MX-4 application on my CPU deteriorating in days.

People might have seen my recent thread where I repasted my CPU and GPUs and had massive temperature drops from it. People also may have seen that just a week later, my CPU temperatures were back up again.

I repasted about two days ago - using a different method this time (put a line on the CPU and let the heatsink spread it, rather than spreading it manually) - and it's happened again. Good temperatures immediately after application (~85C under 100% load at 3.6GHz), and about 48 hours later, that's already up by almost 10C. Idle temperatures are up slightly too.

It's only the CPU application that seems to be deteriorating, the GPUs are still at the same temperatures they were immediately after repaste (80C peak at 100% load, idle around 40-45C).

So, what's happening here, guys? MX-4 is supposedly rated for 8 years, yet it's not even lasting me 8 days. I've taken a look on Google, and found a couple of other threads where people saw rapid deterioration on their MX-4, but all of the cases seemed to be desktop GPUs, not mobile CPUs. Could it be something to do with it being applied directly to the CPU die rather than the heatspreaders found on desktop chops, and then deteriorating due to the very rapid fluctuations in temperature that the CPU can undergo?

I'd like to hear from anyone else who's used MX-4, if possible. I'm also likely going to have to get a different paste to try out (unless it's just that this tube is faulty or something). I'd like to see what people recommend...I know I've asked that before, but I'm asking again. My requirements are that it must strictly be non-conductive - I will not risk using even a slightly conductive thermal paste, in case I make a mess of things - and, preferably, something durable that I never have to re-apply. (That said, I'm going to become used to frequent re-application of paste at this rate...) I'm currently looking at IC Diamond, I'm just concerned about it scratching the CPU die.

Thanks guys.

 

Well i did use MX-4 in past few times, it didn't last long. ICD is better option compare to MX-4 and non-conductive also as you lookin for. Im usin ICD on CPU 3740QM OC to 4.29GHz maximum temp. can peak around 90C but remember i did OC this chip extreme level so i think 90C is fine. At 3.7GHz temp peak around 78C full load.

 

Eh. Getting a tube of ICD. Might as well, I suspect other pastes similar to MX-4 would break down in a similar manner.

 

I won't say I told you so (because I did not, plus it would not be nice even if I had,) because it is best to make these discoveries on your own. There is no substitute for personal experience and the value of exploration when it comes to this kind of thing. MX-4 and TX-2 are probably the two worst thermal pastes I have ever used before. Both of them lasted only a few days before they became worthless. I think my system probably would have run just as cool using toothpaste, LOL. As far as the claim to lasting for 8 years... telling lies for marketing purposes almost seems like the norm any more. The boldness of that claim is what originally made we want to try it. Maybe the shelf life is 8 years, LOL.

IC Diamond is really great stuff... not as good as Liquid Ultra, but I saw that you did not feel comfortable using an electrically conductive solution. I have repasted my M18xR2 and Alienware 18 and XPS desktop CPU with Liquid Ultra and it's doing a marvelous job on all 3 machines.

 

Well, there were a lot of great reviews out there for MX-4, as well as plenty of people who have found it to be great, which is why I went for it. It might perform fine on a desktop CPU with a heatspreader, for all I know. As I said, I suspect that it just can't handle the rapid fluctuations in temperature when applied directly to the CPU die, rather than to the heatspreader (which presumably wouldn't change temperatures quite so rapidly as the die itself).

I'm tempted to contact Arctic Cooling to see what they have to say about it. I'm also tempted to ask the seller for my money back (after all, the 8 year claim is clearly false), but I'm not sure if that's justification to claim that the product is truly defective and that they have to give me my money back.

I would have gone with IC Diamond in the first place, but like I said above, I was worried about it scratching and damaging the CPU die. With MX-4 getting so many good reviews, I assumed it would be almost as good thermally while being easier to apply, easier to clean, and no worries of scratching anything.

I wonder if it's worth replacing the MX-4 on my GPUs too, while I do so on my CPU? Might give it a go, see if it performs much better than MX-4 before MX-4 starts breaking down.

