X870 will not post after switching itself off

[Jerome]

Hello everyone,

Recently completed a new build less than a week ago and today had some time to finally start using it properly. It's been running on/off all week and no issues until now.

Specs:
CPU: AMD Ryzen 7 9800X3D
Motherboard: ASRock X870 Pro RS WiFi AMD Socket AM5 Motherboard
RAM: Corsair Vengeance 64GB (2x32GB) 6000Mhz DDR5 CL30 RAM

Last night I changed the RAM from 2x32GB Corsair Vengeance 6000Mhz CL40 sticks over to CL30 sticks (white to black) and it worked fine overnight and this morning. I had also enabled EXPO in the BIOS settings when setting them in originally.

Went to take a shower this morning and came back to my PC completely off - it turned itself off (not via any sleep settings).

Tried to turn it back on and solid red/orange lights on the motherboard which indicate CPU/RAM based on the manual. It won't post.

I immediately considered faulty RAM so I've swapped out to the original sticks to no avail. Right now I'm using one of the original sticks in the A2 slot (the recommended slot). I've waited for any RAM training for ~10 mins but no change. I've reseated/repasted the CPU. I've updated the BIOS via flashing it via USB to the latest beta version (I had the latest regular version before). I tried to short the CMOS pins with a screwdriver but feel maybe that's not done anything.. I'll have to remove the GPU to take out the CMOS battery.

I've gone ahead and ordered some cheap DDR5 RAM to see if that does anything (in case there's issues with the Corsair 6000Mhz which would be odd considering I've tried two separate sticks). I've also ordered a new Asus X870 motherboard (the cheapest one I can find as the one I have now is no longer in stock).

Does anyone have any ideas what this could be? The strangest part is that it had switched itself off. To say I'm frustrated is an understatement. I haven't had any issues (heating, stability, nor restarts) since the original build.

Many thanks for any advice/thoughts.

 

[Jerome]

To be clear, you're just getting a black screen with those lights?

Correct, nothing on the display from either the GPU or integrated. Simply solid red/orange lights on the motherboard which don't change with time.

If you do take the GPU out, you can try with the onboard/integrated graphics too.

Done, unfortunately no change. I've taken out the battery to reset the CMOS (waited 30 mins or so) but no joy.

What happened when you removed the RAM? Is there a cable or two that might have unseated themselves?

Nothing changed as far as I could tell. It worked fine for a few hours with the new RAM in place last night and this morning it worked well too for a couple of hours.

What cooler are you using?

Phantom Spirit 120 SE. One fan is removed (in the pics) due to changing the RAM, however normally both fans are strapped on.

You could take a picture of the build and we can check you haven't missed anything.

I've attached a few here.

All cables are securely attached from what I can tell. The motherboard power cable was quite tight going through the top so I took the top part of the case off to verify they're all OK and no possible damage etc.

I'm unsure what occurred when the power went off as I wasn't in the room although it was only for the computer (my other equipment attached to the same antisurge lead were all fine).

That's what makes me think it's the motherboard or perhaps the PSU, as there was no BSOD/restart, just simply a fully powered off computer. I'd read an anecdotal comment on a different site regarding the board itself that this particular motherboard has an issue with Corsair Vengeance 32GB sticks running at 6000Mhz; not sure if there's any truth to it but that's why I ordered some cheap RAM to test it out.

The one thing I absolutely do not want it to be is the CPU given the fact it's impossible to obtain a replacement.

Images are here: https://imgur.com/a/JZXI7Hx

Thanks for your assistance.

 

[Jerome]

I'm not seeing anything obvious in the photos. Out of interest, are the lights different if you try to post without any memory installed at all?

Still the same with no sticks installed - both red/orange lights illuminated.

Have you checked the cpu base and socket for any damage?

I had a cursory look when reseating it earlier but nothing to note. I didn't look in depth but will do so tomorrow after the new motherboard arrives. I'm hesitant to do it now as I've already used more paste to reseat it earlier.

I was very conscious about ensuring it sat properly as damage to pins is something I was particularly nervous about in the beginning.

Plan is:

1. Test with a basic RAM stick to rule it out completely
2. Switch to the new motherboard (and check the CPU at the same time)
3. If that fails, I'll try the new PSU (which I may stick with anyway due to it having a 12 pin cable for the GPU instead of having to ues 3x PCIe cables which is a bit messy)

All three of those should arrive tomorrow. At this stage I'm actually hoping it's the motherboard.

Failing that, it's time for a new CPU, which is going to be really disappointing indeed. I don't foresee any situation where I'm not going to be significantly out of pocket (having to wait for a 9800X3D replacement but buying a seriously overpriced 7800X3D in the meantime which I'd have to sell for a loss).

 

[Jerome]

Latest update:

- Tried completely different branded RAM with the existing motherboard, no change
- Swapped the motherboard to a different model, no change
- Updated the BIOS via USB flashback on the different model just in case, no change
- Tried all the RAM I have on the new motherboard to see if it makes a difference, no change

The Asus X870-P Wifi doesn't even have indicators lights which is ridiculous, considering it's more expensive. Still the black screen.

It's now looking to be that the 9800X3D is the culprit. Time to shell out for an inferior processor at a higher price as there's no chance I'll be getting a replacement anytime soon.

