Question Windows 10 system crashing / not waking before and after replacing motherboard and video cards

[JasonTD]

Back in May/June I built a new PC with the following components:

Motherboard: ASUS AM4 TUF Gaming X570-Plus (Wi-Fi) ATX
CPU: AMD RYZEN 9 3900X 12-Core 3.8 GHz
Video card: EVGA GeForce RTX 2080 Ti XC ULTRA GAMING
RAM: G.SKILL Ripjaws V Series 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 3600
PSU: Seasonic FOCUS GX-1000, 1000W 80+ Gold
CPU Cooling: be quiet! 250W TDP Dark Rock Pro 4 CPU Cooler with Silent Wings - PWM Fan - 135 mm
OS Drive: SAMSUNG 970 PRO M.2 2280 512GB PCIe Gen3. X4, NVMe 1.3 64L V-NAND SSD
Case: Fractal Design Define 7 XL

I'm running it on the most current update of Windows 10, with all drivers updated and the BIOS updated to the most current version.

Everything ran great from June through mid-September. At that time, I started having a strange problem where the computer wouldn't wake from sleep in the morning. When I would finish at night I'd put it to sleep (like usual), and the light on the front of the case would pulse. When I would come back in the morning, the light would be solid as if the computer had awoken, but it was not running, and it would not boot or turn on beyond that unless I unplugged the case power, replugged it, and started up from the case power button.

This soon escalated to a point where the system wouldn't boot even after unplugging/replugging. I tried swapping the video card (2080 Ti) out for my old card (GeForce GTX 1070), and suddenly everything worked again. The computer would sleep, wake, boot, run, etc, normally just like it had from June onward. I took this to mean that the 2080 Ti card was bad, so I did a RMA with EVGA. The replacement 2080 Ti card arrived, and I swapped it into the system, and the sleep/wake/boot problems returned. Thinking I somehow got a second faulty card, I did another RMA, and they sent a third 2080 Ti card. Same thing with this one.

During all this, I'd been running with the 1070 card with no problem. But suddenly I started having the same issue with the 1070 that I'd been having with the 2080 Ti cards. The system still would boot, wake, sleep, etc, with the 1070, but every once in a while it would refuse to turn on or would just shut down altogether. Still, it was definitely more likely to turn on and more stable than with any of the three 2080 Ti cards. At this point, I eventually got the system to boot and remain stable with the 1070 card installed, and I just left it running without putting it to sleep for as long as possible. It ran successfully during that time without any crashes.

I tested the PSU with a multimeter and found that there were no problems there. BIOS also showed that the PSU was giving the proper outputs. I tried reseating the two RAM sticks, and that had no effect. This led me to think the problem was a faulty motherboard.

I jumped through a bunch of hoops with ASUS and finally got a RMA for the motherboard. The new one arrived last week, and I swapped it into the system and put the newest 2080 Ti card in. Everything started up perfectly after that, and it's been running great for about a week.

Jump to this morning, and the computer shut off while I was using it. I tried booting back up, and it wouldn't go. I unplugged and replugged the power and then tried turning it on from the power button. When I did that (just like with the previous problem instances) the CPU fan spun up, the motherboard LEDs lit orange, then red, then white, then green, and the system would just shut down and not finish booting.

After that, I swapped out the 2080 Ti card for my trusty 1070. With that installed, the system would turn on and boot to Windows. Shortly afterwards, when I tried to open applications (once with Chrome and once with Excel), it shut down again. I've gone through the attempted boot up sequence numerous times and finally seem to have it running and stable again right now with the 1070. I know it'll give me problems as soon as I reboot or put it to sleep, though, and I'm not able to run it at all with the 2080 Ti card installed. As of right now, the system has been running successfully on the 1070 card without crashing for about four hours.

Can someone please advise me on what to troubleshoot next or how to proceed? After all my troubleshooting so far, I suspect that the issue is either on the OS / software side or is some sort of incompatibility between this particular motherboard and my video cards (especially the 2080 Ti). But I'm stymied since the system has run fine for extended periods of time and has had the motherboard replaced once and the video card replaced numerous times.

Any help would be greatly appreciated. Thanks in advance!

 

[iiSlashr]

Are you using the real 8-pin PCIe connectors for all of the 2080Ti's power input? If you're using adapters, that's why. Also, if you can, return the 2080ti and just get a 3080 if you can find one available, better performing for half the cost.