What would I do for applying IC Diamond on the CPU? I know their recommended method is to apply a pea-sized amount then use the heatsink to spread it, but that's presumably not ideal for the rectangular CPU die. Would a line of the stuff work, or would it be too thick/too much for that to work? Also, would a pea be too much on our CPUs and GPUs? (Compared to the significantly larger desktop CPU heatspreaders)

Thanks again guys, I appreciate the advice.

 

Do the line meathod for the CPU -a grain of rice shape across the 2/3 - 3/4 of the die.
GPU's do a pea sized blob in the middle. Pea sized meaning the form of the shape, not a 1:1 size of a pea. It's probably 1/2 - 2/3 the size of a pea.

 

This is how I do it. The pea-sized ball is right for a square IHS on a desktop CPU, slightly smaller for square GPU die, but the lidless die on the laptop CPU needs it in a line. It will spread in a circle as a pea-sized ball and the line will spread in a rectangular fashion.

While you have it apart, I would go ahead and apply IC Diamond to the GPUs. Leave them mounted in their PCE-e slots and just take the heats sinks off. It would be a shame to have to disassemble the machine again to repaste the GPUs and like you, I don't really expect that the MX-4 is going to hold up for long. The GPUs die does not get as hot, but it does not seem to be very durable.

 

Yeah, I might as well. Hopefully the 1.5g tube is enough for all three applications. It should be arriving tomorrow, I'll post back with how it went after I've repasted.

When I repasted the first time, I actually had to remove the GPUs from their sockets to be able to get the heatsink off...the thermal pads had the heatsink adhered to the card so strongly that I had to apply quite a bit of force to get it off.

I wonder if moving the cards resulted in a very strange issue I seem to be getting now. It's only happening in some games, but I get something that looks similar to screen tearing? It's hard to describe, but after a few hundred pixels from the top of the screen and a couple of hundred from the left side, everything will be shifted right by a few pixels with a black line in between. That flickers slightly between where it's meant to be and being shifted a few pixels. Vsync seems to have no effect on it (whether it's on or off).

 

Hello

I think you'll find it is nothing to do with the paste you are using. Its the placement of your heat sink. Because the die on these cpu's are rectangular shape, and the design of the heat sink you will notice that there is movement if you press on the part where the heat pipes meet the fins. And when you put the top plastic cover this puts pressure on the fins pulling the heat sink at one end. Two options here.
1. Have a look at your heat sink when you remove it, see where on the copper the paste is dry, determine at what angle the heat sink is being pulled and bend the heat pipes slightly to reduce the amount of pressure being put on it when you put the top plastic cover back on.

2. Play with the heat sink for a while start the laptop up before you put everything back together again, maybe leave the 2 back screws loose, maybe only turning once or twice, then tighten up the two from screws tight to try and pull in a way so it sits level.


Sorry if I have not explained this correctly, I hope you understand what I mean.

I have went through 3 different heat sinks and umpteen different types of paste thinking these were the problem, but the fact is, it is the heat sink lifting up slightly at one end of the CPU die, thus creating a dry patch where the pase gets squeezed out too much leaving no more paste in that one area, you will notice this if you check your temps on each core, 2 of them will shoot right up to 99% very fast while benching.

Good luck, this is trial and error, I would spend a whole day, trying different thinks, without building the whole laptop up again as this is a pain to take it apart all the time, especially with those fragile plastic clips on the keyboard surround, which tend to break after one or two removals.

regards

 

Repasted yesterday with IC Diamond. So far so good. Despite once again using what I think was probably far too much paste, the CPU now takes 4 minutes of 100% load before the fan even kicks in to its highest speed, after which it stabilises at 80-84C. (This is at 3.6GHz, no voltage bump.) If it doesn't degrade, I'm pleased with that, so I'll wait and see if that goes anywhere.

A 15-minute FurMark burn-in very slowly climbs and reaches a maximum GPU temperature of 83C, which is...not really as low as I would have liked, but not concerning. (No idea what the secondary GPU temperature was, most of the time my secondary GPU's temperature sensor just refuses to function.)

