're only for the motherboard and gpu though.

[Agall]

12 minutes ago, Xemosh said:

I have these in my system:

- Vision 24pin Extension (Sleeved)
- Vision 6+2pin Extension (Sleeved)

 

I assume these are okay to reuse. They're only for the motherboard and gpu though.

Extensions are 1:1, but modular cables usually are not. I believe the only true standard 1:1 cable are 12VHPWR cables per the PCIe 5.0 standard, something I've asked Seasonic about (including the other stuff). The power supply end has no other standards for how its wired and what connectors they use. The easiest example of this are 24pin motherboard connectors, where the PSU end can be anything from 1:1, 18+10, 24+4, etc.

[Xemosh]

10 minutes ago, Agall said:

Extensions are 1:1, but modular cables usually are not. I believe the only true standard 1:1 cable are 12VHPWR cables per the PCIe 5.0 standard, something I've asked Seasonic about (including the other stuff). The power supply end has no other standards for how its wired and what connectors they use. The easiest example of this are 24pin motherboard connectors, where the PSU end can be anything from 1:1, 18+10, 24+4, etc.

Not sure I understand your response. Can I use the extensions again with a different PSU?

[Agall]

18 minutes ago, Xemosh said:

Not sure I understand your response. Can I use the extensions again with a different PSU?

Yes, extensions are 1:1 by design. 1:1 being that each end of the connector are wired exactly pin to pin, extensions just having a female and male end. Some modular power supplies are 1:1 in this way but with two male connectors, although that's not common anymore.

 

Contrarily plenty of power supply manufacturers don't wire the VGA or EPS 8 pin connectors the same way at the power supply end, so you could end up putting 12V into a ground connection, therefore frying whatever components are downstream. This is even applicable to different models of power supplies from the same company. I imagine there's an internal wiring reason for this, where its easier to bus different connector's 12V, GND, 5V, 3.3V together than having them potentially spread out simply because of modular header connector positioning and internal wiring of the power supply.

 

12VPWR seems to be the only exception I can find, which is by design a 1:1 cable with really no benefit to changing the wiring, so its safe to say that any 1:1 12VHPWR cable will work with any 12VHPWR native power supply, which is unique in the space, even for models from the same manufacturer.

[Xemosh]

50 minutes ago, Agall said:

Using PSU cables from other power supplies is a big big big no no no. Only the component side connectors are standard and PSU manufacturers follow their own standards for the PSU side connectors, so its possible to fry components using cables from other PSUs.

 

Swapping a power supply isn't too difficult, I'd recommend just being careful around some of the other connectors when pulling the cables out to not damage them, especially USB 3.0 headers since those are quite fragile. I'd unplug them from the front and then carefully pull them out through the back, then remove the power supply physically.

 

 

You could just leave the game on, but if you're worried about it, best to defer testing until you're able to monitor the system.

So I'm back at the pc. It didn't crash, but it had went to sleep. I guess it's a standard setting I forgot to turn off after I reset my PC. But I think that proves it's the PSU that's the issue. Or rather, the GPU is the issue and my PSU doesn't supply enough Watt.

So, I'll have to get a new PSU. Do I have to think about size? My current one seems quite small (or average, but I saw larger ones). Would this be a good replacement?

With 75% TDP would I be able to bump up my graphics again? or at least the framerate? Or do you think that would draw too much power? Could I potentionally turn it down to 50% TDP with higher graphics or would that just be stupid?

[Agall]

18 minutes ago, Xemosh said:

So I'm back at the pc. It didn't crash, but it had went to sleep. I guess it's a standard setting I forgot to turn off after I reset my PC. But I think that proves it's the PSU that's the issue. Or rather, the GPU is the issue and my PSU doesn't supply enough Watt.

So, I'll have to get a new PSU. Do I have to think about size? My current one seems quite small (or average, but I saw larger ones). Would this be a good replacement?

With 75% TDP would I be able to bump up my graphics again? or at least the framerate? Or do you think that would draw too much power? Could I potentionally turn it down to 50% TDP with higher graphics or would that just be stupid?

I've recommended on this given issue a few times various %TDP, depending on the variables. I would stick to 75% and see if you run into the issue over the course of a few days, potentially while waiting for whatever new PSU to show up.

 

I have a Corsair RM1000x, only thing to note is what type cables it uses, simply because if you do replace your RTX 3080 with something newer with 12VHPWR, then being able to use a 1st party conversion cable is best. My RM1000x wasn't a 'type-4' that could use their 1st party conversion cable so I bought a Seasonic Vertex 1200W to replace it for native 12VHPWR in my primary rig.

 

My personal recommendations steer towards Seasonic recently, but I've mostly owned Corsair, EVGA, or Silverstone power supplies over the years (Silverstone was the standard for SFX PSUs for a while).

 

Seasonic FOCUS GX-1000, 1000W 80+ Gold, Full-Modular, Fan Control in Fanless, Silent, and Cooling Mode, 10 Year Warranty, Perfect Power Supply for Gaming and Various Application, SSR-1000FX. - Newegg.com

 

If you can find this PSU for about the same price, I'd recommend Seasonic over Corsair any day, otherwise I'd say Seasonic PSUs are worth ~20% more than Corsair, as a rough price comparison. They'll generally underrate the capacity and efficiency of the PSU, so overdrawing usually isn't an issue if you're within the specifications.

[Xemosh]

5 minutes ago, Agall said:

I've recommended on this given issue a few times various %TDP, depending on the variables. I would stick to 75% and see if you run into the issue over the course of a few days, potentially while waiting for whatever new PSU to show up.