 

[JasonTD]

Are you using the real 8-pin PCIe connectors for all of the 2080Ti's power input? If you're using adapters, that's why. Also, if you can, return the 2080ti and just get a 3080 if you can find one available, better performing for half the cost.

My PSU came with multiple PCIe cables, each with a split to two output ends. I'm using two of those (separate cables) and plugging one of the output ends into each of the two 2080ti power inputs (so each cable has one extra output end just dangling). That worked great for for 2-3 months, so it doesn't seem like that is the issue, unless I'm misunderstanding something there. Also, the 1070 is using the same cables (but just one for its single power input) and currently is running okay.

I wish I could swap out the 2080ti for the 3080, but it's too late for a return at this point. When I bought the 2080ti, I thought the disruption from the pandemic was going to delay the new cards into 2021. Felt pretty dumb when they released right after my return window closed!

 

[JasonTD]

Are you using the real 8-pin PCIe connectors for all of the 2080Ti's power input? If you're using adapters, that's why. Also, if you can, return the 2080ti and just get a 3080 if you can find one available, better performing for half the cost.

Quick question, because I realized I might have misunderstood-- When you ask about an adapter versus a real 8-pin connector, can you clarify what you mean?

The two cables I'm using are the PCIe cables that came with my Seasonic PSU. Each cable has an 8-pin connector on one end that goes into the CPU. The other end has two 6+2 pin connectors. I have one of those 6+2 pin connectors plugged into the 2080 Ti from each of the two cables. Is that good or should I be using a different type of cable than what came with the PSU?

 

[iiSlashr]

Quick question, because I realized I might have misunderstood-- When you ask about an adapter versus a real 8-pin connector, can you clarify what you mean?

The two cables I'm using are the PCIe cables that came with my Seasonic PSU. Each cable has an 8-pin connector on one end that goes into the CPU. The other end has two 6+2 pin connectors. I have one of those 6+2 pin connectors plugged into the 2080 Ti from each of the two cables. Is that good or should I be using a different type of cable than what came with the PSU?

6+2 power connectors carry less power than 8-pins. In total, you have 75W from the PCIe slot, 75W from the 6+2, and 150W from the 8-pin, meaning that you're drawing more power than you can provide during spikes in usage, as the 2080Ti frequently reaches ~280W, so it's possible it spikes above 300W in intense load scenarios. Make sure you have two full 8-pins plugged into the GPU, and let me know if that fixes your problem.

Edit: read wrong, you have 2 6+2 pin connectors on it, which means a max of only 225W, the 2080Ti draws 280W regularly so there's your problem.

 

[JasonTD]

6+2 power connectors carry less power than 8-pins. In total, you have 75W from the PCIe slot, 75W from the 6+2, and 150W from the 8-pin, meaning that you're drawing more power than you can provide during spikes in usage, as the 2080Ti frequently reaches ~280W, so it's possible it spikes above 300W in intense load scenarios. Make sure you have two full 8-pins plugged into the GPU, and let me know if that fixes your problem.

Edit: read wrong, you have 2 6+2 pin connectors on it, which means a max of only 225W, the 2080Ti draws 280W regularly so there's your problem.

This sounds very encouraging! I'm having trouble finding full 8-pin cables to order... Are there some particular ones you'd recommend? All the ones I'm finding right now are the 6+2 like I already have.

 

[JasonTD]

Update to this-- I emailed support at BTOS Integration to see if I could order those cables. The response I received back was:

"8pin PCIe connector uses 3 x +12V and 5 x ground wires ( power is supplied by those three +12V wires ).

In Seasonic’s 8pin to single 6+2pin PCIe connector cable, they use 3 x +12V and 4 x ground wire at the power supply side, then at the PCIe connector side, one of the ground wire is split into two ground wires. Since ground wire do not carry voltage ( power ), this does not affect the supply of power from power supply to your PCIe video card.

In Seasonic’s 8pin to two 6+2pin PCIe connector cable, they use 3 x 12V and 5 x ground wires.

So, changing the cable will not solve your problem."

What are your thoughts in light of that reply? I haven't been able to find the 8/8 cables for the Seasonic power supply anywhere so far.

 

[iiSlashr]

Update to this-- I emailed support at BTOS Integration to see if I could order those cables. The response I received back was:

"8pin PCIe connector uses 3 x +12V and 5 x ground wires ( power is supplied by those three +12V wires ).

In Seasonic’s 8pin to single 6+2pin PCIe connector cable, they use 3 x +12V and 4 x ground wire at the power supply side, then at the PCIe connector side, one of the ground wire is split into two ground wires. Since ground wire do not carry voltage ( power ), this does not affect the supply of power from power supply to your PCIe video card.