So...I'll wait and see if this application degrades. If it doesn't, I'm happy. Perhaps I'll try The Witcher 2 a little later and see what that does to temperatures (if it's not still repeatedly crashing with crossfire on for no apparent reason...)

I doubt that's what has been happening here. With MX-4 I didn't see two cores shooting up to very high temperatures after a few days - ALL four cores had worse thermals. Core 0 on my CPU is consistently 10C lower than the rest (at 100% load), but that has always been the case whether on stock paste, MX-4, and now IC Diamond. The temperature spread between the cores was the same, whether it was a stock MX-4 application or one that had degraded after a week.

 

Typically if a thermal paste goes bad that quick then the heatsink mount is not quite right and is moving around.

I used MX-4 for up to a year with no appreciable change. I tend to repaste often so that's about as long as I get to test them.

A stiffer paste will tolerate more movement in this case.

 

This is true. The stock Dell TIM and IC Diamond are both very durable and they are both very stiff. Noctua NT-H1 works good when it is first applied, but the viscosity is so thin that it does not hold up. In the case of low viscosity just the constant pressure between the heat sink and die might be enough to degrade the effectiveness as it gradually oozes or "pumps" out, but movement in a laptop chassis (especially those with more flex) makes low viscosity paste even less durable and more susceptible to "pump-out" problems.

 

How could I avoid that problem, then? Does it mean that I didn't have the heatsink tightened enough?

I'm not really sure what I could have done differently. It would explain the problem, and explains why I don't see it on the GPUs with their different heatsinks. I saw it on two applications of MX-4, and I'm now seeing it with IC Diamond as well. Load temps are a few degrees higher (about 3-4 degrees on average, rough estimate) than they were a couple of days ago, and there's far less time before the fan ramps up to maximum when I go from idle to 100% load. It's nowhere near as bad as MX-4 was, but still concerning.

What exactly do I need to be doing to prevent pump out? There must be SOMETHING I'm doing wrong here.

 

That points to a problem with the screws or plate.

 

Is there anything more specific you could advise? Do you mean that the screws or plate are damaged, or that I haven't tightened it enough/tightened too much/not aligned properly?

 

There is no easy answer here as I am still experimenting with mine. Thought I had it sorted, but my temps are shooting up again. It takes a few days for this to happen, and that is the most frustrating thing. I am sick of tearing this thing down!!

I wouldn't recommend tightening the screws all the way down and I think they still need a little spring tension to work properly.

I still think it has to do with the palm rest base pushing down on the fins/heat pipe and putting extra tension on the back end of the cpu therefore lifting up the front of the cpu heat sink because when I take off there is always a dry patch on the cpu die 2/3's towards the back. This tells me my heat sink is still not sitting level on the cpu die.

I am ready for biting the bullet and just buy an new M18!!! and just keep my 2 780m's and put these into it when it arrives and get rid of the M18x for good!!!

 

You know, I had two M18xR1 and now this M18xR2 and never have experienced this problem you are referring to with the heat sinks. And, I haven't seen other folks complaining about it. It almost seems like maybe a different choice in thermal paste is in order. It probably has more to do with the thermal paste than you might think. What you are describing with temps shooting up after a few day is exactly what I have experienced with inferior pastes like AS5, Ceramique, Shin Etsu, MX-4 and TX-2, and to a somewhat lesser extent NT-H1... but, with IC Diamond and Liquid Ultra I have never had issues like this. I have never needed to monkey around with or finesse the tightening sequence, worry with misalignment, the palm rest putting pressure on the heat sink where it should not be, or anything like that. Perhaps the bare spot on the die you are seeing is evidence of "pump out," which is also more common with inferior pastes. If there is some minor misalignment the thicker and more durable paste might take care of that situation. Whatever the case may be, there seems to be something specific (unique) to your machine or something with the paste, paste application method, or something else that we are not sure of, but it's not normal or common to the product itself.

I think you might end up regretting getting the 18 as a replacement for the M18x. Keeping the Haswell CPU cool is more of chore, so in that regarding you would be jumping from the frying pan to the fire.

