Is that 1600 one of these 12nm refreshes or the original?
Should be the 14nm one. I still really wouldn't expect much difference on the kernel or anything. The Ryzen processors should be sharing much of the same code. It would be crazy to write entirely new code for every CPU. Its a maybe. I would look into other things before suspecting that.
Should be the 14nm one. I still really wouldn't expect much difference on the kernel or anything. The Ryzen processirs should be sharing much of the same code. It would be crazy to entirely write new code for every CPU.
I was thinking about the power stages (P-States?) and voltage curve.
Imho the 12nm should have a lower power consumption.
I'm still thinking the drop to a too low voltage might be the cause as there are always 1-2 cores listed and never all.
With "Typical current idle" my system is much more stable, but unexplained GPU crashes still happen from time to time, bringing down the whole system.
Thanks for the hint, i will have a look for that setting as soon as i check the bios.
I got the same problem using a 1700 Ryzen CPU and 64 GB RAM (in support list of Asrock x370 sli).
My OS is Fedora 32 with a 5.6.14 Kernel.
Bios settings are all at default.
Anybody got the same prob using 2nd GEN or higher Ryzen CPUs?
I solved my mce errors issues with clearing cmos (complete power drain) and bios reflash.
But this didn't solve the freezes that happend from time to time, for that I had to disable the C6 states with the zenstates script (r4m0n/ZenStates-Linux). I had the freezes in two different systems (same cpu but different components otherwise - btw both systems with asrock boards as well, x370 and b450) and the script fixed it in both cases.
With 2nd and 3rd gen ryzens I didn't have any of these issues.
My 1st gen ryzen 1700 only did this on linux until either microcode update
or kernel update solved it few years ago. Make (multithreaded compilation
system) still crashes in high cpu load and that is separate issue but it's
probably unique to this cpu and is 'hardware bug' rather than software bug.
ke 3. kesäk. 2020 klo 18.46 derstef <email@example.com> kirjoitti:
Re: Ryzen Instability MCE bea0000000000108? What do do next? in *General
Thanks for the hints.
I added an small offset (0,05v) to the voltage curve now and set LLC to 3 (on this Board 5 is weakest and 1 strongest correction).
Lets see if it helps.