Background:
Ryzen 5900x on an MSI MEG ACE x570. (2x16gb dual ranked sticks, running 3600cl14, flawlessly. Yes I can also set my RAM to CL18 when I want to test higher frequencies like 4000mhz. So the RAM will do up to 4400 before it runs out of room on the controller etc)
Problem:
FCLK 1800 will work, and the system runs fine, after much tweaking. What does AMD recommend in order to try to boot at FCLK 1900 and above? I can boot up, and I can run fine, but I get WHEA errors about once every 4 minutes. If I had to take a guess: You changed the bridge on the IOD, in order to support 2 CCDs with separate memory controllers, but due to a bug in the synch operations on the IOD; it's trying to join 2 or 4 NUMA nodes into a single memory bus for the motherboard. (Am I right?) But the bridge SUCKS at 1t and 2t command rates, it's built for GDM and there's a bug with the other modes. You didn't tell the motherboard manufacturers that they have to add a 2 cycle delay during a cad bus transition, so the system throws Bus Interconnect errors. (Am I right?)
PLEASE just tell us - what's the bug over Fclk1800? Can I lock in zero or 1 numa node per socket, and add some kind of bus delays somewhere?
Here's the error.
"
Log Name: System
Source: WHEA-Logger
Event ID: 19
User: Local Service
OpCode: Info
A corrected hardware error has occurred.
Reported by component: Processor Core
Error Source: Unknown Error Source
Error Type: Bus/Interconnect Error
Processor APIC ID: 0
The details view of this entry contains further information.
"
Concern:
Also, please tell us: Is it safe to use the computer while these errors are appearing, or does this indicate an unsafe operation condition? Does this indicate we are damaging some components? Would this start a fire? Is there any place we can get more data on this error and the circumstances surrounding the events? Specifically, I'd like to know which " Bus/Interconnect" is having an error?
Helpful information for other customers:
FYI: The number 1 problem with crashing at FCLK 1800 and lower, is people trying to use 1t command rate on the RAM instead of using GearDownMode or 2t command rate. Seriously, if you haven't already tried 2T or GearDownMode, you need to try it, for the max fclk and stability. Also, note that some BIOS have two sections for configuration; One is called "AMD Overclocking" and one is called "Settings". We have to reset the "AMD Overclocking" section to ALL AUTO/defaults... And only use the DRAM settings in the main motherboard settings area within BIOS. Then double check that you're setting ram voltage correctly, per the manufacturer's data.
PS - the new BIOS (AGESA 1.2.0.0+) is the only reason I can run Fclk 1800. I find it disgusting that you sold products that wouldn't work at the advertised frequencies, and it took you 3 months to even get close to a stable BIOS.
Side note: Don't use Ryzen Master Game Mode profile, unless you want to turn off half your CPU. As a matter of fact, why is this option NOT in the BIOS settings? Shouldn't every single feature found inside Ryzen Master (such as "legacy mode" also have a setting in BIOS where we can turn it on and off without using the Ryzen Master Software?