Spooky Electronics Store
Spooky Electronics Store
E:desktop installation in short. A few days back I adjusted CPU parameters in BIOS—multiplier from 34.5 to 42, voltage override set to 0.9V at 1.25V. Performance remained stable; HWINFO64 reported 1.194V during peak load (p95), peak applied voltage never exceeded 1.25V. Last night I attempted to reduce the voltage slightly, started by checking the load line range, moving from 1 to 3. Load voltage came in at 1.068V, with a maximum of 1.4V+. Since then my settings no longer align. I reset BIOS and flashed it (first time using present version AC0, then updated to latest beta AF5). Now the override is set to 1.1625V; current readings show max at 1.263V p95 load (mult 40). Other settings: SOC 1.2V, VDDP 1V, both VDDG 1.05V, DRAM 1.5V. I’m still puzzled about this issue. PBO is disabled and I skipped the AMD overclocking menu after flashing.
I observed that when a CPU experiences insufficient voltage, the bus frequency tends to fall below 99mhz. The impact varies with the severity; clocks can drop noticeably. Have you seen this before? If it happens, the device likely needs more power. However, if both CPU and board behave as expected, degradation might be the cause. A CPU usually provides optimal timing right out of the box. I've also seen processors become sluggish or even unreliable under certain conditions. The same applies to boards, and memory tends to degrade similarly. That’s what I’ve noticed so far.
Bus clock maintains stable 100.0MHz. Override is designed to deliver consistent voltage. Adjusting BIOS to 1.25v yields steady idle; this has been the standard for two years, and cooling seems unaffected. I configured CPU override between 1.1625-1.1875v, frequency between 3400-4000MHz. Results show stable 1.25v idle, 1.2v during p95 runs. At 3400MHz with 1.0v, idle reads 1.188v; p95 at 1.144-1.15v also holds. Previously running at 4.2GHz with 1.25v for several days, voltage stayed within limits. Changing load line caused fan speed to spike and voltage to exceed settings. Resetting BIOS and flashing resolved the issue, but voltage still surpassed target values.
CPU:Override supports three modes—Manual, Adaptive, and another one I don’t recall. The voltage range you mentioned seems unusual. And regarding the 1.625v to 1.875v shift, it might be a typo? Didn't exceed .9v at 3450mhz, but under prime95 it did go up to 1.144–1.15v. That’s typical LLC behavior. Right now, you’ll need to increase voltage further, which can raise it beyond what you’re used to. You’re asking if adjusting the voltage is correct before modifying the LLC circuit.
1.1625 to 1.1875 and 1.175 all led to idle at 1.25v load. Load 1.2v didn't exceed 0.9v; under load it was 0.875v, while idle stayed at 0.9v. Right now 3400 at 1.0v is showing 1.188 idle, with CPU voltage options including auto, override, offset, override+offset and AMD overclocking.
I understood right away from this text—it's about an MSI board. What exactly does it do? And yes, I face issues similar to yours; I’m pretty sure the motherboard is the culprit.
The voltage you select matches your observations closely and stays within the expected range. To determine p-state voltages and clocks, open CPU-Z, navigate to the tab, and click Save report.txt. Load the file and scroll to P-states to see the VID requests at different frequencies. This will give you approximate values for stability and a slight deviation. From my experience, using 3.7GHz with a maximum of 1.212V in p-state mode on my 2700X caused instability. Adjusting the voltage core or increasing the LLC helped bring it to around 1.250V for better performance. This seems consistent across two motherboards as well.
My issue lies in the fact that the voltage I adjusted and what I measured don’t match. It seems Hwinfo64 might be giving inconsistent readings, probably because I altered the load line from 1 to 3 just to test direction, and now it keeps changing even after fixing it. I’m worried about this affecting performance. I don’t need a supernatural fix—just reliable results. Please keep things stable, or else I’ll lose my mind.
HWInfo behaves differently from other monitoring tools due to its algorithmic processing. The data isn’t a direct sensor reading, and delays between measurement and reporting can affect accuracy. Running the system at standard settings shouldn’t cause issues, so the focus is on why specific parameters are restricted.