In Seasonic’s 8pin to two 6+2pin PCIe connector cable, they use 3 x 12V and 5 x ground wires.

So, changing the cable will not solve your problem."

What are your thoughts in light of that reply? I haven't been able to find the 8/8 cables for the Seasonic power supply anywhere so far.

I don't know, you said previously that you have 2 cables that are 6+2 and 8 pin at different point on the cable, so if that's the case you might be able to just use the 8-pins for your GPU.

 

[JasonTD]

I don't know, you said previously that you have 2 cables that are 6+2 and 8 pin at different point on the cable, so if that's the case you might be able to just use the 8-pins for your GPU.

The PCIe cables I have (the ones that came with the Seasonic PSU) are solid 8-pin on the end that plugs into the power supply. The other end has two 6+2-pin connectors. I currently have two of those cables plugged into the PSU (with the 8-pin end). Then I have one of the 6+2-pin ends from each cable plugged into the 2080Ti card. So it's two cables overall, and each of them has one 6+2 connector plugged into the video card and one extra 6+2 connector that remains unconnected to anything. I don't have any cables that are solid 8-pin on both sides and can't find those sold anywhere.

Is there a good way for me to test these cables in particular to see if there might be something wrong with them? I tested the PSU itself but am not sure the best way to test the cables for output.

Thanks!

 

[JasonTD]

Updating this:

I contacted Seasonic support and asked them for help with the potential PCIe issue. They said:

"This is not a correct assessment. A 6+2 is the same as an 8pin. They provide more than 225w power draw. Your PC crashing is most likely a graphics card issue or a PSU issue and not a cable issue."

After troubleshooting with them, the PSU still seems okay. And whatever the issue is seems independent of the specific graphics card or the specific motherboard, given that the same thing has happened with three different 2080 Ti cards, one 1070 card, and two different ASUS TUF Gaming X570-Plus motherboards. When the computer will boot up, it runs for an extended period with no problems at all, no matter what I'm doing on it (gaming, audio/video editing, general work, etc), so I feel like it's probably not an issue with the RAM or CPU. It seems like the system wouldn't be able to run continuously and successfully for this long if something were wrong with them. It currently has been running for a couple of days uninterrupted with the 1070 video card in it. I know I'll likely have trouble again if I put it to sleep or reboot, or if I put a 2080Ti card back in it.

I'd love to get some additional suggestions of things to check. It's a very frustrating problem and seems to only be more confounding by the day.

 

[iiSlashr]

Updating this:

I contacted Seasonic support and asked them for help with the potential PCIe issue. They said:

"This is not a correct assessment. A 6+2 is the same as an 8pin. They provide more than 225w power draw. Your PC crashing is most likely a graphics card issue or a PSU issue and not a cable issue."

After troubleshooting with them, the PSU still seems okay. And whatever the issue is seems independent of the specific graphics card or the specific motherboard, given that the same thing has happened with three different 2080 Ti cards, one 1070 card, and two different ASUS TUF Gaming X570-Plus motherboards. When the computer will boot up, it runs for an extended period with no problems at all, no matter what I'm doing on it (gaming, audio/video editing, general work, etc), so I feel like it's probably not an issue with the RAM or CPU. It seems like the system wouldn't be able to run continuously and successfully for this long if something were wrong with them. It currently has been running for a couple of days uninterrupted with the 1070 video card in it. I know I'll likely have trouble again if I put it to sleep or reboot, or if I put a 2080Ti card back in it.

I'd love to get some additional suggestions of things to check. It's a very frustrating problem and seems to only be more confounding by the day.

If all the 2080TIs you're getting don't work, you might want to just ask EVGA for an RMA, and maybe see if they'll send you the cheaper 3080 instead assuming they have one in stock.

Now that I think about it, it may just be a driver thing... It didn't really cross my mind before, but make sure you use DDU (display driver uninstaller) and reinstall the 2080Ti's drivers.

 

[JasonTD]

Update to this -- After trying messing with drivers, changing out video cards, changing out the motherboard, etc, etc, etc, I finally revisited the power supply. It had tested as working correctly, but I suspected it had to be something to do with power since the problems generally happened during power-up, waking from sleep, etc. Yesterday I picked up a new PSU (EVGA 850 G3 Supernova), and so far the system is running without a problem. I'm going to keep testing it for about a week, and if it continues working, I'll RMA my Seasonic PSU.