 

Well, I'll leave it be for now and see if temperatures get any worse. If it stays below 90C, then I'll be happy...if it doesn't, I'll probably tear it down (again...getting sick of this) and try to figure out what's going wrong by looking at the paste residue.

 

I was responding to rdalcroft and the experience with heat sink problems that he was sharing with us.

I think you will end up being fine with IC Diamond. I suspect MX-4 was the cause of your issues since I had similar rotten luck with it, and so have others in our M18x Community. It's just not a very good thermal paste.

 

Well, it seems that I've got similar problems. It's worse again today, the CPU is now touching 90C within 40 seconds of running OCCT and peaking at 91C, averaging about 89C. This is up from averaging 83-84C and taking four minutes to hit 87C where the CPU fan kicks up to the highest setting, which I had a couple of hours after repasting. Less than 72 hours since I've repasted and the average temperature is up by 5C, with the fan hitting the highest state within about 25 seconds rather than four minutes.

Something is badly, badly wrong with the way I'm aligning or screwing in the CPU heatsink, it seems. I just wish I could tell what...it doesn't seem like something that you could really screw up very easily in such a way to cause results like this. I didn't see it degrading like this with the stock Dell paste (at least, so far as I remember, but I never kept a record of what the temperatures were immediately after the triple pipe heatsink was installed).

 

That's warm, but not hot. Your TJ max is 105° so your peak is about 15°C under that. Also understand that programs like OCCT and Prime95 induce temps that are artificially high and it is unlikely you will see similar temps doing "normal" stuff. I never use OCCT (or Furmark) because they induce unreasonable stress that doesn't really help me with anything. Honestly, I don't see a huge problem with 90°C running OCCT. Even XTU benchmark/stress test creates more heat than you would encounter during normal use.

Try resetting BIOS defaults and see if CPU temps improve. Maybe something with the CPU power settings is messed up and causing it to run a little on the warmer side. Once in a while the BIOS settings can get corrupted. It might not help, but it certainly cannot hurt. Do you have the original stock QM heat sink or an XM 3-pipe? That would be worth buying if you don't already have one.

 

I know it's safe, but also far from ideal. Running cooler also means running quieter, which is always nice.

I understand that OCCT's load is unrealistic for anything else I'd hammer the CPU with (i.e. games), but the simple fact is that three days ago, when I applied IC Diamond, it took 4 minutes of OCCT before the fan even hit its highest speed, and it stabilised at a temperature of 82-84C on the hottest core. Now it takes about 30 seconds for the fan to hit its highest speed, and it stabilises at 89-91C. And it could just keep on getting worse, as I observed exactly the same thing happening with MX-4.

Nothing's changed in BIOS settings, I've checked.

I'm running the XM 3-pipe, and I'm damned sure that it should be getting better temperatures than this with IC Diamond. Keep in mind that I'm running at 3.6GHz with no voltage bump, far less than what some of you guys are running at, yet I still hear of others getting peak temps in the 80s while running above 4GHz.

If it's any worse over the weekend, I'll open it up and take a look. And I'll take pictures this time.

 

What are your ambient room temps, bro?

Just to make sure I should ask, your CPU heat sink does not have a thermal pad on it like shown in the photo below, does it? If it does, take it off and leave it off. You do not want that on there. This used to cause problems when they were shipping with that thermal pad in place.

 

Ambient is 'whatever I feel comfortable at'. It does vary and I haven't measured it, but it's safe to assume that it's somewhere in the region of 22-25C at any point. Changes in ambient are definitely not enough to explain the difference I'm seeing after a few days, definitely not after seeing it on three applications and two different pastes.

I never noticed a thermal pad on the heatsink, I'll double check on my next inevitable teardown. (I'm assuming you're talking about the massive brown crap under the contact area, and not the black border around the copper?)

 

Right... where that brown pad is (not the black border) should be bare aluminum. They used to ship with a thermal pad there and they stopped doing it because it caused overheating. It should be bare and the surface mounted components on the motherboard above that part of the CPU heatsink should also be bare.

