Testing the NVIDIA GTX 970 with IC DIAMOND

[frostdragon]

IC-DIAMOND BEFORE AND AFTER TEST


We heard good things about MSI's gtx 970 4-g gaming graphics card, so we decided to purchase one.

We thought that we would give the the good old 'tried testing and true' IC DIAMOND Before and AFTER test.

That is with stock paste and after application of IC diamond.

We tested the graphics card with fans on auto (GPU IDLE AND FULL LOAD) and Fans on full.
and then the same again with IC DIAMOND

ALL ambient room temperatures were recorded. To obtain this we ran the graphics card for a FULL hour and then the GPU was left to cool down naturally for 30 mins. ONLY THEN where AMBIENT room and idle temperatures recorded.

ALL ambient temps were recorded by the use of a K-PROBE and voltcraft meter. GPU temps recorded by MSI after Burner and Kombustor.

Testing motherboard was MSI XPOWER 2
CPU 3960X INTEL (at stock) (water cooled cpu )

Removing the stock heatsink. PLEASE excuse the poor photographs as my main camera was at the office, whilst I was at home conducting this test, so I had to make use of my phones camera

PLEASE NOTE we only tested the GPU CORE and not the memory.

TO remove the heatsink from this card, involved the simple removal of four screws.

Only to be confronted on what looks to be the usual stock paste that is very liquid, which gives rise to concerns of reliability over time. This is worrying given the amount of watts that go through any graphics card in general.

Let us clean that GPU CORE to a nice shine shall we?

Now apply a small amount of IC DIAMOND TIM to the gpu CORE

All we have to do is perform testing.

The next post will show the before and after testing and the ambient room temps.




To be continued....... IN NEXT POST

 

[frostdragon]

testing section BEFORE ICD

TESTING SECTION

As previously shown in the above post.

We tested the graphics card with fans on auto (GPU IDLE AND FULL LOAD) and Fans on full.
and then the same again with IC DIAMOND

ALL ambient room temperatures were recorded. To obtain this we ran the graphics card for a FULL hour and then the GPU was left to cool down naturally for 30 mins. ONLY THEN where AMBIENT room and idle temperatures recorded.

ALL ambient temps were recorded by the use of a K-PROBE and voltcraft meter. GPU temps recorded by MSI after Burner and Kombustor.

Testing motherboard was MSI XPOWER 2
CPU 3960X INTEL (at stock) (water cooled cpu )

Please note that when the fans are set to AUTO They do not engage until the GPU reaches 63C


BEFORE ICD

Ambient room TEMPS 21.3C

GPU IDLE FANS AUTO

GPU IDLE FANS FULL

GPU FULL LOAD FANS AUTO

GPU FULL LOAD FANS FULL

 

[frostdragon]

tESTING AFTER ICD

TESTING WITH THE APPLICATION OF IC DIAMOND

Same procedure as above except....

AMBIENT ROOM TEMPS

23.8C

The above image also shows the importance of testing the room temperatures by the use of proper equipment. Cheap standard room thermometers are often inaccurate and can give false results.

IC DIAMOND APPLICATION

IDLE FANS AUTO

GPU IDLE FANS FULL

GPU FULL LOAD FANS AUTO

Have to retest on this one as I cannot find the file


GPU FULL LOAD FANS FULL

Conclusion and fans auto gpu full load results to arrive soon, however on average so far we are seeing a 5-6c drop in temperature all across the board. So in other words the marketing ploy that other TIm manufacturers like to inform us that all Compounds perform only one degree of each other ,so what is the point etc etc. Well given the evidence shown above this is CLEARLY NOT the case at all!

We see on average a 5-6c and in one result 7c temperature drop from stock paste. THIS is when we factor the AMBIENT ROOM TEMP differntial

However even after the evidence presented above, if you do not take our word for it, then PLEASE TRY FOR YOURSELVES! We are giving away TEN TUBES TO TRY on YOUR graphics card.

 

[frostdragon]

ICD is not electrically conductive or capacitive buddy! IT is ideal for GPUS and mobos etc for the reason.

Also the viscosity of the product is very high, thus being again ideally suited for the above and for reliability.

As for warranties.... I KNOW EVGA still allows you your warranty if you remove the heatsink, A far as MSI, I think they do also.

The main issue is this, TEMP performance is only one thing. But given hoq liquid the stock paste is then one can reasonably be concerned of performance over time. That is with the high heat of a GPU then there is a good chance the stock paste will dry out. Thus when you have a thicker paste the more reliable it is as it physcially creates a gasket seal, allowing no air to enter the joining of the surfaces.

Golden rule is this:

Thinner the compound less reliable over time
Thicker the paste is the longer it lasts before re-application.

 

[frostdragon]

This product has never scratched or stained any product fella. That is usually the marketing ploy of those whom are in the employ of certain other companies.

