Overclocking: Core Voltage adjusting ?

I've built two computers but have never played with overclocking. On the rig in my signature below, i used Asrock's F-streaming performance tuning utility, to get my system to 4.2 MHz and it's been stable at that frequency for almost 2 years now.

I use this computer, primarily for for video rendering and one of my software programs is a core hog, running all the cores to 97-100% load.

Yesterday i started researching overclocking basics, and found an interesting youtube video with one of the engineers from Asus giving a "how-to-tutorial", and one thing i noticed was that he indicated walking core voltage up in small increments, after increasing clock freq, and looking for stabile settings.

I bumped my clock freq to 4.3 using the Asrock tuning utility, and rendered a video file that i had already rendered, just to compare time to render. 4.3 knocked about 1 minute off a file that took 23 minutes to render at 4.2, and completed the file with no issues except temps jumped up to 81-84C range.

Using Intel's XTU utility, i went in and dropped core voltage in .005v increments. Asrock's utility had set the core voltage at 1.220 in "adaptive" mode. I left it adaptive and walked it down to 1.175 where it brought temps down a bit to 80-81C with the occasional bump to 82C on a couple of the cores. I monitored temps in both Intel's XTU as well as HWMonitor CPUID (both reported identical temps, btw.

As i adjusted voltage downward, at each increment i'd render the video file for 10-11 minutes to see what happened to temps. I noticed that as i got down to 1.190V, individual cores would report actual voltages slight higher than the value I'd adjusted to. Also i noticed a trend that as i lowered voltage, the actual values intel was reporting grew in terms of distance from the setting or target value. At 1.175V i noticed that difference grew to it's greatest distance, from a setting of 1.175V it was reporting 1.182V at times. So at that point i set core voltage back up to 1.180V and left it at that.

My question is, am i risking any damage to my CPU or mobo dropping the voltage like that? I'm only asking in reference to electrical values. Temp wise i'm going to add a 2nd fan to my cooler to see if i can't take another couple of degress C off.
 
Solution
It is impossible to damage a chip by undervolting it, only OVERvolting is able to kill your CPU.

You're doing good so far, going down on the vcore. But don't use just video editing as a stability tester, use OCCT or Aida64 and stress test that way.

The reason why the voltage is going up when at the 1.175 mark is because your Load Line Calibration needs to be tweaked a little bit. Load Line Calibration or LLC is a key feature in most motherboards that allows you to prevent what your experiencing, also known as vBoost (when voltage goes beyond what you specify) and vDroop (when voltage goes BELOW what you specify).

There are 5 levels typically in an LLC, start at level 1 and work your way up until you find you get none or a very little...
First off, adding a second fan wont really help the temps, but mainly know as long as you don't do anything stupid, lowering your voltage should not damage the cpu, in the case that the cpu doesn't have enough voltage the pc will either turn off, or black screen either way don't worry this happens to everyone. when this happens you know how low you can bring the voltage. :)
Hope this helped.
 
It is impossible to damage a chip by undervolting it, only OVERvolting is able to kill your CPU.

You're doing good so far, going down on the vcore. But don't use just video editing as a stability tester, use OCCT or Aida64 and stress test that way.

The reason why the voltage is going up when at the 1.175 mark is because your Load Line Calibration needs to be tweaked a little bit. Load Line Calibration or LLC is a key feature in most motherboards that allows you to prevent what your experiencing, also known as vBoost (when voltage goes beyond what you specify) and vDroop (when voltage goes BELOW what you specify).

There are 5 levels typically in an LLC, start at level 1 and work your way up until you find you get none or a very little amount of vboost/vdroop.

Here's a good technique on lowering voltages quickly and without wasting a lot of time.

1. Continue to lower voltages and stress test with Aida64/OCCT for 5 minutes.
2. Once you go unstable and BSOD (or black screen or freeze up), go back to the previous voltage that seems stable.
3. Stress test for 10 minutes.
4. If unstable, go up again on voltage and repeat step 3 if necessary.
5. If you are stable, run OCCT/Aida64 overnight. If stable, congrats!

Oh and one more thing, what CPU cooler do you have? If you upgrade to something beefy like a 240mm AIO or a high end dual fan tower air cooler, you should be able to get 4.5-4.8ghz at around the 1.250-1.3v mark. 1.175v is very very low, so low that you're basically at stock voltages.
 
Solution
appreciate the responses

cooler is a Noctua NH-U14S.

What about cache voltage - will bringing that down help any in lowering heat? It's currently set, by the Asrock utility, at 1.250V and in Static mode. I was going to try changing to Adaptive and lowering the voltage some
 
Yeah that CPU is mid-range for a CPU of your caliber. If you upgrade the CPU cooler, you'll get way higher clocks. If that's something you're up to.

Anyways, back on topic, don't touch the cache. Lowering voltage shouldn't do hardly anything in terms of temps. I am unaware that you could change the cache voltage to adaptive though, I'd google it and see what others say about adaptive voltage on cache.
 


it can be changed to adaptive in the intel XTU utility under Advanced Controls > Cache controls.

Lowering voltage on the CPU did lower temps, in fact more than i expected. at least while rendering a video file. I just ran OCCT for 5 minutes and temps went up to 87, so OCCT is definitely throwing a heavier load at the cpu.