Hello ROG community,
I think I found a weird behavoiur of my X670E-E Mainboard.
I just bought a new system with the following components:
Newest BIOS 1602 ist flashed.
From the beginning I had the problem that in 1-2 out of 10 reboots or cold starts, my comoputer won't POST. Mostly I received Q code 46, sometimes 0d or 00.
If POST was successfull, system was always stable while running. It also passed various stress tests.
I never had POST problems with the EXPO profile I or II deactivated. Only with the EXPO profile activated.
I contacted the support of my OEM and Asus as well as AMD.
I finally found out, that the key was to change 1 voltage setting.
My RAM runs at VDD and VDDQ voltage of 1.35v.
In the bios I saw that my CPU VDDIO / MC voltage remained at around 1.13v even with EXPO profiles activated. The setting was "Auto".
In the web I read that the VDDIO / MC voltage must be increased by the mainboard while using EXPO profile. This does not happen in my case automatically.
So I manually set the VDDIO / MC voltage for the CPUs memory controller in the BIOS to 1.2v. No other voltages were changed. And since then: No instabilities, no POST problems at all.
This is good news, but there are two reasons why I am posting this:
Thanks a lot for your help in advance..
Auto rules for VDDIO scale with memory frequency, however, some CPUs may require more than others. 1.2v is fine if it aids in POST stability. Auto rules are not a one-size-fits-all solution and success cannot be guaranteed. After all, this is overclocking.
To fall in line with AMD's revised voltage recommendations, later BIOS auto ruling is likely going to be in the > 1.1 region for 6000MT. However, as you have discovered through tuning some CPUs may need a little more than this.
Thanks a lot for fast reply.
What I don't understand though is, that my mainboard probably does not behave like described by you: "Auto rules for VDDIO scale with memory frequency".
I observed the voltage being around 1.13 in both scenarios, with EXPO activated or deactivated.
This then seems to be a BIOS bug oder a hardware bug, or am I wrong?
I would absolutely understand your reply if without Expo the voltage was 0.9v, with Expo the mainboard addresses 1.1v e.g. but my CPU needs 1.2v. But in fact without seeting the voltage manually it always remains at aroung 1.13v.
It won't be a bug, no. I can't speak for R&D as to why the auto rules changed, but it's likely that a majority vote of CPUs do not require much of an adjustment. Usually, these rules are defined on a worst-case basis, but some CPUs may need more for upward of > 6000MT or with more than 1DPC.
SOC voltage is more important for memory stability, so this voltage does scale. Again, how much is needed still depends on the CPU sample.
I just wanted to inform you that I sent the computer to the OEM (RMA) and he replaced the CPU due to this behavior. He stated the CPU was defective and lowering the voltage is only a workaround to get defective CPUs to work but problems could increase in the future with those CPUs.
Now my POSTs always work perfectly fine with standard Expo settings.
Tuning voltages when we run the CPU out of spec is often expected and something system integrators should be aware of. Who is the OEM?
Glad you were able to get it resolved.