Question I need some help troubleshooting a PC that won't boot ?

Friisern

Distinguished
Mar 10, 2013
12
0
18,510
Hi there!

I recently bought a new GPU and PSU and everything seemed to go swimmingly during my first few hours of playing cyberpunk. Then I tried to hit the overclock preset in the AMD Adrenalin software, which instantly resulted in a black screen and the PC crashing. It took a few tries to get the PC to boot, as only the lights and fans of all the components inside the PC would turn on, but not the the monitor, keyboard or mouse. After I got the PC to boot, I started up cyberpunk again, and started to fiddle around with settings in both the game and Adrenalin. A new crash happened, and again it took a few tries to get the PC to boot, same issue as mentioned earlier.

I started up the game again and played a bit without fiddling with any settings, and had no issues.
Then I closed the game and once I reached Windows, the screen turned black after a few seconds and i crashed again. I have since tried back and forth with cyberpunk and less demanding games and the pattern seemed to be that if I pressed "exit to Windows" while playing, the PC would crash, but if I exited to main menu first, then exit to Windows from there, I was safe and no crash would happen.

Now after trying this to make sure that was the problem, the PC will not boot back up. All fans and lights are on inside the PC, no lights to monitor, mouse or keyboard. I have tried removing the GPU and using the integrated graphics, tested with only 1 ram stick, multiple slots with both and taking the motherboard battery out and putting it back in, all to no avail. I don't really have many components to troubleshoot with at this moment, but any help pointing me in the right direction is much appreciated.

System specs
GPU: ASRock Radeon RX 7900 XTX Phantom Gaming OC
PSU: Corsair RMe Series RM850e
CPU: AMD Ryzen 7 5700G
Motherboard: ASUS ROG Strix B450-F GAMING II
RAM: Corsair Vengeance LPX DDR4 3200MHz 16GB (2x8GB)
SSD: WD Black SN850 500GB M.2 NVMe SSD


This is the last i saw before my PC took a vacation
 
Was the video taken with 7900 XTX in the system or with iGPU?

Since what there is, is high level or artifacting. Could even be BIOS corruption.

PSU: Corsair RMe Series RM850e
Mediocre quality PSU, most likely the culprit of all your issues.
With that high-end GPU in the build, bare minimum i'd expect to see PSU wise would be either: Seasonic PRIME, Corsair AXi or Super Flower Leadex Titanium.
pcpp: https://pcpartpicker.com/products/compare/vkYXsY,2BcRsY,WrcgXL/

So, my 1st order of business would be new PSU, at least 1kW unit (1000W). But that doesn't fix the artifacts/corruption.

Depending which GPU was in the build, you're looking towards new dGPU or iGPU. Since latter is part of the CPU, this means new CPU.
 
That video was taken with 7900 xtx in. I kind of emptied my cash reserve going for those parts, so those psu's are out of my price range at the moment, although I have room to stretch it a bit further in terms money. Is it really necessary to go that high in wattage? The peak draw of the card was around 480w, that would leave more than enough power for the other components, or am I mistaken?
 
Is it really necessary to go that high in wattage? The peak draw of the card was around 480w, that would leave more than enough power for the other components, or am I mistaken?
Sure, GPU is 355W and it pulls 480W under max load, but have you also factored in transient power spikes? 🤔 Since if you include that as well, where GPU can pull ~2 times of it's rated TDP, you're looking towards ~700W for transient power spikes.

With PCs, having more wattage doesn't hurt. But coming short - will.

That video was taken with 7900 xtx in.
Hmmm.... it is difficult to diagnose it, since while there was serious artifacting/corruption when dGPU was in, now the entire build doesn't POST.

I'd breadboard the build. Since you'll have to take it apart regardless.

So, pull MoBo from PC case and put it on any cardboard box (e.g MoBo retail box does fine).
Remove: GPU and all storage drives, including OS drive.
Also remove all RAM, except 1 DIMM.
Hook monitor to MoBo and look if you can get an image on power-on.

If not, then one of the following or all of them, are dead: CPU, MoBo, RAM.

Since you have 2x sticks, you can try with 2nd stick as well. But always use only 1 stick at a time.

If still no dice, it comes down to either CPU or MoBo. And here, it's 50:50.
Usually, when PSU acts up, MoBo is 1st to go.

You could also consider your 7900 XTX dead, since good working GPU doesn't produce artifacts, let alone the severe ones you saw.

Though, for 2nd opinion, haul your PC to PC repair shop and pay them for diagnostics. They can find out if CPU or MoBo is toast (or both).

