Asus Sabertooth X79 VRM Temps almost 100°!

Hey everyone, I've noticed that, with AI Suite Thermal Radar, the vcore temp is actually the VRM's above the CPU socket. During Prime95 stress testing, they eventually get up to and fluctuate between 80-96°C depending on the FFT size. I have two 140mm fans in the front (intake), a 140mm at the bottom (intake), and a 140mm at the back (exhaust). My CPU cooler is the H100i exhausting air out the top with two 120mm fans.

CPU: 3930K @ 4.6Ghz
RAM: 16GB Corsair Vengeance (4x4GB)

Digi-Power page:
LLC: Medium (25%)
Everything else: Auto/default

OC page:
OC type: Auto
BCLK: 100
Turbo: 46
DDR: 1600Mhz
VCORE: offset + 0.010
VCCSA: 1.1v
VTTCPU: 1.1v
Everything else: Auto/default

So with my voltages being very reasonable and my LLC only at Medium, why are my VRM's getting so hot? Is there a setting in Digi-Power that, when left on Auto cranks up a voltage that makes the VRM's run extra hot? I would think I have great cooling in my case with all the fans. My VCCSA temp only gets into the mid-70's max. So what gives with my VRM (Vcore) temp hitting 96° at times? It actually got up to 98° once and I noticed that my CPU throttled to 3.2Ghz briefly and then back up to 4.6Ghz. It did this a few times until I stopped Prime. My actual cpu core temps were in the low 70's at that point.

I've noticed that with gaming, 3d rendering, video encoding, compositing, etc....the VRM's never get above 80° - only with Prime95. So should I even worry about it?

Welcome to X79. That platform, no matter what board and what cooling is on it, tends to need active cooling on the VRM area once you start to overclock. Even at stock speeds under load, the heatsinks on all X79 boards I have tested were warm to the touch.

What clockspeeds are you at that you needed to raise VccSA and VttCPU?

Currently humming along at 4.6Ghz on a 3930k

4 or 8 slots populated for memory? You shouldnt need to adjust Vtt or anything unless you are overclocking by bclk/strap and not multiplier, or have 8 sticks in...

I have 16GB of Corsair Vengeance 1600Mhz (4x4GB) so, four sticks.

I originally left vccsa and vttcpu on auto/default but XMP setting for 1600Mhz is 1.3v for vccsa. I thought that was too high and knocked it down to 1.1v. I turned up the vttcpu to 1.1 because my system was 24-hr Prime stable on blend tests but was unstable at idle or menial tasks. It seems to have helped.

OC is 100 Bclk and 46 turbo

Vtt has nothing to with idle stability IIRC (not 100% on that). Are you using LLC (Load Line Calibration)? If you have that enabled, may want to boost the base voltage up a notch and lower the LLC level so your base voltage is higher and load voltage remains the same.

My settings:

vcore: offset +0.01v
vccsa: 1.1v
vttcpu: 1.1v
LLC: Medium (25%)

The only lower option for LLC is "Regular" which, I assume, just eliminates the droop but doesn't add any boost? If I leave LLC on Auto, it actually boosts my vcore even higher than on Medium. Would leaving any of the other Digi settings on Auto raise my VRM temps?

Would the 1.1v vttcpu voltage add any heat to the VRM's? In Thermal Radar, my vttcpu temp is even lower than my vccsa temp. I see people OCing to 4.6Ghz+ with higher vccsa/vttcpu voltages - like 1.2-1.3v. My 1.1v/1.1v seems fairly low and don't raise any warning colors (yellow/pink/red) in the bios.

I dont do offset, I used fixed, Offset only raises the voltage when under load. Perhaps switch to fixed and leave things the same? I dont know. Its been a while since I have been on this platform.

I believe offset also allows the voltage to drop when idle. For instance, when my cpu is idle and running at 1.2Ghz, the voltage is a mere .085v which is nice.

How would I go about adding a fan to the VRM heatsinks? Any recommendations? Placement? Size of fan?

Anything that gives airflow to the area is fine. I like 60mm X 15mm AMD heatsink fans for that purpose, though they aren't the quietest.
I would drop the oc till they have airflow, x79 cpu power bits are known to explode.

I'm looking into a 20CFM 60mm fan and a 30CFM 80mm fan that I can hang with tie-wraps from the top grill of my case. Should do the trick!

But my question remains that if it's already a Prime95-stable system and the VRM's never get above 80c when doing even my most intensive normal workload, should I even worry about the fan for the VRM's?

Well, that depends on how you look at it and how often you clean out your computer. If you clean often you'll probably be Ok. If you don't, those temps will go up as dust accumulates.
Also silicon degrades faster the higher the temps are. Between the two you could in theory have them go blammo in the future sometime.
Most likely you're fine, but is it worth the risk?

Alright. I'll try the 80 first and if it doesn't fit I'll try the 60. I'll post results later!

Bobnova said:

I would drop the oc till they have airflow, x79 cpu power bits are known to explode.

Click to expand...

I would really like to read an article about that if you could possibly provide a link to one.

PolePosition said:

I would really like to read an article about that if you could possibly provide a link to one.

Click to expand...

I heard of this issue as well and googled searched it for you:

https://www.google.com/webhp?source...f.&fp=17ff08c2ceac14d5&biw=1335&bih=910&ion=1


Wait for it...*

By that search, seems to be the Moffsets on Gigabyte boards, whereas on the Sabertooth, we're talking military specs, which I haven't heard any of them blowing up until now, and that is the article I want to see and not some search results which I could have typed "X79 vrm explodes' myself, which includes articles on exploding budgets.

LOL, you are so predictable. See that "wait for it *" in m previous post? I posted that there as a placeholder for this copy/pasted response I had ready as I knew you were going to make this about the Sabertooth.

Its not about the Sabertooth here. He BobN spoke in generics. It happens. Yes, it happened to Giga's mostly, and some Asrock's, but as BobN mentioned, it can happen to VRM/Mostfet. Overheating is part of the reason this can happen. Sabertooth or $200 X79 board, it can happen.

Click to expand...

BobN said:

x79 cpu power bits are known to explode.

Click to expand...

BobN said:

Also silicon degrades faster the higher the temps are. Between the two you could in theory have them go blammo in the future sometime.

Click to expand...

I didnt see him mention anything specific about the Sabertooth, did you? I dont care if its a Golden Unicorn, if its made out of silicon VRM's they can explode too! No board, cheap to expensive is immune to this behavior really.

Sorry, I didn't know what you meant with the "wait for it".

You mean there have been VRMs to explode on other platform besides the X79?
Or is this something that has ONLY happened on a X79 in the history of computers?

Its also my understanding that there are various types of silicone, including high temp silicone used in race car applications for mounting gaskets.

Guess that is where a 5 year warranty comes in handy.

I just read the title of this thread and thought that is what the discussion is about.

You shouldn't know what that meant... it was meant for me, to later show you I knew EXACTLY what you were going to say in response.

I'm certain all platforms have had that happen at some point, absolutely. I know P55 did it to me (EVGA). What is your point? All we are saying is the X79 platform's VRM's notoriously run warmer than say Z77/Z68/P67...And because of that with overclocking, more care should be used to keep those cool. Stop acting like a cornered animal, LOL!

There are various types of silicon, yep. However, I dont think the same silicon used in race car applications are used in wafers for IC's. But admitedly, this is out of my wheel house and I am just taking a guess.

Now's a good a time to ask as any. I've heard VRM's are safe to 100°c and I've heard they're max is about 120°c. Anybody on here know what the safe range and max are for VRM's?