Really aggravated.

Photos of the CPU and previous motherboard pins, can't see any indication of damage: https://imgur.com/a/6fCGxPw

 

[Jerome]

Does sound like the CPU to me. That's always a possibility when you overclock the RAM, but it's extremely rare. Perhaps this is a weakness with this CPU? I don't know, I am not familiar with it's architecture.

Anyway, if you look at the tech specs of the RAM, the CL30 is higher voltage than the CL40. My guess is the higher voltage blew the memory controller.

I enabled EXPO when I installed the CL30 RAM sticks. Just noticed regarding the voltage, I believe it's 1.40V instead of 1.35V. It worked fine overnight and into the next day when I tested a few games but then suddenly went.

I feel very unlucky indeed if that's the case.

Didn't you get the memo? That's a high end feature these days!

Pins and base look fine to me, so I'd agree with the CPU being dead. 7600x is a reasonable sub until you get your replacement back.

If it ends up being the CPU I'll keep the original ASRock X870 Pro RS board, which was almost 20% cheaper and has more features too.

Thanks for the advice on the 7600X. I think I'll go ahead with that. In my haste I was considering purchasing a scalped 9800X3D out of pure frustration but now I feel it would be unwise.

 

[Jerome]

Would it be faster to RMA the CPU direct to AMD?

You're probably better off going direct to AMD if you've had it longer than 28 days, especially given the 9800x3d shortage.

I received it just under 2 weeks ago so I'm not sure the best course of action.

Would you think it's still preferable to go direct to AMD instead of the retailer?

 

[Jerome]

I submitted an RMA request direct to AMD, let's see how it goes.

In the meantime a new PSU arrives today so I'll give that a go. Perhaps the CPU power module on my existing supply is faulty.. doubtful but we will see. If it's not the case then it has to be the CPU (or I have a dead motherboard and the second motherboard was DoA, which again is doubtful).

7600X arrives tomorrow to finally answer that once and for all.

In the meantime I've reconnected the laptop I used previously as my workstation (i7-7700HQ, 16GB RAM, 1050Ti). It'll do but it's certainly not as good as the new one!

 

[Jerome]

Current: Corsair RM1000x

New: Corsair RM1000e

The current one was used less than a week. The GPU low voltage light appears correctly when the power button is on and then disappears when the computer is switched on fully indicating there's sufficient supply through the PSU itself, but perhaps specifically a fault with the CPU module or wires leading to it. Wishful thinking I suppose.

If it's not the PSU then it'll be going back as I understand the RM1000x is superior in quality.

 

[Jerome]

I believe it's 3x PCIe cables connected to the splitter provided in the GPU box. It's an Asus ProArt GeForce 4080 Super.

No issues with the GPU. For the purposes of testing I've also performed all diagnostics with both the GPU in and out (using the integrated HDMI port otherwise).

 

[Jerome]

Unfortunately after switching the PSU (I attached the Motherboard and CPU cables only to test) no change.

I've since removed the CPU and sent photos to AMD as per their request. I'll test the 7600X when it arrives tomorrow.. let's hope..

I feel AMD will be able to provide a replacement CPU quicker than a retailer. We'll put that theory to the test.

 

[Jerome]

Over on a different forum it appears that others are also having the same issues with 9800X3D's dying after a few days of usage. One common theme is that it occurs after EXPO is enabled (which is what happened in my case too, albeit the day after).

I wonder if there's a bad batch or issue with quality control to allow this to happen.

 

[Jerome]

I'm sure everyone is sat here waiting for an update to this nailbiting saga.

After receiving a 7600X, I can confirm the system now posts. Therefore it was a faulty CPU (which I'm led to believe is exceptionally rare in comparison to other components).

AMD have accepted my RMA and it's going on a trip to the Czech Republic it seems.

Now time to rebuild the PC with my original (ASRock) motherboard and get all this excess hardware sent back (where possible).

 

[Jerome]

Is true, we don't see it on here very often, but glad you got to the bottom of it!

Thanks! I'm certainly relieved. Testing performance at 4K it's not too far off the 9800X3D so it'll do in the meantime. I'm not into any competitive games and have a 60Hz 4K monitor so as long it's above 60FPS it suffices.

Did AMD give any ETA on how long the RMA would take?

Glad you narrowed it down

Sadly no. The guy I was in touch with in the ticketing system approved the request and DHL will collect it on Friday (as I'm working until then). Hopefully not too long but I know it's dependant on 9800X3D stock availability.

Maybe, to be on the ultra safe side, get RAM that doesn't need 1.4v to run at 6000/C30 EXPO?

Any recommendations?

I have some 6000/C40 1.35v sticks that I bought originally which I could use. I have the 6000/C30 1.40v sticks in now. I haven't changed any settings besides going from 4800Mhz to 6000Mhz in the BIOS. I'm not overly interested in overclocking or undervolting so whichever RAM works best out of the box honestly.

It's Corsair CMK64GX5M2B6000Z30 (2x 32GB). It's not on the motherboards QVL list admittedly. (Mobo: ASRock X870 Pro RS WiFi)

--

Just to note I've put the case back together with the original motherboard, etc. All works well. So it was certainly only the CPU that failed, for whatever reason.

I appreciate everyone chipping in with advice, it helped keep me sane.