Comprehensive Memory Overclocking Guide

[XenosTech]

Now that I'v seen this post again, I have to sit down and tinker with this 3600 and see how the memory behaves

[Benji_w]

I'm wondering if my IMC is faulty on my 6700k with z170 board. I cannot use more than 1.150v for VCCIO however VCCSA can go all the way up to 1.35v no issue. If i set VCCIO to anything over 1.150v the audio becomes completely distorted and unbarable. The system also appears to lose its snappy-ness?

[MageTank]

10 hours ago, Benji_w said:

I'm wondering if my IMC is faulty on my 6700k with z170 board. I cannot use more than 1.150v for VCCIO however VCCSA can go all the way up to 1.35v no issue. If i set VCCIO to anything over 1.150v the audio becomes completely distorted and unbarable. The system also appears to lose its snappy-ness?

VCCIO isn't something that provides stability the higher it gets. It's actually very sensitive to "voltage holes" where certain values (both low or high) are simply unstable. For example: I've had scenarios where 950mv was stable but 1000mv wasn't. 50mv difference resulted in complete instability. 1050 was just as stable as 950.

 

I do not like the modern trend of throwing more VCCIO/VCCSA at things for stability, but I am seeing it far more often with most memory overclockers these days. My testing just doesn't show good scaling. The same can be said with VDDP on AMD. If you pay attention to AM5 and the latest 1.0.0.7B AGESA firmware, they actually lowered VDDP from 1.15V down to .9V on most boards because vendors were throwing way too much voltage at XMP kits, resulting in instability and higher boot times. People attributed faster boots to MCR, but it was actually the VDDP voltage.

 

For your issue in particular, might be worth trying a different BIOS. The noise you are hearing can be caused by impedance which would explain why VCCIO is impacting the noise. I would also recommend experimenting with lower VCCIO voltages, I promise you can find stability just as easily with less voltage.

[Benji_w]

On 8/1/2023 at 2:31 PM, MageTank said:

VCCIO isn't something that provides stability the higher it gets. It's actually very sensitive to "voltage holes" where certain values (both low or high) are simply unstable. For example: I've had scenarios where 950mv was stable but 1000mv wasn't. 50mv difference resulted in complete instability. 1050 was just as stable as 950.

 

I do not like the modern trend of throwing more VCCIO/VCCSA at things for stability, but I am seeing it far more often with most memory overclockers these days. My testing just doesn't show good scaling. The same can be said with VDDP on AMD. If you pay attention to AM5 and the latest 1.0.0.7B AGESA firmware, they actually lowered VDDP from 1.15V down to .9V on most boards because vendors were throwing way too much voltage at XMP kits, resulting in instability and higher boot times. People attributed faster boots to MCR, but it was actually the VDDP voltage.

 

For your issue in particular, might be worth trying a different BIOS. The noise you are hearing can be caused by impedance which would explain why VCCIO is impacting the noise. I would also recommend experimenting with lower VCCIO voltages, I promise you can find stability just as easily with less voltage.

I wasn't aware of that, does VCCSA follow the same trend? if so i'll do more testing with that. As for my VCCIO voltage i think i've found the sweet spot for it so will just leave it pinned there. My board has the current available bios and i won't be getting any updates now as its Z170.

[MageTank]

5 hours ago, Benji_w said:

I wasn't aware of that, does VCCSA follow the same trend? if so i'll do more testing with that. As for my VCCIO voltage i think i've found the sweet spot for it so will just leave it pinned there. My board has the current available bios and i won't be getting any updates now as its Z170.

It does, but to a lesser extent. When I used skylake, I didn't exceed 1.15v on either rail, but once I got my OC dialed in, both were stable at 950mv. VCCSA tends to scale a little better though, so having it be higher than VCCIO wouldn't be uncommon, just make sure you avoid pushing too high. Most of the people recommending 1.3-1.35v on VCCSA back then had no idea about IMC longevity on Skylake, they were just taking a guess at it, lol.

 

As for the BIOS, newest isn't always the best, especially when it comes to memory stability. If it supports rolling back, might be worth trying a different BIOS from an earlier version. I don't know if people discuss BIOS testing when it comes to memory OC on LTT, but the OCN forums might have older threads where this was investigated on the Intel CPU's sub-forum.