VRM Cooling?

[yoadknux]

Hi,

I own an X79 rig, specifically my motherboard is Asus P9X79LE, my CPU is a 4930k @ 4.4GHz with voltage set at 1.36V, RAM is TridentX 2400MHz CL10 @ 1.65V, and VCCSA and VTT are both at 1.1V. system is running good and stable. However I want to push it even further, specifically 4.5 or 4.6GHz on my CPU. I have a Corsair h80i which is a so-so cooler, temps on prime95 highest at about 75c after a few hours. 4.5GHz and 4.6GHz can both boot, but voltage crosses 1.4V (I think something like 1.45V boots with 4.6GHz). Problem is I cannot get neither 4.5GHz nor 4.6GHz to be Prime95 stable, they can pass cinebench or intel burn test, I think RealBench as well, but Prime95 never. So I'm starting to wonder, maybe sustained heat is the problem, and specifically maybe on the VRMs. Is it worth it to place a fan specifially at the top heatsink of the motherboard? If so, which size and what direction (directly on it, to the size, below it, etc)

Wanna hear recommendations! If you think I'm chasing ghosts here don't hesitate to tell me! I love learning stuff here!

 

[DaveB]

Can't hurt to aim a fan at the VRM heatsink. Especially since you're water cooling so no fan on the CPU.

 

[EarthDog]

Does that mobo display VRM temps through software? AI Suite if it was around?Does the VRM heatsink get warm, hot, or HOT, to the touch under load?

That board was pretty solid so, it may be chasing ghosts... but you won't know unless you try. Point a fan at it, try to cool it off.

 

[yoadknux]

Cool, before I place the fan, let me just show you the cooling configuration right now:

I have a total of 3 intake fans: 2x120 at front, and 1x230 (or something like that) on the side. As for exhaust, It's one 230mm fan at the top, and if we're counting it, the 2x120mm rad of the CPU at the back. It can be shown here (sorry for the dust on the PSU shroud... I'll clean that later). See here:

https://imgur.com/a/Qv2tYgo

This is specifically the heatsink that I consider cooling:


I have access to 40mm fan, 60mm fan, 80mm fan, 120mm fan and even 140mm fan. There's no placement for it but with zipties I can get anything done, lol. Which one should I use and where should I direct it?

 

[EarthDog]

Id run the 40 or 60 and point it at the heatsink.

Dont hesitate to try things on your own!

 

[wingman99]

The VRM should work fine up to throttling point. You would know it is throttling by the reduced clock speed until the VRM cools down to reset the full processor clock speed.

 

[maxfly]

from your pics it looks like you really arent going to see much if any improvement in temps. your exhaust fans are right on top of it as is. pushing more air into one or the other may make your temps worse by changing the case flow.

 

[Alaric]

You could try flipping the top fan in the picture over.

Bearing this

may make your temps worse by changing the case flow.

in mind.

 

[DaveB]

from your pics it looks like you really arent going to see much if any improvement in temps. your exhaust fans are right on top of it as is. pushing more air into one or the other may make your temps worse by changing the case flow.

A little 40mm won't affect the case flow at all, but will provide some airflow over the VRM heatsink. It helped my mini ITX experiments when I was running an overclocked i7-8700K which overheated the VRMs. No fan was needed for an overclocked i3-8350K, stock i5-8400, overclocked i5-8600K or stock i7-8700.

 

[ihrsetrdr]

 

[Alaric]

^Been there, done that. LOL

 

[yoadknux]

Update: I have done stress tests, where the idea is to get a feel of how hot the VRM heatsink is.

I've ran Prime95 for an hour and touched the heatsink. It was too hot to touch for more than 4-5 seconds, compareable to the backplate of a 1080ti after it runs Furmark. But I knew that this isn't really qualitative info, I searched for info in the BIOS but it only has CPU and motherboard temperature. Luckily for me, I've found out that my Scythe Kaze Master Flat II fan controller has the ability to measure external temperatures using, errr... some thin material. Maybe it's a thermocoupler? Anyway, I've placed right between the tightest fins of the heatsink, this time ran Intel Burn test on very high settings, and.... temps were about 65c. Done it again with Aida64, same result.

I'll throw in the 40mm fan and measure it again and see how it goes. Then I'll try to continue overclocking.

 

[UltraTaco]

If you can hold your finger on it more than a second, it's not hot enough. If throttle happens at 100°C, try sticking finger into boiling water. If it feels the same, then yeah it's probably throttling

Hehe that Fan-Mod™, been there too.

 

[Alaric]

try sticking finger into boiling water.

This is where someone says "Don't try this at home, kids!".

 

[wingman99]

Update: I have done stress tests, where the idea is to get a feel of how hot the VRM heatsink is.

I've ran Prime95 for an hour and touched the heatsink. It was too hot to touch for more than 4-5 seconds, compareable to the backplate of a 1080ti after it runs Furmark. But I knew that this isn't really qualitative info, I searched for info in the BIOS but it only has CPU and motherboard temperature. Luckily for me, I've found out that my Scythe Kaze Master Flat II fan controller has the ability to measure external temperatures using, errr... some thin material. Maybe it's a thermocoupler? Anyway, I've placed right between the tightest fins of the heatsink, this time ran Intel Burn test on very high settings, and.... temps were about 65c. Done it again with Aida64, same result.

I'll throw in the 40mm fan and measure it again and see how it goes. Then I'll try to continue overclocking.

When you stress tests is the processor throttling the clock speed? If it does not the VRM is not overheating. I have not seen any VRM problems overclocking unless they overheat or shutdown.

 

[yoadknux]

When you stress tests is the processor throttling the clock speed? If it does not the VRM is not overheating. I have not seen any VRM problems unless they overheat or shutdown.

There are some things I've noticed about clock, usage and BCLK when I run Prime95:

- Most of the time, everything is normal: BCLK 100, multiplier stable, CPU usage 100%.
- On overclocks that I still cannot achieve such as 4.5GHz, CPU usage can drop to 91%-93%, but clock remains 4.5 (Throttle?)
- Weirdest thing, probably not related to overheat: BCLK increases from 100 to 104... with usage still 100%. This has happened when I switched to 2400MHz DDR3 RAM, and EarthDog said it's normal. Problem is it increases CPU clock to almost 4.6GHz... So I've tried something funny, I set the strap to Auto, lowered the multiplier, and changed BCLK in BIOS to 104, and... it goes up to 108... wtf

 

[wingman99]

When stress testing the processor and it stays at 4.5GHz the VRM is not throttling the clock speed. With prime95 when workers fail and drop work per core the utilization % is reduced, you can see this happening with HWmonitor CPU utilization.

When overclocking memory from stock speed the BCLK is increased to meet the memory clock speed that the CPU memory dividers can't provide, it is normal. If you run the memory at BIOS default you will see.

 

[Organik]

My VRM is usually like 42c .... Also I have taken out 2 of the fans mini fans that sound like a jet took them both out,, now silent system, much better.

 

[UltraTaco]

These VRMs though...
Always center of topic!

 

[yoadknux]

When stress testing the processor and it stays at 4.5GHz the VRM is not throttling the clock speed. With prime95 when workers fail and drop work per core the utilization % is reduced, you can see this happening with HWmonitor CPU utilization.

When overclocking memory from stock speed the BCLK is increased to meet the memory clock speed that the CPU memory dividers can't provide, it is normal. If you run the memory at BIOS default you will see.

What do you mean by "overclocking memory from stock speed"? This is the stock speed for those memories. I haven't enabled XMP but those are the timings written on the sticks.