So, for bare minimum, you're looking towards:
* new, great quality, 1kW PSU (either any of the three i already linked, or i can give you a bit cheaper, but still great options to choose from)
* new GPU (you should be able to RMA your GPU, IF AsRock covers it. They may not, since if they find out that GPU was killed by PSU, it doesn't fall under GPU warranty)
* new MoBo or CPU (which one exactly, better to do PC diagnostic in repair shop) Chances that both RAM sticks died on you, are rare.

But i'd get new PSU regardless. Since when PSU already killed hardware, it will kill replacement components as well. And PSU warranty does not cover other components it killed. Hence why never cheap out on PSU. And also why never buy used PSU either.
 
  • Like
Reactions: PEnns
I plugged in the old GPU, and the PC booted up without any issues. This leaves me to believe that the GPU is the problem, and possibly the integrated GPU in the CPU. Now I'm uncertain of the right cause of action, but after talking with the shop, i could just hand in the old PSU and get another model if I wanted to. This seems to me to be a possible first step. But if my PSU is so mediocre, why is it listed under A-tier at https://cultists.network/140/psu-tier-list/ . Am i reading that list wrong, or are you suggesting that it is mediocre due to low wattage?
 
I plugged in the old GPU, and the PC booted up without any issues. This leaves me to believe that the GPU is the problem, and possibly the integrated GPU in the CPU. Now I'm uncertain of the right cause of action, but after talking with the shop, i could just hand in the old PSU and get another model if I wanted to. This seems to me to be a possible first step. But if my PSU is so mediocre, why is it listed under A-tier at https://cultists.network/140/psu-tier-list/ . Am i reading that list wrong, or are you suggesting that it is mediocre due to low wattage?
It's possible your old GPU draws less current than your new one, thus less stress on the PSU.
 
It's possible your old GPU draws less current than your new one, thus less stress on the PSU.
Yea, at this stage I'm not really sure what to do. Cash out and upgrade the PSU and test it with the 7900 xtx? RMA the card and get an ASUS TUF card instead as that is similarly priced (the other xtx cards are 200-500$ extra, no chance i'm doing that). RMA both or just the card and test it with the RM850e PSU? If i just RMA the GPU, i might end up with the same problem again if the PSU is faulty. I also don't really want to give ASUS any more money, but that is the only XTX option if i don't want to sit for weeks waiting for the repair technician diagnose the Asrock GPU.
 
But if my PSU is so mediocre, why is it listed under A-tier at https://cultists.network/140/psu-tier-list/ . Am i reading that list wrong, or are you suggesting that it is mediocre due to low wattage?
Cultist.network is biased and favors Corsair PSUs above all else, hence why Corsair PSUs are rated far higher than they actually are.
E.g RMe has edit note of [14], which means:
[14] Potential ATX 3.0 units either not tested for compatibility (no Aris review), not received the certification (no entry in Intel’s DB), having problems with passing ATX 3.0 tests (failed Aris review), or without proper reviews in the first place to put them in the normal priority subtier.
Yet, RMe is put into normal priority subtier, while on the same time, Seasonic Vertex, with edit note of [14] as well, still remains in the speculative category.

Another prime example would be their PSU buying guide,
link: https://cultists.network/2606/psu-buying-guide-september-2021/

Where almost in every metric, Corsair PSU is #1. Except 650W and 550W unit, but even in those two categories, Corsair PSU is still listed.

But if you were to look at reputable (unbiased) PSU list, like one in Hardwarebusters,
link: https://hwbusters.com/best_picks/best-atxv3-pcie5-ready-psus-picks-hardware-busters/2/

Then you can see that on every category, there are completely different units that are best. Moreover, every category is subdivided based on Overall Performance, Best Average Efficiency, Least Noise, Best Transient Noise, Best Ripple and Best Voltage Regulation. And top 3 is listed in each subcategory. So, you have a wide selection of PSUs to choose from, rather than the cultists.network "Corsair is #1" mentality.

Yea, at this stage I'm not really sure what to do.
Your build failing to POST is biggest concern, since that indicates issue with CPU, MoBo and/or RAM. Or with PSU, which powers them all.

Namely this part:
I have tried removing the GPU and using the integrated graphics, tested with only 1 ram stick, multiple slots with both and taking the motherboard battery out and putting it back in, all to no avail.
Without GPU, your PC should work fine, since CPU has iGPU in it.

I'd still look for new PSU, since while the build booted with 2nd GPU, with less load on PSU, doesn't make PSU sound.

