0

I just upgraded my server from a ryzen 5 1600X to a ryzen 5 4600G. During the boot process, the server turns on correctly and even gets past the BIOS splash screen, then displays GRUB. Proxmox then tries booting and displays:

Loading Linux 6.x
Loading pve

the screen then turns off (not black, off), my keyboard lights turn off and i can no longer turn off the server by pressing the power button for a long time. the fans inside the case keep spinning. the POST LED for the CPU briefly flashes, then the LED for "DRAM" turns on and the system stays like this until power is cut with the PSU switch. On reboot, the system takes a while to get to the BIOS splash screen, as if it were retraining RAM timings.

This behavior also happens with a spare windows install, but nothing is ever displayed after the BIOS splash.

here's all the steps i took to try to fix this issue:

  1. install BIOS updates to a version that supports the 4600G, both to the main and backup BIOS
  2. install EC FW Update Tool as required by gigabyte when the BIOS update instructions said to
  3. Installed a dGPU (a really used GT 730 I previously used with the 1600X)
  4. swapped back to the 1600X (it then booted correctly) and updated the system with apt
  5. Installed firmware-amd-graphics from the second item in this link installed with dpkg -x + cp -r + update-initramfs ( can't install through apt because it would uninstall the proxmox kernel)
  6. looked in journalctl -t kernel for any error messages, there were no messages at all from attempted boots with the 4600G
  7. ran memtest for 1 full pass. no RAM issues. this ram previously passed a full 48 hours memtest with no problems and is fairly new.
  8. Updated BIOS to the very latest version (54c)

The fact GRUB runs (and so does memtest) points me to a video driver issue on linux. I know from experience that ryzen APUs on linux require some special config, but i cannot for the life of me remember what this config is. The CPU may also simply be defective, but i doubt it.

System config:

proxmox 8 (up to date as of 2024-08-31) (problem also happens with a random windows 10 ssd that i have laying around, it works with the 1600X)
Ryzen 5 4600G
Gigabyte AB350 gaming-3 rev 1.0 (BIOS version F52e/F54c)
GT 730 (problem also happens without it)
GSkill Aegis K2 DDR4 2666 32 Go C19 (2*16), in slots marked red on the motherboard
Hard drives all report no SMART problems
Cooler master MWE 550W bronze V2 PSU
Improve this question
8
  • 1
    ummmm Youv'e did most of the troubleshooting. Have you tried to disable the iGPU in BIOS?
    – Netan
    Commented Sep 3 at 20:55
  • 2
    Try to boot into windows several times. It should offer you to boot in safe mode. Maybe that'll work? Also, remember, it might very well be a fault CPU.
    – Netan
    Commented Sep 3 at 20:57
  • 2
    Also, since that's affecting both Windows and Linux, I still suspect the BIOS/hardware; Have you tried to reset the BIOS after CPU replacement?
    – Netan
    Commented Sep 3 at 21:01
  • 1
    @Netan: i have not tried disabling the igpu, i was unaware that was an option and thought it disabled itself when it detects a dgpu. I am also leaning towards faulty CPU. i am in contact with the seller's support team but their response time is very slow. the BIOS has been reset multiple times due to BIOS updates and downgrades. Thanks for your input, will try to boot into safe mode and disalbing the igpu.
    – Themoonisacheese
    Commented Sep 4 at 8:09
  • I can't tag you for some reason? Anyhow - I meant, like, disable iGPU in BIOS. I also offered to try Safe Mode as an alternative OS with minimal drivers ( as memtest managed to boot )
    – Netan
    Commented Sep 4 at 21:28

1 Answer 1

Reset to default
0

Gigabyte motherboard says there is an issue with rev 1.0 motherboards, your chipset and Ryzen 5000 CPU so could be your issue:

Improve this answer
4
  • I know that there is some circular logic in AMD's marketing terms, but as far as i can tell the 4600G is not a 5000-series CPU. 5000 series are zen 3, 4000 are zen 2. maybe i'm wrong because it does count because it came out later? I'll ask gigabyte support.
    – Themoonisacheese
    Commented Sep 5 at 7:09
  • I am equally, probably more confused by AMD terms. I do think reaching out to gigabyte is best. I'd also try with just one RAM stick, then try a different single stick. I know you did memory tests, but different CPUs, usually in dual CPU PCs, have different memory requirements sometimes. Hardware troubleshooting usually involves removing anything but the bare minimum, then adding components one by one. I'm thinking try CPU with 1 ram, no storage, nothing plugged in except monitor & keyboard let it sit in BIOS for awhile. Maybe try live Linux, sit at installer prompt for a bit
    – gregg
    Commented Sep 5 at 12:41
  • I did also try with only one ram stick and the issue persists. i need storage for the bug to happen because it only happens after boot, but i'll try to unplug everything and only use a live linux stick, though i don't suspect that'll be any different. thanks for your input.
    – Themoonisacheese
    Commented Sep 5 at 12:47
  • 1
    Using a live linux stick has the same effect. i'm becoming more and more convinced the CPU is defective, but i'll wait for the sellers and gigabyte's response.
    – Themoonisacheese
    Commented Sep 5 at 18:00

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .