Odd power(?) issue

Ok.... I'm at a loss with this one. Hoping someone here may have some insight.... maybe came across the same issue in the past and might have an idea on what could be the problem.

Asus TUF Gaming X670E Plus WiFi
Ryzen 9 7950X
64GB (4x16GB) DDR5 5200 (running at 3600mhz)
NVidia RTX 4090 FE
Corsair RM1000e 1000W PSU
GPU connected with CableMod 12VHPWR 4>1 adapter

For a while, in my NZXT H7 Flow, i was using the NZXT vertical GPU bracket. But i kept getting this issue where out of nowhere, my display would go black and as far as i can tell, the fans on the 4090 (if not every fan in the computer) cranks up to 100%, and the lighting on the 4090 shuts off. But the computer itself didn't crash. A video that may have been playing at the time this happens can still be heard playing as if nothing happened.

I thought to myself "Maybe it's an issue with the riser cable on the vert mount?" So i took the card out and put it directly in the slot. Seemed to have solved this issue, as it hadn't done it for well over a month now.

Until yesterday. Computer was just sitting at the desktop when i went upstairs to grab something. Came back down and heard the fans cranked, and no lights on the GPU.

I'm kinda leaning towards it being a PSU issue more than a GPU issue, but i could be wrong. Maybe also a possible motherboard issue? I'm assuming if the board for some reason suddenly drops it's connection to the PCI-E slot, there's a good chance it could cause the lighting on the card to go out? Or just unstable power from the PSU causes it to shut off to protect itself somehow? I'm at a loss. Never encountered anything like this in my many many years of PC building and ownership.

 

Have you disabled the iGPU on the 7950X?

Do you need 64GB of RAM? why not run 2x 16GB config at > 5600MT?

 

Yes, the iGPU is disabled.

And mostly running the 64gb cause I got one of the pairs for free. Was actually debating buying a 32GBx2 kit instead so I can turn EXPO on.

 

So you are running 2 kits of 2x16GB? Your system RAM might not be fully stable.

Ryzen4 benefit from RAM speed too, so it's better you get 2x32GB 5600MT RAM

 

Well, In my experiments when I (thought) I figured out that it was the PCI-E riser cable on the vertical mount, I was going between using 2x16GB at EXPO 6000mhz, 2x16GB at the standard 3600mhz, 4x16GB at EXPO 6000mhz, 4x16 at the standard 3600mhz.... and possibly some other settings.

Just had the issue happen yet again. Double checked this morning if the iGPU was disabled... and it in fact wasn't, so I disabled it. But still.... same issue. Screen goes black and eventually into power saving mode, fans crank it to 11, Youtube video still playing in the background, and can press the space bar to pause and resume the youtube video like normal. It's not a full system crash or visual glitches like one would expect from core hardware functionality issues. Part of why I'm personally leaning towards it more likely being a PSU-related issue. Which kinda sucks.... brand new PSU.

Just figured for sh*ts and giggles I'd take out two sticks and run 2x16 @ 5200mhz, which is the max technically supported by Ryzen 7000 series. Would be nice if it were just a RAM stability issue caused by using 4 sticks..... but I'd be surprised if RAM stability would cause the GPU to cease being powered, and Windows to still seem to be running just fine in the background.

 

RAM instability can lead to Windows corruption that cause some very weird issues, display corruption is one of them.

BTW can you check gpu-z and see if your 4090 is running at x16 PCIe lanes? if it runs at x8 then there could be something wrong with the VRAM

 

Yes, it's running x16.

And.... yeah, i could get RAM instability causing windows issues and display CORRUPTION. But I don't see how issues with RAM instability could cause the video card itself to just.... stop. Not have any sort of graphical issues/artifacts.... no BSOD..... not even a sign of it still working. The GPU just seems to no longer receive power.. LED's shut off on the card itself, but everything else seems to be still functioning as if nothing happened.

Gonna re-setup remote desktop. Then the next time it happens, i'm gonna see if I can still see what SHOULD be on screen that way..... maybe. I'm really close to just biting the bullet and picking up a new PSU.... maybe a nice Seasonic or something....