There has to be a simple explanation. Maybe your heat sink is deformed or it starting to lose some of the gas inside of the copper tubes. Whatever the case is, it shouldn't be too hard to remedy... there are only a couple of options other than paste... bad heat sink or something goofy gone wrong with the CPU. The latter is highly unlikely. Heat sinks can go bad, so that would be my next suspicion. Given your machine used to not run too hot and now it does even after repasting, I am starting to think maybe your heat sink is bad. If you still have the original part, it might be worth experimenting to see if it cools better than your 3-pipe. If it does, then that is a dead giveaway the 3-pipe needs replacement.

 

I'm fairly sure that there wasn't a thermal pad there. Actually, I'm almost certain, I would have noticed and remembered. But I will double check if/when I open it up again.

If there was a problem with the heatpipes, then I don't think a repaste would alleviate the problem temporarily. It might be deformed, but I don't think there was anything up with the base plate. I'll take a good luck at it the next time I open it, maybe photograph it a bit to see if you guys can see anything wrong.

I don't have my original heatsink, unfortunately. I paid Dell for the heatsink and they performed the upgrade when they came to replace my screen assembly due to a defect, the technician took away the old heatsink.

Thinking back, I was hitting 100C after a couple of minutes at 100% load with the stock paste before I applied MX-4 for the first time. Unfortunately I can't remember if THAT degraded over time from the original application, I should have kept records, but thinking back, I think I was fairly happy with the temperatures immediately after the new heatsink was installed...and I doubt that the triple pipe heatsink would work very well on the XM if it was hitting 100C with my 3720QM at 3.6GHz. I'm starting to lean towards a heatsink deformity as being the cause.

I'll try another repaste over the weekend, and this time I'll photograph everything to see if you guys can spot anything I'm doing wrong, including the old paste and my new application before I reinstall the heatsink.

Just to try and cross off possible causes...could using too much paste cause an issue like this? (I'm certain I'm not using too little) Could overtightening or undertightening the screws cause problems?

What would I do if we do suspect that it's the CPU heatsink, should I call up Dell and query it? I'm not sure if they'd replace the heatsink just because I say my CPU is running a little toasty, I doubt they'd agree to replace it under warranty if I admit to repasting it, and as for buying another one, I'm unfortunately currently very short on cash (short enough not to be able to afford the 24 GBP or so that a new triple pipe heatsink would cost) and won't have enough expendable cash to buy a new one until mid to late April.

 

Over-tightening can cause deformation, but that's not very likely based on the way it is designed. The screws bottom out before the spring is totally collapsed to prevent this from occurring.

The tubes have a gas inside that can leak out over time if there is a pinhole in the solder. That could be what is happening.

They may cover under warranty. I've seen them do it many times. Upgrade parts purchased directly from Dell/Alienware are generally covered for the duration of your original warranty. I'd say there is a good chance they'll cover it. You repasting it won't affect that, and if nothing else you have taken an appropriate step in troubleshooting by having tried.

Your 3720QM should not be getting that hot. You're running stock power settings, so excessive voltage is not causing it. So, with neither MX-4 nor ICD curing the problem that leaves only the heat sink in my mind as being the problem at point.

Call them... I bet they take care of it for you under warranty.

 

I have found one way to get my R1 screaming in pain from heat even with ICD. Running Skyrim in Ultra settings with about 30 plugin mods to enhance the graphics even further with my 780Ms at PT 146% with an offset of 128/307 and my 2960XM @ 4.1GHz. After playing Skyrim for about 30 minutes under these conditions my R1 sounds like a jet engine with my 780Ms sitting between 90% to 98% usage at near 90C temps on both GPUs. With all those mods, though, man does Skyrim look amazing! Really presses my rig to its absolute limits and the FPS stays fluid 90% of the time except in the most complex of scenes.

 

Jeez. 90 degree temps?! Is it summer where you are ?