Steps would be;
1. New PSU.
2. Breadboarding with new PSU, to make sure CPU-MoBo-RAM combo works.
3. Breadboarding with new PSU and old GPU, to see if build boots. (Currently does boot with RMe.)
4. Breadboarding with new PSU and 7900 XTX GPU, to see if artifacts/corruption remains.

If step #2 fails, but step #3 and #4 make the PC boot, iGPU inside the CPU could be damaged. This would mean CPU replacement, despite the build working with dedicated GPU. Since it won't be long until CPU starts to throw out BSoD errors.
Step #4, with new PSU, would tell if 7900 XTX is gone for good.

Other, good/great quality 1kW PSUs, but cheaper options, would be:
Seasonic, pcpp: https://pcpartpicker.com/products/compare/jJMTwP,RqYmP6,m72WGX,W2zXsY/
Super Flower, pcpp: https://pcpartpicker.com/products/compare/rt848d,DMbTwP,cmrRsY,yxzhP6/
Corsair, pcpp: https://pcpartpicker.com/products/compare/jRvdnQ,LTmNnQ/
 
I just sent the PSU back, and I'm getting the Corsair RM1000x 1000W tomorrow. Something that i have failed to notice/mention that has been going on this whole time after the shutdowns is that the PC takes longer to post(?). After i press the power button, everything inside starts up normally, but the keyboard, mouse and monitor takes ca. 7 seconds extra from what they used to to receive image/light. During this time, the top right light on the MB next to the CPU starts out as orange, then switches to red, then goes to white and then the monitor, keyboard and mouse turns on. I don't know if this is an indicator of anything, just thought it was worth to mention. As of now, i have temporarily plugged in the old PSU and going to try to play some games to see if anything else is different. Thanks for all the help so far, and hopefully I can get some more answers tomorrow when breadboarding
 
From my testing, everything posts when the PC is laying down. Once I erect (lol) the PC, it does not post. I tried to put slight pressure on the GPU when it was laying down to simulate the massive weight of the GPU, and then the artifacts appear on the monitor. Either the MoBo PCI slot can't handle the weight, or the card it self can't handle it. And since no artifacts appear when I flex the old GPU, it seems as though this GPU is too big for it's own good. The GPU did not come with an extra support slot like the other 7900 xtx models does, so there is a high chance of the some internal stress in the card causing it to malfunction. I have also tried to tighten the screws on the GPU, but they seem to have been tighten properly.

Update: this GPU is going back via RMA. Tried to play some overwatch 2 at max settings (which should be a breeze for this GPU), and then closed the game and the PC instantly crashed and i had problems starting it again. There seems to be multiple issues, not this the sagging, so a new GPU is in order. It's as if it can't handle the drop from high wattage use to low, and it just calls it a night.
Thanks for all the help to everyone who replied!
 
Last edited:
Either the MoBo PCI slot can't handle the weight, or the card it self can't handle it. And since no artifacts appear when I flex the old GPU, it seems as though this GPU is too big for it's own good.
Every GPU that is 3 slot or more wide, should have GPU support under it. Else-ways, the GPU sag can cause all sorts of issues. Artifacting and even breaking the PCB of GPU near PCI-E slot or bust open the MoBo PCI-E slot.

It's as if it can't handle the drop from high wattage use to low, and it just calls it a night.
Did you test that with Corsair RMx 1kW?

There seems to be multiple issues
Yes, and hence why it is difficult to diagnose. Only option is component replacement, to figure out what causes it.
 
Did you test that with Corsair RMx 1kW?
Yes, still crashed when exiting games. Although something dawned on my last night. I have been using 2 PCIe cables, one single and one daisy-chained. Did some research and there seems to be a lot of conflicting info. Some report that they have been using it like that for a year without any issues, while others report it causing instability, and even melted cables. That might be a possible cause to the crashes, as, from my understanding, 2 cables can only deliver 600w. I'm still going to RMA the card because of the fact that the PC isn't posting anymore when the card is horizontal, but going forward, I will dedicate 3 single PCIe cables to be on the safe side.
 
as, from my understanding, 2 cables can only deliver 600w.
No.

One 8-pin PCI-E cable is rated for 150W.
One 6-pin PCI-E cable is rated for 75W.

So, when you have two dedicated cables, with 8-pin PCI-E at the other end, you're looking at 300W.
Three dedicated 8-pin PCI-E cables would bring it to 450W and 4 would bring it to 600W.

Hence why there are 4x 8-pin PCI-E connectors on 12VHPWR adapter for RTX 4090, to deliver 600W.

Nvidia_12VHPWR_adapter.jpg