 

1. Get your pc 100% stable
2. Try the box 16pin from nvidia

 

TUF GAMING X670E-PLUS WIFI BIOS 1222
Version 1222
9.09 MB
2023/02/24
https://dlcdnets.asus.com/pub/ASUS/...SUS-1222.zip?model=TUF GAMING X670E-PLUS WIFI
------------------------------------------------
NVIDIA Releases 531.29 WHQL
https://www.guru3d.com/files-get/geforce-531-29-whql-driver-download,1.html

 

Oh I totally forgot, have you updated the GPU firmware with this tool?
https://www.techpowerup.com/301693/...nd-that-can-only-be-fixed-with-a-vbios-update

Also you might want to freshly reinstall nvidia Driver too, just to make sure the previous installation is not corrupted

 

you're confusing the crap out of yourself here, an unpowered graphics card doesn't have a fan running at all, the crash to black screen is a crash to protect mode for any number of reasons, one of them being unstable ram.

 

I'm really starting to regret ever having built this damn computer..... or having gone team Ryzen again with it....

To cover the bases from previous replies:
"Get your pc 100% Stable" - It was running flawlessly for the ~2 months since removing the vertical GPU bracket until just the other day.
Motherboard BIOS update - been running it since it came out
NVidia 531.29 - see above
GPU Firmware - already updated
fresh driver install - already did

One reply I skipped.... using the NVidia nightmare fuel adapter. I pulled out my CableMod 12VHPWR cable earlier this afternoon and swapped to the NVidia adapter. I haven't really been using the computer long enough to notice whether or not the same black screen issue will arise..... but ever since switching to the adapter, I can't manage to play more than a round or two of Modern Warfare 2 without it crashing out. But there's a good chance that's just MW2 being MW2.

edit: Ok.... might NOT only be MW2 being MW2. BF2042 just flat out crashes to desktop with no error or anything after a minute or two of playing..... which NEVER crashed before. I wish i had the disposable income to rip the 4090 out of this computer and throw the remainder at the freakin wall. This is like..... I can't even count the number of computers i've built from the ground up or upgraded in their entirety..... and it's the ONLY one i've ever had any issues like this with.

Thinking I might just say heck with it and try putting the CableMod cable back in....

 

https://photos.app.goo.gl/evd4C6sj95EDa2MH9

...... I'm totally lost. Unplugged both hard drives. Removed one NVME drive. Wiped the one still installed in BIOS.... WILL NOT boot from installation media to try to reinstall Windows.

 

select the media from uefi?

 

Selecting it in UEFI, pressing F8 on startup for the boot menu to select it. Going into BIOS and removing the NVME as a bootable option THEN trying to boot from the installation media. Nothing. All I get is this message. Enter to retry basically just makes the screen refresh. Same with F8 for boot options.

Only thing I can think of to even ATTEMPT to do now is put a SATA SSD with a working install of Windows on it, and try to do something that way to try to get Windows back on the NVME drives. But that'll have to wait till tomorrow afternoon.

 

then your media isn't made correctly.

 

Figured I'd work on recreating it last night before I went to bed. Would be odd, considering that'd be the first time I ever had an issue with the Windows installation media creation tool goofed. Gonna test it when I get home later.

Sucks that this crap goes down when there's a double XP event in MW2.

 

I use Rufus to create windows installation usb.

To install Win on NVME drive, choose GPT as partition scheme

 

Well.... I don't think it was necessarily the installation media that was borked. After remaking it last night using the MS tool, I tried booting to it when I got home. Still errored out. At one point, it miraculously made it to the point where I'd choose which version of Windows 11 to install, and THEN there was a BSOD. And tried again, BSOD. And again, BSOD. So I thought..... lets try reflashing the latest motherboard BIOS. Tried using EZUpdate in the BIOS settings to reflash it.... and that's leading me to believe either it just doesn't like trying to flash the same BIOS over itself using EZUpdate, or my CPU might actually be partly to blame because when it rebooted prior to starting the flashing process, it hung with the CPU LED light up on the motherboard and went no further. So.... used the BIOS Flashback option, and that worked like a charm. Got windows installed again, just finished installing drivers, and MW2 is about done installing..... so we'll see how this goes.

Though.... I'm wondering if that whole previous Windows installation was borked from the get-go, because it always seemed like it took an unnecessarily long amount of time for programs to open. Not so much right now, so far.

 

Looks like you are on the right track , hopefully you will it fixed.

I had some RAM problems with AM4 platform too (r5 3600 + x570), sometimes BIOS would just fail to boot and reset itself, not gonna mess with AMD platform again as I haven't had any problem with Intel