 

Pft pt of 146%. I've been up to 180%

 

75°F ambient room temp. Only Skyrim with all the mods installed running my GPUs at almost 100% for 30 minutes and up get my temps near 90°C. Everything else stays at 81°C or less depending on the load.

 

Not trying to impress with numbers. Just pointing out that certain scenarios are going to heat your laptop up no matter what. Real Vision ENB plus numerous other Skyrim mods on top of that tax even the best of systems over a sustained time frame.

 

You might appreciate a video I did with ENB and multiple mods. I'm still surprised your temps get that high even at that load and OC. My temps never go past 77degrees under very similar conditions. Although my ambient temps are a fair bit lower right now (let's see when it's summer here). Yes modded Skyrim is a monster isn't it ?!

 

I would be interested to see your plugins and your enblocal.ini file. Here are mine. Maybe I will try yours and see what the results are. I have repasted 3 times now and I think I am done for awhile.

 

I would highly recommend getting some pressure indicating film (they're pretty good on giving out free samples, and the sample film is large enough for multiple uses if you cut it up) and see if there are any areas of bad contact with the heatsink.

Just ti give you an idea of what it would look like:
View attachment 109211 View attachment 109212

Red = area of good contact (contact pressure >28 psi)

As you can see the heatsink used in the left picture has pretty good contact with the CPU die, with the only deadspots at the corners. The one on the right however seems to suffer from a concave surface, since peripheral contact is good but contact along the center is very spotty.

 

It's good stuff, I am tempted to get some.

 

Perhaps PM if you before we get anymore off topic - Skyrim mods could even be thread worthy - I dunno, but here's all 50 of them for ya:

 

This is bizarre, temps are back down again today. Maybe it was just a warmer day when I was getting 90+C, but now it's back down to averaging 85C. Odd. I'll keep an eye on it and see how it changes, hopefully I won't need to repaste after all.

If I do though, I'll take a look at that pressure film. Looks like it'd be good to diagnose the issue.

 

Hello, I have the exact problems that you described but I have an Asus G750JZ.
I applied (2 times now) Arctic MX4 on both CPU and GPU and it degraded in about 4-5 days. When it was "fresh" pasted I had 70 degrees C on CPU and 85 degrees C on GPU and now temps are 83-84 degrees on CPU and 85 on GPU. So its kinda funny and wierd that GPU temps are the same but CPU degrades. Also worth mentioning that I had the same GPU temps with the stock Asus paste.
The application method doesn't seem to count since 1st I tried "spread" and 2nd I tried "line".

I am not sure what to use because theoretically my lappy is still on warranty (no seals since they replaced my cooling system in service and they did not apply new ones ), so Liquid Ultra is a no go, since it may corrode or atleast leave marks and IC Diamond will probably scratch the dies. I am tempted to order Gelid CG-Extreme but I am not 100% sure.

 

Gelid and IC Diamond are both excellent thermal pastes and IC Diamond is superior to MX-4, TX-2, NT-H1 and AS5 in my testing. If you really want to control your temps more effectively, go with Liquid Ultra.

Have a look at this for more insight: [Death Match] Thermal Paste Showdown: IC Diamond vs Noctua NT-H1 vs Liquid Ultra

You may also want to look at the contact between your heat sink and die. If they are not mated well you are going to have thermal management problems. The Asus G-series has never been noted for good cooling, so there is probably some room for improvement in this aspect of design. You can test this with pressure sensitive film. The third image in the first album in this post shows an example of that.

 

Well, honestly reading your post about Liquid Ultra confirmed my fears. It does indeed corrode the copper, or atleast leaves clear marks on the heatsink. So that's not an option since my laptop still has warranty and with today's quality of electronics...it's good to have some insurance. Honestly, the only reason I would consider Liquid Ultra is the fact that it doesn't seem to degrade. The temperature differences vs. a normal TIM is not worthed (ofc if the normal TIM doesn't degrade) for me since I have no plans to overclock my CPU (4700HQ is junk in overclocking anyway ).

