30+ mV difference between BIOS CPU Vcore and CPU-Z reported Vcore

[scanny]

I'm still pretty new to overclocking (although I can tell already I'm quite hooked :). This is my first overclock.

I've been studying this forum for several days now and have already learned enough to get my EP45-UD3P/Q9550 box running stably at 3.8 MHz (475x8). Many thanks to all of you who have contributed to this great knowledge-base.

Just a quick question.

I'm noticing what seems to me to be a pretty big difference between the CPU Vcore I set in the BIOS and the one reported by CPU-Z (and every other monitor I have that shows Vcore, e.g. Everest, etc.)

BIOS Setting: 1.37500
CPU-Z Reading: 1.3440
Delta: 31 mV

I have Load Line Calibration (LLC) switched on, before it was even greater under load, when the sag kicked in.

My question is, is it normal to observe such a large delta between the BIOS setting and the sensor reading?

And if so, should the maximum rated voltage for the CPU (1.3625 in my case) be based on the BIOS setting or the one the sensor is showing?

Thanks much for any guidance you can offer on this.

[scanny]

Re: 30+ mV difference between BIOS CPU Vcore and CPU-Z reported Vcore

Ok, after further study on this forum I think I'm able to answer my own question, for the benefit of any other newbs who might find this thread on search.

I appeal to more experienced and wiser overclockers to either confirm or correct my statements.

----

There are three voltages related to CPU core that are of interest to overclockers. I'm not sure they all have official names, but for the sake of discussion I'll use the following three terms:

<ul>
<li>CPU Core (BIOS)</li>
<li>CPU Core (Sensor at idle)</li>
<li>CPU Core (Sensor at load)</li>
</ul>

CPU Core (BIOS) is the voltage level set in the BIOS (say 1.2500 for example)

CPU Core (Sensor at idle) is the voltage that actually makes it to the chip (say 1.2160 in this example) when the system is in an idle state, as evidenced by appearing in a sensor reading utility such as CPU-Z. This voltage is lower than the BIOS setting by roughly 30 - 40 mV due to voltage drops on the PC Board traces and other resistive elements between the supply and load.

CPU Core (Sensor at load) is the voltage that makes it to the chip when it is drawing full current, such as when running Intel Burn Test (IBT). The difference between this and the idle voltage is referred to as "Voltage Droop". I expect this is due to larger currents being drawn through the same resistive elements that cause the sensor@idle voltage to be lower than the value set in the BIOS. Note that settings such as "Load Line Calibration" (in Gigabyte BIOS) can counteract this effect.

From an alternate perspective, there is one base voltage (CPU Core (BIOS)) and two voltage drops, CPU Core drop, and CPU Core droop.

CPU Core drop is the difference between CPU Core (BIOS) and CPU Core (Sensor at idle).

<b>CPU Core drop = CPU Core (BIOS) - CPU Core (Sensor at idle)</b>

CPU Core droop is the difference between CPU Core (Sensor at idle) and CPU Core (Sensor at load).

<b>CPU Core droop = CPU Core (Sensor at idle) - CPU Core (Sensor at load)</b>

Maximum applied voltages for a particular chip stated by a manufacturer refer to CPU Core (Sensor at idle), the voltage that is actually applied to the chip, regardless of what value is set in the BIOS.

Hope this is of help to later newcomers like me.

[Lsdmeasap]

Re: 30+ mV difference between BIOS CPU Vcore and CPU-Z reported Vcore

You got it!

[scanny]

Re: 30+ mV difference between BIOS CPU Vcore and CPU-Z reported Vcore

Thanks so much for the confirmation Lsdmeasap (may I call you Timothy? (Leary :)?). I knew those electric circuits courses in engineering school would come in handy one day :)

I can't tell you how gratifying it is to have attracted the attention of such a luminary on this forum, if even just for a confirmation of my learning :)

I am so grateful for all the work you have contributed here. I've learned a very great deal from you over the dozens (hundreds?) of posts you've contributed that I've read and taken guidance from.

I'm just finalizing a 31% overclock of my EP45-UD3P / Q9550 (C1) rig (3.7 GHz), final P95 24hr stability test currently on hour 10. I had 3.8 GHz stable, but needed to back down the air cooling fan dB a bit and found an inflection point at 3.7 GHz. At every challenge I've found the next answer in these forums with your voice invariably in the solution set. It's been a great learning experience and a lot of fun. My new rig will make a great Hackintosh.

Will post my final BIOS settings for a final review as soon as I hit the 24 hour mark with P95.

All my thanks again for all you have contributed to me and to all of us who look to this forum for advice.