Ryzen 7 3700x with kernel 5.12.11-1.el7.elrepo.x86_64 Panic

Hi All!

I currently use AMD Ryzen 3 2200G (built in graphics) with kernel 5.12. I’m trying to upgrade it to Ryzen 7 3700x (no GPU, but insert some old Nvidia which initializes well)… but get kernel panic. Do you have any idea where to start?

Maybe try another kernel or reinstall it to fix possible initramfs issues:

I’d have to do it with Ryzen 2nd gen (Ryzen 3 2200G ). There were some strange things with installation of NethServer: I had to install it in virtual machine, update kernel to 5.12, and then use this installation with Ryzen 3 2200G. It is very stable. Then I had some config restore and strangely enough, Ryzen 3 2200G accidentally started with kernel 3.10 (or 3.11 I don’t remember) while it couldn’t even go through installation of NethServer with it.

I should have taken a picture of the kernel panic message, but I vaguely remember it was about SATA or some ATA… But all things are the same except CPU (Gen 3 vs Gen 2). Since it gives me kernel panic, then at least HDD works - it loaded kernel! Memory works… What could go whrong USB (which I don’t use) or energy saving…

I’m currently looking into difference between 2nd and 3rd Gen to see if I could switch off something in BIOS.

BTW BIOS works and it is of the latest version (50D, while 3rd Gen support came with 40F).

@sashaluda maybe it’s the “perfect occasion” to go for Proxmox.
This will allow you to get the full experience of your APU capabilites, and have Nethserver run without any hassle.


Yeah… I think I’ll have to do it eventually. Only right now I’m a little overwhelmed with other responsibilities at my work. I hope this old dog can learn some new tricks :slight_smile:

So, I installed a copy of Promox… and as I was installing it, I realized: I will have this server installed on my hardware, which I plan to change once in a while… So, instead of reconfiguring NethServer (which turned out not so difficult with the option of changing lan interface while importing config) I will have to reconfigure Promox.

Doesn’t this defeat the purpose of installing Promox? (so, I don’t have to reconfigure things)
I simply shift the problem from configuring Centos 7 to Debian 11. Yes, they are realy different in their HW support approach… but I would simply add one more layer of trouble (learning new trick)…

So, I would like to try to solve Centos trouble first… only if I get really stuck – I’ll continue my exploration of Promox…

How do I install kernel 5 through Software Center? Or, maybe you think that some other kernel should be good for Ryzen 7 3700x?



Not really.
No matter what hardware, a Proxmox install (vanilla) takes about 30 minutes.
Disaster recovery on any hardware is basically install Proxmox, then restore VMs.
You can easily spend a day playing with Kernel builds, but that’s “fun”, not really professional!

The last 5 years I have only installed virtualized servers… :slight_smile:

My 2 cents

I just wanted to exclude an initramfs error by reinstalling a kernel (which creates new initramfs), it can also be done on command line.

So, I’ve done some reading and experimenting:

  1. I found out that only starting from kernel 5.13 that Ryzen 7 3700x is fully supported.
  2. I installed latest kernel 5.14… (from ELRepo | kernel-ml)
  3. Took out Ryzen 3 2nd gen, inserted Ryzen 7 3rd gen → got kernel panic. This time I recorded it on my cellphone :slight_smile: (later rerecorded it with boot parameter set debug=all)
  4. Took out Ryzen 7 3rd gen, put back Ryzen 3 2nd gen → all is well.
  5. Looked carefully at another gen 3 CPU available: Ryzen 5 3600
  6. Decided to try it… → kernel panic
  7. Put back Ryzen 3 2nd gen and took some break for thinking and studying the recording…

So, I’m planning to look at the video and see when the first error occurs…

So, looking at the video (don’t laugh), I found out that the very first error happens after trying to allocate some memory (system freezes for some time at that) and eventually it hits:

general protection fault, probably for non-canonical address 0x78

Prior to that I saw the work with GPT of two disks, and read of initframfs… but all gets stuck after this message.

I already found a recommendation to run memtest. I’ll do it when I get a chance.

Besides that 2nd and 3rd gen can simply differ by supported speeds of RAM, and my Ryzen 7 could be trying higher frequencies… so the fault could be with RAM.

Ryzen take a lot advantage of XMP and faster memory modules. But the DDR4 Specs are… lower, a lot than what published as “supported”. Moreover, getting greedy with faster memory speed can lead to not nice results, if the timings are not “well supported enough”.
Safest setting shuld be 2400/300mhz, then upscale to 2666/333Mhz. Also, for desktop-level mainboard (when not using branded business-class desktops) refer to QVL should be… nice.
Someone says “i want 3200” (which means 400mhz of clock) but not all the sticks can relate reliably with mainboards at that speed. Even when it’s “certified” in some way by XMP.

CentOS 7 is not that picky on hardware, but as any other Linux OS don’t like at all unstable and unreliable RAM setups. So the issues you encountered maybe was not due by kernels…

Take your time, take your tests.

Well… It is SOLVED! Believe it or not, the problem was in the memory. I had two DDR4 4GB modules 2666, and it looks like with 2nd gen Ryzen they were working as 2133 and all was fine and dandy. But once I installed any 3rd gen CPU, both modules started to work as 2666 and began to crash.
As of today, I’ve already done several installations (restore of backups – both config and data) on several configurations of the hardware, and now can do it with my eyes shut without any worries.
In the end I’ve got 3200 8GB single DIMM (planning to add another one later), 3rd gen Ryzen 7, B450 MB. Things work very smooth and snappy.

Thank you all for help.


Take a test if any bios updates of your MB can improve the management of the memory sticks.
And anyway… use Memtest86+ to assess stability of your config before running the system.

1 Like