And regarding the contact, idk...it seems fine honestly. CPU the temps are fine on the initial application and the TIM mark looks ok. GPU temps are not so great and the fact that the TIM doesn't seem to matter makes me wonder. No difference between the stock Asus paste and MX4, but oddly enough no temp difference due degrading. What's more wierd is the fact that the paste on the GPU was kinda dry, opposed to the CPU where it seemed to have perfect consistency and great TIM mark. What I noticed on the GPU that the pressure of the cooling system might actually be too high since the PCB seems to be curved abit.

Also, what do you think about the GPU thermal pads. The stock ones suck...they basically transformed into a thick sticky paste so I am unsure what to do. Replace them with 0.5mm pads (either junpus jp-p600 or Phobya Thermalpad XT) or just use TIM. I saw that someone on this forum just used normal TIM. You can see pics here - http://forum.notebookreview.com/asus-gaming-notebook-forum/726296-asus-g750-disassembly.html On the top post is normal TIM applied and lower on the page you can see how the stock one looks like.

 

You should never need to use paste instead of pads. It gets so messy.

 

Whoa! That looks like a mess in the making. I don't recommend using thermal paste where a pad would normally go in most circumstances, and especially to the extent in that example. That's probably going to turn into a nightmare, LOL. Not only is it going to make a terrible mess, the air gap will be too wide for paste to be effective. Pads are used in the places they are for good reasons. Speaking of warranty concerns... that is definitely a good example of one.

The difference in temps between Liquid Ultra and anything else is nothing short of phenomenal. Maybe not that big a deal for someone that does not overclock, but for anyone that does it means the difference between managing the temperatures or not being able to overclock due to thermal throttling and thermal shutdowns. If you looked at the screen shots I posted in the first link you would see that the difference is incredible.

Liquid Ultra does not corrode copper. It does eat up aluminum. Liquid Ultra is silver and what you see is silver metal that is in pores of copper, so there is a contrast in color. This is not harmful. In fact, I think it is more likely helpful because it increases the density of the copper plate and should have a positive effect on thermal conductivity even using a different paste later. I'm not sure why this would have any effect on your warranty at all. If the warranty allows you to disassemble the machine and apply new thermal paste, a square patch of copper taking on a silver-colored tone should not matter. It is common to have the copper heat plate color altered by thermal pastes. With most thermal pastes it is usually darker stains rather than a silver color.

I suspect the problem with the MX-4 is a combination of bad fit and "pump-out" of the paste. If the two surfaces are not flush where they come together, with even pressure on all 4 corners, the "pump-out" problem will be worsened. The viscosity may also be part of the problem. A very thick paste like IC Diamond offers more resistance to this. Some people complain about thick thermal paste being "too difficult" to apply, but this is advantageous for durability. I found a number of pastes that I have used were not very durable because they quickly ended up squished out onto the CPU chip instead of remaining in place on the contact surfaces between the die and heat sink. I applied MX-4 twice and in both cases I began having thermal management problems with overclocking within a matter of days, so you're not alone in that aspect of the experience. The only difference is that you are not overclocking. I ended up throwing the unused portion of the MX-4 syringe into the garbage because it did not meet my expectations.

The only way you can accurately determine the quality of the contact between the heat sink and die is to measure it with something like pressure-sensitive film. There is no way to accurately "eyeball" the fit or judge based on how it feels. Increasing pressure is less effective than improving contact. Increasing pressure may help to minimize the consequences of poor fit by forcing the two surfaces into mating together, but if the fit is really poor increasing pressure won't be enough and may actually cause physical damage to the processor or PCB if the amount of pressure applied is excessive.

If the idea of using Liquid Ultra has you concerned, I would highly recommend IC Diamond. In my opinion it is the best non-metallic thermal paste. On your stock-clocked CPU it should last a very long time and be very effective as controlling the temps. The pads you mentioned are good products and there is no reason not to use them if they are not overpriced and readily available to you. I use Fujipoly Extreme pads from FrozenCPU.com.

