CPU: 3700X
Motherboard: Aorus B550 Elite
RAM: 8GBx4 Corsair Vengence LPX 3200
GPU: PowerColor 5700XT
PSU: Cooler Master MWE 1050 V2
Built in 2020.
Since last month, my PC started having random reboots and giving āMachine Check Exceptionā error, similar to these:
https://old.reddit.com/r/AMDHelp/comments/190mkn0/5950x_whea_error_18_machine_check_exception/
https://old.reddit.com/r/AMDHelp/comments/qia2e7/whea_18_critical_error_computer_goes_black_restart/
https://old.reddit.com/r/buildapc/comments/150m14n/pc_randomly_restarts_whealogger_id_18/
And now from the last 3 days the system doesnāt boot. When I power on the computer, all fans start spinning but keyboard and mouse LEDs donāt light up. Pressing CTRL+ALT+DEL doesnāt reboot system neither does pressing the power button for few seconds.
I suspect that motherboard has gone kaput and isnāt completing or even starting the boot process, which is why keyboard and mouse arenāt getting any signal or power from motherboard or why restart or power down functionality is working.
Before the system stopped booting, I was trying to solve the machine check exception error by updating BIOS, updating chipset drivers, changing BIOS settings etc. But now Iām thinking none of it couldāve helped because the board itself was deteriorating.
Also during that time, I would randomly get display glitches (pic below) which could only be solved by restarting the machine so I was suspecting it might be GPU that was causing the problems.
Sometimes it would show chessboard like pattern. I guess this was also because of some issue with mobo-GPU connection?
Anyway before changing the board is there anything else I can try? Changing it is a pain so Iām trying to avoid that. š
Remove the video card, use inboard video, and use a single stick of ram. If itās still wonky try a different power supply.
A second idea if you cannot source another gpu; Change PCIe slots. While I believe your video card is going bad or at a minimum: overheating: consider moving it to the other x16 slot on your mobo as it IS technically possible the slot is having issues. It also doesnāt hurt to spray your case out with canned air to get the heat capturing dust out.
I tried without GPU and the problem persists. Iāll probably take the whole cabinet to my friendās place when heās available and try replacing components one by one.
for whatever itās worth, my powercolor red devil 5700 died in a similar way to this several years ago. Got to a point where it failed to output any display signal at all.
Powercolor had an insane return rate for their NV10 GPUs, at least in Europe.
Are you getting WHEA logs here? Do they implicate a specific component?
for whatever itās worth, my powercolor red devil 5700 died in a similar way to this several years ago. Got to a point where it failed to output any display signal at all.
Thatās what I thought too but as Iāve mentioned, keyboard and mouse are not getting any power from motherboard.
If it was GPU issue, those would have active lights with only monitor not getting any signal from GPU.
Are you getting WHEA logs here? Do they implicate a specific component?
When the system still used to turn on properly, Iād get these in Windowsā event viewer:
Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: This kept changing
Provider Name: Microsoft-Windows-WHEA-Logger
Event ID: 18
When I started looking up information about this error I found out it could be caused by literally anything. Faulty CPU, board, memory, PSU. Sometimes setting CPU voltage around 1.30-1.35 helped. In one case the guy replaced his custom power cables with default ones and that solved the issue.
Oh, cache hierarchy errors were decently common with Matisse too. I think this one may be your CPU my colleague hit this the other day with their olā 3700X. I donāt suppose you could RMA that?
I agree with Brkdncr, this sounds like a video card issue.
Iāve had this problem multiple times before. The combination of display glitches and the fans spinning but no numlock or keyboard functionality simply points to the video card first.
In short, during POST of the BIOS it attempts an init to the display, fails and then stops attempting the boot sequence. It, the video card, is just as important during init as the motherboard registers, RAM and CPU all starting.
So, start with video, see if it works without the 5700XT and using the onboard or some other cheap pcie card. If that fails too, then itās most likely the mobo as assumed. This just doesnāt sound like a mobo issue though.
Thereās no onboard video for this setup* but Iāll try to boot without GPU.
*Only Ryzen CPU with G suffix have integrated graphics. Plus, my CPU is Zen 2 which didnāt have integrated graphics at all.
In short, during POST of the BIOS it attempts an init to the display, fails and then stops attempting the boot sequence. It, the video card, is just as important during init as the motherboard registers, RAM and CPU all starting.
I wasnāt aware of this. I assumed display check would be last considering there can be systems without display.
Let me check if I get NUM LOCK and CAPS LOCK LEDs after removing GPU.
While I call it init, many will assume I am referring to the boot init, but I am actually referring to the bios initialization (init).
That said, most BIOS inits go in this order:
-
Power Detection check. If enough power, proceed.
-
CPU program link, CPU calls the BIOS to basically wake up and run the bios program.
-
Ram Detection check. If RAM is present, the BIOS will use about 64k to load from ROM to RAM (called the bios reserve area) that then does the next steps.
-
Hardware Detection check. Identifiers of the hardware are detected, enumerated, configured and initialized.
-
Boot Sequence is initialized whereby the BIOS does a handoff to the bootloader.
Itās during the hardware detection phase when the display is initialized by the gpu and you often see it displaying the bios version, then counting RAM. If the gpu is working BUT the display out isnāt, itāll actually continue to boot 100% of the time (it doesnāt care). If the gpu hardware itself doesnāt respond correctly to the BIOS request however, it sends the hardware detection of the BIOS into a loop or shuts the system down, never getting to the final step: boot sequence.
Depending on the bios type, it may or may not show numlock. Iāve also seen it act differently on UEFI enabled systems than when itās set to classic bios. So, it just depends.
Regardless, see if you can source another pcie gpu for testing this. It only takes a minute and tbh, it doesnāt hurt to have a cheap used pcie vid card in your pc tools for such things.
Good luck!
You canāt see in the pic but the white glitch columns arenāt continuous. Those are white lines interspersed with whatās on display.
I had a 3700x CPU that would lock the system up randomly. I was on Linux with an Nvidia GPU so the symptoms donāt match up. I think itās worth borrowing a friendās CPU, or buying a used one locally, to test. If you buy a used CPU locally for $50, you should be able to sell it for $50 after your test.