I have a question for the community that arose from my last thread. That thread was more specific about the hardware I was working with, this question is more general.
The gist of that thread is that I seem to have received two faulty motherboards in a row with the exact same failure scenario. Both have working IPMI but neither will power on under any circumstances. When I say “won’t power on” I mean absolutely nothing happens when pressing the physical power button or using the power on feature from the IPMI. No fans, no LEDs, no error codes, nothing. I tested with multiple known working power supplies and multiple cases. There are existing reports out there of others having the exact same issue with this motherboard so my case isn’t unprecedented, but the chances of someone getting two faulty motherboards in a row with the EXACT same failure are incredibly small. I have built plenty of computers in my life, but this is a clusterf*** of epic proportions.
The big issue in my scenario is that I don’t have a known working processor to test with. I have a used AMD EPYC processor that I got a good deal on on eBay, so I am unable to say definitively if it is or is not working. This makes it difficult confirming if the motherboard is indeed faulty or not.
In an attempt to test if the motherboard is faulty, I tried powering up the system with no CPU or memory installed at all. AFAIK, the motherboard should still power up but immediately fail to post because no CPU is detected. I got the exact same result though, no power whatsoever from the physical power button or toggling the power in the IPMI. This leads me to believe that the motherboard is faulty, but I can’t find any results out there on the internet if my test should work with my specific hardware or not.
So I come to you with a question, levelone community. Can a CPU be so faulty that it completely prevents a CPU from powering on? I would assume that if a motherboard has a faulty CPU installed that it would at least power on but immediately fail to post and either display an error code on a debug screen if it has one or have a beep code from the motherboard’s speaker. That is just an assumption though, and I would greatly appreciate any input I can get from others.
I am wondering the same thing. However I don’t think I am able to even get far enough into the boot process for this to be an issue. My understanding of how this feature works in EPYC processors comes from this reddit thread:
Certain AMD chips have a PSB, a complete seperate ARM cpu and OS, that regulates the CPU. If the PSB was instructed to lock the CPU to a code signing key that signs the bios, when the CPU boots up, it will look for that key. If it doesnt see the key, the CPU will HALT and the PSB will fire off an error message. The key is stored in efuses, and (in theory) is unremovable/unbypassable.
In order for the system to verify if the processor is locked or not, it needs to receive power. My system isn’t receiving any power at all, so there is no way it could be even getting as far as to check if the processor is locked.
It’s all good, it happens. I agree with Marc’s point in all honesty, it’s bullshit that hardware makers even offer it is an option to OEMs. If there was a CPU manufacturer that offered the same feature set at a relatively similar price point without the ethical dilemma, I would happy give them my business. Let’s face it though, that’s nothing but a pipe dream.
It’s an odd failure scenario isn’t it? One would assume that at the VERY least the system should power on but fail to post. How could the circuitry that provides the vendor locking functionality even receive the power it needs to prove to the system that the CPU is vendor locked if the system won’t power on at all? I mean I guess there could be a tiny little battery or capacitor in there, but I haven’t heard anything about that in my research and I would assume AMD wouldn’t want to have to deal with squeezing that into the die.
Regardless of the vendor locking though, I would still assume that a motherboard with no CPU or memory in it should still power up and attempt to post if someone tries to boot it up, it would just fail to post almost immediately. That isn’t even happening in my situation. It’s really hard to believe that I could get two totally dead motherboards in a row, but all evidence is starting to look like it.