 

Well, I heard that Liquid Ultra is made out of Gallium not Silver and it does corrode the copper aswell, but slower. I don't want to argue here but I saw some tests from people that I really trust (lab501.ro)
As for warranty...well technically I am not allowed to open up the laptop and such (except HDD and RAM area) but my laptop was sent to service for cooling system replacement and they did not put any seals afterwards

I am not sure what to order...since neither Gelid GC-Extreme and IC Diamond are available where I live(Romania). Regarding IC Diamond I have my concerns, mainly about die scratching. Other available TIM's that are worth mentioning that I can buy are: Antec Formula 6 or 7, Prolimatech PK-2 or PK-3, Dimastech HTX-EE, JunPus JP-D9000 or JP-DX1, Phobya NanoGrease Extreme or Xigmatec XI-3 HDT.

 

Yes, it is Gallium. Did someone post that it was silver? (I mentioned silver as I was referring to a color, but never intended to communicate that it was made of silver.) It hasn't corroded my copper for 6 months and shows no evidence of corrosion, so I'm not too worried about it.

User Manual: Coollaboratory Liquid Ultra

I have been using IC Diamond for years with good results. I haven't had problems with die scratches from using it, but I am careful about how I remove it. I soften it with alcohol and it comes off easily without scratching. I have seen people complain about it causing scratches, and since I have not experienced that I think it may have something to do with how they handled it. If they tried to scrub it off without dissolving it first, then it is certainly conceivable that it might happen.

IC Diamond - Removal Instructions (also shows photos addressing the concerns about scratches)

I would still recommend checking the fit of the heat sink carefully, as no paste will be effective if the heat sink is not sitting properly on the die.

IC Diamond Troubleshooting - Contact and Pressure

Fujifilm Prescale® Pressure Indicating Films

 

Ok, well the problem is that I cannot seem to find pressure film in my country. But I highly doubt that pressure might be the problem. If the heatsink wouldn't sit properly on the die, temps would be bad from the 1st minute the paste is applied.

I guess I will go ahead and order IC Diamond and JunPus thermalpads. Do you think 1.5 grams of paste is enough for CPU + GPU? Also is a 100x100mm thermalpad (0.5mm thick) enough for GPU's memory chips and VRM?
For IC Diamond application what method should I use? Spread? Or thin line for CPU and pea size drop in center for GPU?

 

Yes, one tube of ICD7 should be enough, a tube of ICD24 is enough for multiple repastes. The 100x100 sheet should be more than adequate pad to cover everything. I cannot confirm if 0.5mm is correct for your application or whether multiple thicknesses are needed. You would need to inspect and measure the pads to find out if they are all the same and what thickness is required.

Use the thin line application for best results. See the photos below and the comments demonstrating the effectiveness of the thin line approach.

<iframe class='imgur-album' width='100%' height="750" frameborder='0' src="http://imgur.com/a/Rkmto/embed"></iframe>

 

The issue with MX-4 is that it's simply not viscous enough and easily pumps out under sustained periods of heat. When freshly applied MX-4 is as good as any non-metallic paste. What I've observed however is that the temps gradually deteriorate over a period of about a month, even if I'm not doing anything terribly demanding. At the end of the one month period, temps are usually 3-4C higher than what I started with. If I go through a period of heavy usage (gaming, computation etc), then the temps can start going up in as soon as 2 weeks.

Thankfully once temps have climbed by 3-4C it appears to level off and hold there for a couple more months, though the longest I've gone without repasting is about 3 months, so I really couldn't tell you how long those "already deteriorated" temps would hold up.

 

A tad too much paste in the pic but it will be much better than the stock application.

 

If you look at how it spread out in the last photo it just barely went past the edge of the die, so it is exactly the amount needed for this particular machine. It takes a little less on the M18xR2 to get the same coverage as what is required on the Alienware 18. This will vary by the amount of pressure the heat sink applies to the die, (less pressure will spread less,) and the viscosity of the paste being used (thin paste tends to pump out more that a thick paste like ICD). The right amount for adequate coverage on a particular machine will be discovered by repetition. Too much makes a huge mess and too little doesn't work well. Had that been NT-H1, for example, it would have been too much with some wasted paste being puked out the sides.