Importance of active VRM cooling?

pgi947 · 2 Feb 2013 at 22:53

Evening all!

I thought I'd put this here as its more to do with GPU Clocking then watercooling

I'm just putting together some orders for the last parts of my build, 7970 waterblocks are becoming a complete pain to order, due to stock levels, prices amongst other things

I have noticed however that some waterblocks don't appear to have active VRM cooling (as in water passing over the VRM area).

The Alphacool blocks (V1, V2 have active cooling but cost as much as EK nickel blocks which voids the point of me buying AC) I have been looking at only have a strip of copper sitting over the VRM bank.

Some other blocks don't even have this, the area is totally neglected! *cough* XSPC *cough*

So my question is, how is this likely to affect temps and over-all overclocking performance? Would a solid strip of copper be enough or is it likely to hinder my overclocking?

My initial understanding was a cooler core means cooler VRM's, how true is this?

Gregster · 2 Feb 2013 at 22:57

With out any prior experience of water cooling, I can tell you with GPU knowledge that the VRM's will play a big part. Maybe spot cool fans would work. You just need to keep them under 120C or it will create a Black hole effect and consume the Earth and it will be the end of mankind.

I may exagerate a little though

Bacon? · 2 Feb 2013 at 23:02

Only a tad

pgi947 · 2 Feb 2013 at 23:30

What is the melting point of copper?

Non active block:

Active block:

Minstadave · 2 Feb 2013 at 23:33

Wouldn't mind if you could put a decent air cooled heat sink on the VRMs but that non active block covers them without offering any cooling.

andybird123 · 2 Feb 2013 at 23:35

little difference - full cover blocks without water flowing around the actual VRM area are still covered by the same baseplate... VRM's don't need to be kept as cool as GPU's themselves, so what little temp difference there is in the base plate from the VRM to the GPU/memory area will have little effect on overclocking ability

I would have thought

edit; sorry, I can see some of them it's via the top plate - but even still, this still has direct contact with water and the baseplate so the heat will still be getting spread from the VRM's to the water

pgi947 · 2 Feb 2013 at 23:39

Minstadave said:
Wouldn't mind if you could put a decent air cooled heat sink on the VRMs but that non active block covers them without offering any cooling.

The non active block still uses thermal pads, with the copper strip acting as a heatsink.

andybird123 said:
little difference - full cover blocks without water flowing around the actual VRM area are still covered by the same baseplate... VRM's don't need to be kept as cool as GPU's themselves, so what little temp difference there is in the base plate from the VRM to the GPU/memory area will have little effect on overclocking ability

I would have thought

This was my initial thinking, then realised I really have no idea on the intimate ins and outs of what parts need to be the coolest on a GPU

So in short, as long as the copper block is being cooled (which it is) I'm good to go?

andybird123 · 2 Feb 2013 at 23:50

pgi947 said:
This was my initial thinking, then realised I really have no idea on the intimate ins and outs of what parts need to be the coolest on a GPU

So in short, as long as the copper block is being cooled (which it is) I'm good to go?

from the reviews I've seen of a couple of active vs. non-active, the actual overclocking results were determined more by the actual GPU temp than whether it had active VRM cooling or not

the alphacool is marginally better than the XSPC, but the EK is the best of the 3
watercool heatkiller GPU 7970 is about the best I've seen reviewed, and that has both the best GPU temp and active VRM cooling (and it's the same price as the EK in euro land)

pgi947 · 2 Feb 2013 at 23:58

Cheers Andy. Think I'll go for the alphacool blocks then

A little googling I found this, its not aimed at WC but dear god look at the VRM temps on that XFX 7970 DD!! Its shocking, worse than a reference cooler!!

http://www.hardware.fr/articles/853-17/thermographie-infrarouge-cartes-graphiques.html

dave_beast · 3 Feb 2013 at 00:13

I have an Alphacool 7970 block, my card never goes over 49c in game on a single 240mm 60mm AC rad.

hedgey · 3 Feb 2013 at 00:20

Uncle Petey · 3 Feb 2013 at 00:29

^HAHAHA!

I want to offer constructive advice about VRM cooling, but A: I know squat about waterblocks, and B: I can't stop laughing at the above post.

pgi947 · 3 Feb 2013 at 00:29

dave_beast said:
I have an Alphacool 7970 block, my card never goes over 49c in game on a single 240mm 60mm AC rad.

Nice, what clocks Dave?

Got any pics? Thinking about the polished stainless top version

Edit: also what readouts do you get from your vrms while under load?

Lol @ hedgey

Yaayuh! · 3 Feb 2013 at 08:49

Put it this way.
My 7970 doesn't like anything over about 70c on the VRMs when overclocking. It's very easy for them to get higher than that when pumping the volts.
Luckily the Matrix Platinum Fan2 is linked to the VRM temperature, so it's easy to keep the temps down.
When I questioned ASUS to why this is the case, they said that the temperature tolerance of the VRM is pretty much the same as the GPU when overclocking it.

If you have a choice to actively cool them, do it.

Minstadave · 3 Feb 2013 at 09:09

pgi947 said:
The non active block still uses thermal pads, with the copper strip acting as a heatsink.

Yes that is obvious, however a copper block has very little surface area and that will mean it will heat up, it won't be cooled effectively by the contact with the rest if the waterblock, meaning you'd be better off without it and sticking a decent air cooled heat sink on.

sieze2 · 3 Feb 2013 at 09:16

Minstadave said:
Yes that is obvious, however a copper block has very little surface area and that will mean it will heat up, it won't be cooled effectively by the contact with the rest if the waterblock, meaning you'd be better off without it and sticking a decent air cooled heat sink on.

Look if you consider the waterblock a huge lump of fast heat conducting copper, and keep it at 50 degrees with water flow, now connect it to the VRM's, and explain to me how its not going to suck the heat out the VRM's?

chris-82 · 3 Feb 2013 at 09:24

I personally think that paying the extra for vrm cooling is definetly worth it I had a Msi 7950 twin frozr and the main thing that held back the oc was the temp of the vrm's every time they wen above 73c it would crash I see that vrm's maybe rated to 120c but they cause crashes way before that temp (in my experience) I'd rather pay a little more now for a better cooling plate than later down the line having to buy another entirely if it becomes a issue.

Minstadave · 3 Feb 2013 at 09:26

Because copper will not conduct heat well over more than a few mm, which is why heat pipes are used in air cooling. Secondly that bit of copper over the VRMs is connected by two tiny bridges of metal to the main body of the waterblock.

dave_beast · 3 Feb 2013 at 11:15

pgi947 said:
Nice, what clocks Dave?

1200/1650mhz are my daily clocks

Got any pics? Thinking about the polished stainless top version

Mobile phone pic apologies, i've got the polished black one, it's basically a dark mirror finish.

Edit: also what readouts do you get from your vrms while under load?

VRM1 73ºc and VRM2 60ºc during a run of Heaven 3.0 at OcUK settings at the above clocks.

Gregster · 3 Feb 2013 at 11:33

I know I can hit 5.1Ghz on my CPU and it is the VRM's on the mobo stopping me from going higher. If I was to cool them, I could get more (so I am lead to believe).

I don't know enough about them but do know if they get too hot, they can't regulate the voltage properly.