Intel CPU overvoltage protection settings
Intel CPU overvoltage protection settings
When I first moved to Intel, I learned about these PLD voltages and thought they mattered. At first setup I experimented a bit but didn’t see much effect. I saw clock stretching and didn’t connect it to that. Later I realized low PLD voltage causes clock stretching. With +75mV it was slower than active clock during stress tests, but raising it to +120mV gave a stable 6000MHz across all cores. I also noticed less stuttering in games when it was set correctly. After that, I adjusted all other PLD settings to +120 because I’m concerned low values might cause similar issues. I’m curious if these parts actually do clock stretching and whether higher values impact their lifespan, though I haven’t seen big temperature changes. My rig has an i9 14900K with disabled E-cores and H-types, locked at 6GHz all-core and 5GHz ring, DDR5 7200CL34/44/44/30 with 1.435V from GB GZ790 Aorus Elite AX RX 4090 gaming x trio.
It shouldn't be this way, the limits you can set are around +150mV according to Intel specs, and most of what I've seen works fine at that level. In reality, the PLL voltages for the IMC don't really contribute much beyond that, and they tend to be at optimal levels—like +105 for the 13700K except for IMC, which is usually +90. Higher voltages probably won't make a big difference.
I haven't seen much impact from imc pll voltage on memory stability. It might matter more if my RAM was pushed to very high speeds like 8000mhz or above, but my motherboard can't go higher than 7200 regardless. What are your thoughts? Also, how do these numbers influence clock stretching? Do they affect just the CPU or other components too?
It can help with 7200, though that depends more on the CPU, and from what I've seen it doesn't do a whole lot on 14th gen chips either. none of them really should, though AFAIK there's no clock stretching circuitry in the memory system, only the actual CPU core.
Heard mixed things about 14th gen memory controllers—some say they’re better, others doubt it. Still struggling with stable 7400mhz and the removal of the 7300mhz choice. Which clock settings need more attention? I guess I should focus on the ring clock first, since it’s the main concern. The system agent doesn’t seem to have much info either, and I’m confused about why clock stretching isn’t an option—it would simplify things.
Generally speaking yes, though the outcome often depends on the silicon draw. Many cases show 13th gen chips outperforming 14th gen in memory controller performance. Wow! Madness!'s 13700K reaches up to 8266MT/s on four DIMM boards. Buildzoid's 14900K hits 7600MT/s on the same setup. It shouldn't be this way, as the ring AFAIK lacks clock stretching circuits. The minimum and maximum ring frequencies are mainly related to Ring Downbin, which operates at lower speeds at specific stages (the exact behavior varies by board). This affects stability more than raw speed.
I grasped that PLL overvoltages enable smaller voltage drops during adaptive voltage scaling. This shouldn’t impact clock speeds, though it might only slightly improve overclock stability—rarely enough to notice. It mainly affects chip longevity and offers minimal benefit in most cases. In situations where PLL overvoltages seem useful, they’re uncommon. The higher frequencies you observe might stem from a BIOS issue. Modern CPUs are designed with adaptive behaviors, so adjusting PLL settings can unintentionally influence performance.