Random crashes with X670E Hero

[Foz]

Before i go into the nuts and bolts let me spec out what i have.

X670E Hero - BIOS 1303
7950X3D
ZOTAC 4090 AMP AIRO
G.SKILL Z5 Neo EXPO 32GB 6000MHz F5-6000J3038F16GX2-TZ5NR
CPU and GPU custom watercooled (EKWB parts)

So the other night while attempting to play Jedi Survivor the onboard Bluetooth disappeared (not in device manager etc), odd I thought but not a deal breaker and not worth the 2 months i was quoted i would be without the board, anyway my fault finding was as follows:-

Applied my clonezilla image of my machine in case it was a rogue update, Image is from 2 month ago when the system was built (clean image OS, drivers and AV) but the device was still missing.

Clean build of Windows 11 Pro from freshly downloaded ISO, no device.

Reached out to ASUS who said run the diagnostics in MyASUS app but there are none.

I had some new fans which i was waiting on installing on my Rads and thought no better time than now, they all connect to a fan controller so all i did was unplug 6 and plug in 6 to the same controller (all noctuas) and tidy the cabling, now the bluetooth is showing again but randomly the PC now just shuts down, no BSOD logs but 6008 IDs in event viewer, to power back on I need to either turn off the PSU and turn back on (Seasonic TX1300) or use the onboard ReTry button.

I'm at a loss and kinda out of ideas so any help is welcomed.

[Țăranu]

I am doing now test with RAM set on Asus Tweaked and AI Overclock + FCLK set on 2033 or what ever is the first step abova that sweetspot of 2000. Try that.

[zack79]

I am not convinced it's a psu problem. My evga 1300 is running another build with no issues. 

[Dabz]

We know that the issue extends beyond just the PSU since users have reported successfully resolving it by swapping either the PSU or the motherboard. As I mentioned in my previous post, the problem likely stems from compatibility issues with my specific motherboard, CPU, GPU, and BIOS combination. Additionally, several users have mentioned that the issue started after a BIOS update.

I have personally conducted tests involving LLC adjustments to address vdroop and have consistently found positive results.

However, the crux of the matter is not the individual workarounds. The key focus should be on ASUS and Seasonic collaborating to identify the root cause and develop preventive measures. This way, they can provide a comprehensive solution for everyone affected. It's clearly not reasonable to expect users to purchase multiple 500€ PSU or motherboards as a fix.

[ItsTrigger]

In terms of bios updates, what was the latest AGESA version that supported the X3D CPUs that was mostly working fine with Seasonic PSUs?

I got worried by the CPU burnout scare and updated my X670E-E to 1416 as soon as I received it. Tempted to try and older bios this time.

If I go down that route, do I just need to put a cap on the SOC voltage or is there anything else I need to be aware of?

[zack79]

That's the same question I started off with when I opened my asus ticket. I think the oldest that supports is like 0944 or something like that. Good luck finding it. Asus took it off their page when all the soc stuff kicked off.

[Flakey]

Some update from me: There were no issues since the last crash reported here until today. Today, the display froze (probably driver issue, but nothing logged by Windows). Afterwards, I updated my GPU driver and then I had two consecutive crashes into the state that requires power cycling the PSU. One was soon after the reboot just on the desktop. The second one was on the Windows login screen before I even entered the PIN (I guess there can be no less load than this...).

I contacted Seasonic yesterday and they are aware of the issues and this thread. Their support responds quickly and they recommended I open an RMA request, which I did. They are currently checking stock and are getting back to me once they know what they can send me as a replacement.

[ItsTrigger]

Bit delayed, got my replacement board back today, just assembled the system with the replacement PSU, will see how that goes.

Out of interest, of the Seasonic PSUs and such involved here, what manufacturing dates are on the problematic units? My original was from Aug '22, the replacement is dated March '23.

EDIT: Well that lasted all of about 2 hours before shutting down in the exact same way.

[McDaniel]

Psu seasonic 1300 from August 22

It's a bummer that your new one shows the same behavior 😔🤦🏻‍♂️🤷🏻‍♂️

[Flakey]

Mine is also from August 22. Assuming the serial number starts with YYMM.

[Foz]

Thought i'd just chime in as th OP, support has been laughable at this point. I returned my PSU to scan.co.uk for a replacement and then received a message from ASUS in line with this thread saying that I should RMA my PSU, i advised i had done this and was kind of ignored and the same information was just repeated or copy and pasted then no reply when i asked if Seasonic was the cause then.

The funny thing is the company i work for is going down the AI/ML route like most and wanted to buy a few test units for dev (chonky laptops as a POC before spending serious cash on nvidia H or L platform servers), anyway while having lunch with some colleagues and talking about my experiance with this motherboard etc the CTO who was wandering by overhears the conversation so sits down with us to explain he was looking at this dev kit and he was down to MSI, ASUS and Dell XPS, he asked me to link him this before heading off to consider options (If you're reading this Garry "Hi" it was great to talk to you).

Personally i think my journey with ASUS will end here and i'll be looking at other option in future builds and with what i recommend to friends, family and in passing, who knows maybe with the next AGESA release other fixes get bundled in but my trust has taken a hit here.