Thread about Intel Skylake/Coffeelake voltage specifications.
Thread about Intel Skylake/Coffeelake voltage specifications.
The suggested voltage comes from Asus, aiming to keep things simple for everyday users under typical usage. However, the details are much more intricate than that. Check the Intel datasheets. https://www.intel.com/content/www/us/en/...vol-1.html Review the electrical specs. You’ll notice the mention of "loadline slope within the VRM loop capability." This refers to the standard VRM loadline or Intel’s default calibration. There’s no minimum listed, which is because it’s meant to operate around 1.6 mOhms (2.1 for six-core chips). A lower number reduces voltage drop, but the design assumes a certain drop. Raja from Asus explained in a meeting that loadline tuning is mainly for enthusiasts, not for standard users. They didn’t aim for high LLC levels since customers wanted it. Intel’s own guidance uses the 'AC Loadline' setting and adjusts voltages like 'Thermal Velocity Boost' to lower voltage by 1.5mV per degree Celsius drop starting at 100°C down to -150mV at 0°C. You’ll also find maximum current and voltage limits there. All these figures connect together. It follows a straightforward rule. These numbers apply only to peak performance (where the chip stays safe long-term). This includes auto-voltage control. The maximum voltage is calculated as R × I, which keeps the system safe up to 100°C. Current remains variable, while VID and resistance stay constant. The max VID is determined by multiplying resistance by current, subtracting from the ideal value. For example, with 1.6 mOhms VRM and 100A draw, the calculation shows a safe load voltage of about 1360mV. Using this logic, you can see why Intel recommends around 1.35V for typical use. When you plug in higher loads—say 125A for games like Battlefield 5—the math confirms it stays within limits. This explains the seemingly simple advice, but it’s based on careful engineering.
The goal of boosting performance is to exceed expected limits, isn't it? The voltage patterns you described are typical for Intel CPUs since the introduction of high turbo clocks on Coffee Lake. That 1.4V figure likely stems from the idea that these processors don’t need such power for a decade or more, making this voltage sufficient for at least five years before resale value drops. My 2600k serves as proof of this concept. The previous owner (my uncle) used it at 1.47V and handed it over after five years, claiming it reached 5.05GHz. Even though it now caps around 4.75GHz at that setting, I still kept it that way—only using it about three months a year, so I’m comfortable with the wear over a full year in just three months.
Yes, the issue wasn't due to a 1.47V BIOS setting. It was caused by the loadline specs being broken at that voltage. If Intel vdroop had been followed, the problem wouldn’t have occurred. You need to understand the specs and how the motherboard calibration affects the defaults. I’ve seen this happen with two 2600k units.
Under the minimum LLC configurations, all boards operate with the CPU vendor spec vdroop. Therefore, maintaining the lowest LLC settings is recommended, while increasing the base voltage to offset the vdroop. This typically results in a voltage of around 1.6V or higher for your board.
An aggressive LLC accelerates chip degradation more than a less aggressive one, especially at higher voltages. Running at 1.6V with Intel VDROOP will cause degradation even without sub-ambient cooling. A 1.35V setup with moderate LLC offers a reasonable compromise. Going to 1.40V with maximum LLC is highly detrimental.
Likely quite significant when using ambient cooling. Anyone missing direct die and delivering 360 rads three times or more wouldn't be able to run a 1.327v VR VOUT chip at 193 amps while staying below 100°C.
This video explains why excessive load line calibration is problematic and highlights the benefits of vdroop.
What works best for you depends on your priorities. One option maintains a steady 1.408 in idle, while the other stays at 1.45 but drops to 1.408 when the second highest LLC kicks in. Your HWINFO after two hours of intense gameplay shows the second one never falls too much (at least in software tests).
A 0 mOhm loadline is simply terrible. I wouldn’t rely on the second highest LLC on any board without top-tier VRM, particularly at that voltage. The spikes will be just as problematic. I’d only accept a 1.45V bios voltage without any LLC at all. There’s a high chance of accelerated wear with any LLC except the lowest ones. I explained the calculations earlier (the 6-core CFL was built for a 2.1 mOhms loadline, meaning there’s a specific voltage/current relationship that shouldn’t be strayed too much from the higher amps side—lower sides are always safe). Since Asrock’s LLC is inverted compared to Asus’, I wouldn’t exceed LLC3 with such a high voltage. I’d stick to LLC2 only at a much lower voltage (not above 1.30V). I’m not even certain if Asrock’s LLC1 is truly zero ohms—check the buildzoid video for details. But this is your CPU; you’re free to adjust it. Not every CPU is created equal—some can withstand more stress than others.
Uhm... without an LLC at 1.408v my speed drops more than 160mv, which means I’d be running around 4.7 to 4.8 GHz instead of the expected 5.2. That would let me run my CPU at stock speeds, which isn’t really the goal of overclocking. I understand why a 0 mOhm LLC is risky, but using anything above 1.3v with an LLC at level 2 seems reckless. It’s all about pushing performance, not just keeping things safe. I’m not paranoid—overclocking is about getting the best out of the hardware. There are many CPUs in the millions now, and I’ve never heard of any issues from running them at high voltages with LLC. Some people even run their CPUs at 2V and they work perfectly fine. I’m considering sticking to 1.425v at level 2 for now, unless there’s a future risk of degradation or failure. If the CPU starts to fail soon, I’ll bring it up here.