 

I have a Corsair RM1000x, only thing to note is what type cables it uses, simply because if you do replace your RTX 3080 with something newer with 12VHPWR, then being able to use a 1st party conversion cable is best. My RM1000x wasn't a 'type-4' that could use their 1st party conversion cable so I bought a Seasonic Vertex 1200W to replace it for native 12VHPWR in my primary rig.

 

My personal recommendations steer towards Seasonic recently, but I've mostly owned Corsair, EVGA, or Silverstone power supplies over the years (Silverstone was the standard for SFX PSUs for a while).

 

Seasonic FOCUS GX-1000, 1000W 80+ Gold, Full-Modular, Fan Control in Fanless, Silent, and Cooling Mode, 10 Year Warranty, Perfect Power Supply for Gaming and Various Application, SSR-1000FX. - Newegg.com

 

If you can find this PSU for about the same price, I'd recommend Seasonic over Corsair any day, otherwise I'd say Seasonic PSUs are worth ~20% more than Corsair, as a rough price comparison. They'll generally underrate the capacity and efficiency of the PSU, so overdrawing usually isn't an issue if you're within the specifications.

I'll look into the Seasonic PSU. I don't plan on replacing the GPU anytime soon. My GPU before I upgraded to the 3080 was a 1080.

I'll stick to 75% and potentially changing the graphics and framerate again, until I can figure out a solution to the new PSU.

 

Thank you so much for your help! You've been really helpful, and I hope getting a new PSU solves the issue. 

[ceph]

Hey buddy! I have the exact same problem with diablo4. Same same same thing. Did you find the solution? Is it the PSU finally? 

[Xemosh]

On 6/2/2023 at 8:17 PM, Agall said:

I've recommended on this given issue a few times various %TDP, depending on the variables. I would stick to 75% and see if you run into the issue over the course of a few days, potentially while waiting for whatever new PSU to show up.

 

I have a Corsair RM1000x, only thing to note is what type cables it uses, simply because if you do replace your RTX 3080 with something newer with 12VHPWR, then being able to use a 1st party conversion cable is best. My RM1000x wasn't a 'type-4' that could use their 1st party conversion cable so I bought a Seasonic Vertex 1200W to replace it for native 12VHPWR in my primary rig.

 

My personal recommendations steer towards Seasonic recently, but I've mostly owned Corsair, EVGA, or Silverstone power supplies over the years (Silverstone was the standard for SFX PSUs for a while).

 

Seasonic FOCUS GX-1000, 1000W 80+ Gold, Full-Modular, Fan Control in Fanless, Silent, and Cooling Mode, 10 Year Warranty, Perfect Power Supply for Gaming and Various Application, SSR-1000FX. - Newegg.com

 

If you can find this PSU for about the same price, I'd recommend Seasonic over Corsair any day, otherwise I'd say Seasonic PSUs are worth ~20% more than Corsair, as a rough price comparison. They'll generally underrate the capacity and efficiency of the PSU, so overdrawing usually isn't an issue if you're within the specifications.

 

Not sure if you're still following this thread, but I'll try anyways.

So since I started this thread, I've been playing with a power limiter on my GPU on 50%. I finally just installed a new PSU with 1000W, and thought I would be able to play games properly again. But my pc just crashed again as usual, so getting a new PSU wasn't the solution.

[Xemosh]

On 6/2/2023 at 5:48 PM, GuiltySpark_ said:

Every single time this scenario comes up. I mean EVERY time, its a dead/dying GPU. Specifically something in the power stage. Same symptoms, PC goes black but doesn't turn off and the GPU fans go full blast. Countless times I've seen this. Can happen at random but always under load and some games are much more susceptible. 

 

The PSU in this instance clearly isn't tripping a protection circuit and shutting the PC completely off. 

What causes the gpu to start dying? I've never had this issue with a GPU before, which I've had for a longer time

[Agall]

On 4/27/2024 at 7:03 AM, Xemosh said:

 

Not sure if you're still following this thread, but I'll try anyways.

So since I started this thread, I've been playing with a power limiter on my GPU on 50%. I finally just installed a new PSU with 1000W, and thought I would be able to play games properly again. But my pc just crashed again as usual, so getting a new PSU wasn't the solution.

Sounds like a GPU issue. I would install HWinfo64 and track some data for it. Unless you happen to be using the same cable extensions or such.

 

TUF-RTX3080-O10G-GAMING｜Graphics Cards｜ASUS USA

 

There's an October 2023 release date vBIOS for the card, I would recommend installing this to see if it fixes the issue. Otherwise, you'll just need to send a ton of data from HWinfo64, like 12V voltages, temperatures of all GPU sensors, boost clocks, TDP, etc. Basically the whole section for the GPU and more.

 

 

[Xemosh]

6 hours ago, Agall said:

Sounds like a GPU issue. I would install HWinfo64 and track some data for it. Unless you happen to be using the same cable extensions or such.

 

TUF-RTX3080-O10G-GAMING｜Graphics Cards｜ASUS USA

 

There's an October 2023 release date vBIOS for the card, I would recommend installing this to see if it fixes the issue. Otherwise, you'll just need to send a ton of data from HWinfo64, like 12V voltages, temperatures of all GPU sensors, boost clocks, TDP, etc. Basically the whole section for the GPU and more.

 

 

Tried the BIOS update, and it didn't fix it.

How do I get that data? I would assume, the data is interesting right before it crashes, but can I get it to save the data constantly?

[Agall]

41 minutes ago, Xemosh said:

Tried the BIOS update, and it didn't fix it.

How do I get that data? I would assume, the data is interesting right before it crashes, but can I get it to save the data constantly?

I'd just continuously monitor it. I'd find a relatively stable test and see if you're getting high wattage spikes or a +25C hotspot to gpu core temp delta, or crazy dips in core frequency, etc.