Z390 UD - LLC issue Question about the topic
Z390 UD - LLC issue Question about the topic
Vcore is the section you need to understand. The idle voltage is marginally greater than the load voltage. Reduce the LLC by one level. The maximum LLC that avoids flatlining or overshooting is optimal.
I've seen others discuss this in forums—some say VR Vout is more precise than Vcore. That claim might not be entirely correct. The VR Vout reading you're referring to is typically lower than Vcore, which aligns with your observation. Let me know if you'd like the exact value.
VR VOUT represents the ultimate voltage after the voltage drop. It's precise, aside from brief spikes. https://www.overclock.net/forum/5-intel-...st27736104 VR VOUT reflects what the CPU actually gets post-VDM. The input voltage coming from the CPU depends on the BIOS setting when using a fixed core voltage. This value excludes any fluctuations and can only be measured with an oscilloscope. The actual supply to the VRM comes from the CPU's VID output on AUTO VCORE, but only when the DC load is kept at 0.01 mOhms (which equals 1 in Gigabyte BIOS). This applies before any VRM droop adjustment takes place. The input limit stays at 1.520V for Auto vcore if the 33-hour command isn't run.
You're interpreting the relationship correctly. VR Vout should indeed be higher than Vcore, which explains your observations. Your measurements show stable values within expected ranges, so it's likely your sensors are functioning properly. No major issues detected.
VR VOUT is the correct value for your tests. Jurrunio's answer was incorrect. Review your other discussions on overclock.net and Reddit to understand how VR VOUT is determined and how to ensure its accuracy. Note that some high-end boards require a slightly lower voltage than what you input, which can cause issues due to a known bug. VR VOUT becomes unpredictable when C-states are active and the voltage is reduced.
You're still experiencing Vdroop around 1.273-1.24 VR Vout under load without overshooting. It looks good to keep the LLC on turbo—your precision in maintaining VR Vout gives you confidence. TY!
Also, should I be concerned about the 1.32v peaks on Vcore? I haven't confirmed yet, but they appear when the system shifts from load to idle and during light tasks like browsing. It seems like a brief voltage spike, likely related to Vcore. VR Vout never exceeds idle voltage (1.273v). Is this typical behavior?
The peak on the 'vcore' sensor isn't a concern since the vcore sensor is sensitive to power plane impedance. Problems with the vcore sensor have persisted since the introduction of the super I/O chip and have intensified on certain platforms as core counts rose. Maximus XI boards include a diode connecting the Super I/O chip to the VRM, enabling recalibration for accurate voltage readings from the VRM—similar to what's found on VR VOUT in Gigabyte, Asrock, and MSI boards. This feature is absent on Maximus X and older models. Onboard sensors can't capture transients; they occur in microseconds and are marked in Intel documentation as "o_vs" or similar, requiring an oscilloscope for detection. LLC=Turbo isn't problematic, but raising the BIOS voltage and setting LLC to high can improve stability during transients (keeping load VOUT consistent). This enhances transient response, reducing the jump from VR VOUT to transient VMIN readings. I switched from LLC Turbo to LLC High on my Aorus Master, increased the BIOS voltage, and achieved smoother performance—allowing stable 5.1 GHz playback of Battlefield 5 at 1.375V with LLC High, without cache L0 errors. Previously, I used 1.270V LLC Turbo at 5 GHz and 1.340V at 5.1 GHz. A video illustrates these changes.