So I have the following hardware that have been running proxmox for like a year now, well until yesterday
Hw:
MINISFORUM AR900i Motherboard, Intel Core i9-13900HX,
24C/32T, Up to 5.4GHz, with Intel HM770 Chipset,PCIe 5.0 x16 Slot,
4×M.2 PCIe4.0,DDR5,HDMI/DP/USB-C, RJ45 2.5G, 3X USB 3.2 Gen 2,WIFI6E&BT5.2
I have it populated with 96GB Crucial Mem Kit DDR5-5600 PC5-44800 CL-46 SO-DIMM
I noticed yesterday when I was outside and trying to remote into it, nothing was responding … Upon checking this morning by connecting a console HDMI cable, I saw it was trying to boot proxmox and it would fail immediately. Tried booting other OS via USB (ventoy)… RHEL installer, Fedora, Ubuntu, even Windows… All would result in the same failing to boot behavior… I then tried moving around my memory sticks and trying to boot with other RAM… same results…
Running a MemTest, no errors but it wouldn’t pass either, on Pass2 of the memtest it would reboot.
I tried different PSU, disconnecting all PCIE, NVMEs…etc, same result.
So I ended up going into the BIOS, changing power mode to “Max battery lol” and locking p-cores to only 4 out of 8 and also locking e-cores to use only 4… then the machine would boot normally into any of the OSs… including my Proxmox.
I think I have a failing CPU, what do you guys think?
There was a video by Jayz2 with “I had problems → I downclocked to make sure the CPU died → Yep, it died”.
Although I don't really trust this fellow...
Although I don’t really trust this fellow. This CPU was “a friend who bought a maxed out system with 14900KS didn’t know that intel had problems” (yes, it is possible, but feels like it was “for a video”), after which, without any explanation “hey guys, my intel cpu also died, I will switch to popular amd now…”.
Feels like that kid, who broke his phone to get a new one.
Yes I seen it but man I really thought given that the i9 13900HX is a mobile chip and I am running it with low utilization(1-5%) … I’ll be okay …
I hope Minisforum support would come to the rescue otherwise this gonna suck
But a mobile chip? Yeah, that would be quite strange. Unless you had something running, that peaked it to 100% while you weren’t looking.
I have a corp laptop with intel’s 15 gen (I don’t know how to pronounce it even…). With an i7 mobile chip. And can say that it was running quite hot (to a boiling point of my frustration, I would even say).
well yeah, It was running Proxmox with a lot of VMs… and been running 24/7… even thought the total utilization is low… I bet some workloads would push the clock speed.
Not really, I mean the Minisforums bios is kinda limited anyways and I just didnt bother in the past poking around with it. I kept it running stock/default BIOS options.
It may be a mobile CPU, but it stock boosts to 5.4 GHz. Its been a while since all the videos about this Intel fiasco, but wasnt the max safe clock speed that doesnt boost into damaging voltage levels found to be 5.3GHz?
From what I remember, both Wendell and GN found the boost behavior stayed at proper voltage when set to 5.3 and below, while hitting unsafe levels at anything 5.4 and above.
Update: Minisforum support has sent the bd790ix3d replacement board for an additional fee(120$)+returning the old board.
After receiving the replacement board. I plugged the same components (memory, PSU…etc) and everything sprung back to life. Confirming the degraded i9 CPU being the cause.
The experience with minisforum support took around 2 weeks with a lot of confirmations and testing, including filming a whole video and sending it to them. Support replies China timezones (12hr different than US)… and to be honest not the best support experience in terms of response speeds. I’m glad and thankful they did replace my board though.
Did they let you keep the dead one? I have a modest collection of failed HX cpus I really want to send to the FA lab people but root-causing has been… difficult.
As for the failure, I had to prove it to them by making a youtube video showing how it has failed and how disabling cores makes it boot again, along with memtest and everything, If you wish, I can share it with you privately but I do not wish to make it public. If that helps in your investigation of the HX chips.