Let us examine this fully

OCUK has sold what around 20,000 tubes of this product? They have had ZERO returns and ZERO returns due to warranty issues either and that says it all

 

[frostdragon]

Hhehe nothing is as good as watercooling fella. IC Diamond does help but it cannot perform miracles. To be honest any company that says otherwise is being dishonest with their customers. Water cooling can achieve (on full load as much as a 30C TEMP drop) I have only seen ICD perform as much as a 10c drop on a GPU at full load and only then for the reason that it was incorrectly mounted (both times before and after)

Unlike other manufacturers we base our data from ACTUAL give away tests and the aforementioned is pretty conclusive.

 

[frostdragon]

YUP the icd will certainly help with that.

Best thing to do is after application of icd

RUN your gpu at full load for about four hours (curing time) and test after you will notice a temp drop

http://www.overclockers.co.uk/showproduct.php?prodid=TH-000-ID

 

[frostdragon]

Rtho782

You will have great temp drops.

Remember to run the gpu at full load for around four hours, let it cool down for about an hour then run and test again, you will see the temp drops on full load

 

[frostdragon]

heheh thats strange fella never heard of that one before.

As for stain on copper is it something like this?

 

[frostdragon]

Let's say I want to do my desktop CPU (1150), two 980s, and my laptop (Ivy Bridge cpu and 650M).

Is one tube going to be enough?

One small tube does around 3-4 to 1366 IHS sized cpu

So yes you should have enough

 

[frostdragon]

Darker, happens with some other pastes too, personally not anal about the backside of a cooler-as long as it completely removes from the core in case of warranty purpose-that's all I'm concerned with.

Pros-Cost, works really, really well dropping the temps considerably lower than other pastes.

Cons-Harder to apply/remove, my tube increased viscosity and hardened to the extent it was unusable over the space of a few months-despite the cap firmly replaced and stored in the same container as other pastes.

Hey there, I am curious about this. Would you like me to send you another small tube, see what happens?

regards

Frosty

 

[frostdragon]

send me a trust request please buddy!

Btw how long has the MX4 been on your GPU?

 

[frostdragon]

I'd like to try this on my msi 970, already checked with msi that removing the cooler doesn't void warranty

Good man as I have the same card good to know as the standard paste they put in it is quite worrying, as you will soon find out.

PM trust me your addy and ill get some sent out.

regards

frosty

 

[frostdragon]

That would be great thank you

I'm guessing it's a blockage, Maybe there's a clump that formed and it's blocking the nozzle as It came out perfectly last month when I used it on my 4790K

When i get the messages into my inbox i will send out accordingly

regards


Frosty

 

[frostdragon]

Should I send you my address or do you need to wait for authorisation from your higher ups ?

Heheh had to smile at that one

Please send me your address via trust message.

In addition we are ONLY sending small tubes out for those hwom are testing on GRAPHICS CARDS /LAPTOPS OR MOBOS only. The reason is that we have all our data on desktop cpu's already

 

[frostdragon]

Frostdragon; did you put a big splodge of IC paste in the centre and little dots all around? It looks like it from the pic. I think with IC diamond you need to put a line going across the centre rather than a pea shaped amount. As 555BUK says, application is the more important thing. The problem with adding additional splodges is that when they meet under compression they form a small microscopic line at the join between them.

I also think you may have applied too much paste. I used to deal with folks who thought applying a thick layer was good as it acted like a gasket but it's there to simply fill in all the holes between the mating surfaces so as to stop any (microscopic) air gaps. (I used to do this stuff for a living - you can tell!)

Thick paste isn't necessarily better than thin - depends on the heat impedence and transference of the product. Arctic Silver 5 is good because the silver molecules never form a flat surface so don't trap heat in. I suspect IC diamond is on the same principle. With the amount of paste you have there a lot will have spilled over the GPU heatsink edge and it could also form too much of a barrier. Any product that stays liquidy (technical term!) rather than forming a solid gasket over time is preferrable. Arctic Silver 5 is great in that regard with it's 200 hour burn-in time.

And I recommend using alcohol handwash to give the mating surfaces a really good finish before applying new paste (if you don't have Arcticlean or isopropyl alcohol).

Personally I'd go for TUNIQ TX-4 over IC Diamond. I did use IC-D but it's a real pig to apply properly and I was always concerned about it's spreadability and long-term performance - all the stuff in tubes got solid after two years of storage. Anyway, just my 2 pennies worth!

Nope i used Isopropyl mate dont worry. Also those little splodges were just access on the end of tube that had come out-being a yorkshireman well and true I hate waste so put the pslodges on.

Actually on CPUS with an IHD LARGER than 1366 cpu it is recommended to put a large splodge in the centre and smaller towards the corners. We recommend that for socket 2011 and 2011 pin-3

 

[frostdragon]

Will purchase a second card soon, in order to test under SLI

 

[frostdragon]

This round we are keeping the tubes to be sent out rather low, due to the low returns last time and in addition we would hate to interfere with OCUK 's sales of the product:_

Will let you all know how it goes

Frosty

 

[frostdragon]

Nice one Scotland.

Might get a bit more out of it once its gone through its 4 hours continuous burn-in (cure time) on that GPU.

 

[frostdragon]

still awaiting more to arrive from hq fella