MX-4 application on my CPU deteriorating in days.

I suppose I cheat a little and make the paste thinner in the middle, tricky to get right without fiddling with the application but it helps reduce the amount needed for full coverage.

 

brother Johnksss taught me about this problem.. and ill pass on the advice he told me to combat it..

1. you cant just tighten your heat sink down to your cpu.

2.throw the normal notion of your doing it right out the window

3. when your re-pasting before you put the laptop all back together leave the top off of your laptop and power on your computer..
open throttle stop and Wprime , run a 32m at 8 threads and look at your temps.. this is the base line..
Now start the big boy test.. and slowly adjust the screws holding your cpu down, like if you see some cores getting hot than others you adjust the tightness of the cooler..

4. As Brother Johnksss told me.. you might have half your cooler barely tight while the other half is snug.. its about finding that balance and what works for you..because no 2 laptops will be the same and none of them will tighten up the same and if you just tighten them down more than likely you will get hotter temps..

 

wait? that black thing going around is bad? all mine have it.. does it need to come off?

 

No, not the black shroud, the thermal pad. I think the problem with the thermal pad being there is that it interferes with the mating of the GPU die and heat sink. When I look at the air space, there is almost none (paper thickness or less). I believe that having the thermal pad tips the heat sink out of alignment. A tiny glob of thermal paste on top of each component might be effective, but I have not tried that. I haven't any compelling incentive to do so because I am not having any thermal management issues.

 

Do any of you use arctic silver 5 I use it on everything and it seems to get better as it gets older

 

Old news, extremely inferior paste and has a 200 hour or so burn-in period.

IC Diamond murders it across the board. It can drop your temps by up to 10C if you're in the 70-100C temperature range.

 

AS5 is what I always used before I discovered IC Diamond. I'd never want to have to go back to using it.

 

Ordered some Gelid GC Extreme, will see how it holds up against MX-4. According to this HWBot thread the Gelid GC Extreme subjectively felt far more viscous than the MX-4, so I'm hoping that this will decrease issues with pumping out under extended use. I've just about had it with having to repaste every danm month because MX-4 can't keep its **** together (pun hahaha).

Plus it seems like Gelid GC Extreme gives better temps than MX-4 anyway. I'm boycotting IC Diamond on moral grounds so that's out of the question.

 

MX-4 was terrible in my experience. MX-4 and TX-2 never lasted me more than a few days of overclocking... total waste of money buying them. I still haven't tried the Gelid GC Extreme yet. It is kind of hard to feel motivated to use anything except Liquid Ultra any more. Since I know nothing else can touch it, there's not really a good incentive to experiment with other options. What's the deal with IC Diamond? What are they doing that is immoral?

 

IC Diamond fiasco
in a nutshell

This just left a horrible taste in my mouth.

I know what you mean about Liquid Ultra, I can't see myself ever going back to regular pastes after using it on the CPU.

 

I haven't brought TIM for a long time because I brought a heap of AS5 so what is the best Tim out there now. I just read the ic diamond post from n=1 jeez I wont be using that

 

Liquid Ultra

 

Coolaboratory Liquid Pro Vs Artic Silver 5 Test results seem's to be legit

 

Spent the night goggling and I think I may have found just what I'm looking for: Shin Etsu X23-7783D

Look at the graph for "Poor Contact". This is exactly what I'm looking for, a thermal paste that it'll hold up even when contact pressure is insufficient (as is the case with the GPUs in my Clevo). The viscosity value is already quite high at 200 Pa·S, more than twice that of MX-4.

Apparently there's an even more viscous version called the X23-7921, and has both higher thermal conductivity and lower thermal resistance compared to the 7783D. According to the data sheet the 7921 has a whooping 360 Pa·S viscosity value, which is nearly 5 times that of MX-4.

If these pastes are as good as they look on paper, then they should perform better than ICD7 without the die etching issue.

 

Let me know what you think of them, I was looking at GC Extreme today myself

Sent from my HTC One_M8 using Tapatalk

 

Doesn't liquid pro perform better than ultra if looking at purely lowest temps?

Sent from my Nexus 5 using Tapatalk

 

Those clocks aren't really extreme. Liquid Ultra does it's best when you treat it mean.
I've used Artic Silver pro, which was smashed by IC7. Then Liquid Ultra came along, pointed and laughed at them both.
It's that much of a difference. Liquid Ultra, especially on the CPU is the business.

 

You guys have finally convinced me...

I just ordered a Liquid Ultra kit... I was still on the fence but Mr. Fox's review was on the Amazon page and pushed me over lol.

Hopefully I can get my 4940MX to 4.3GHz soon

 

Be sure to follow my recommendations for application. ("Paint" a thin coat onto both surfaces. Applying it to one surface and leaving the other one dry did not work well.)

 

I will do that. I'm a bit nervous because the way the chassis is designed you have to slide the heatsink over the die but it shouldn't be an issue as long as I make sure not to have any extra on either surface.

For giggles I used your settings for 4.3GHz just now, throttled halfway through an XTU benchmark (we are talking 97C with 40% throttle) but it *did* achieve 4.3GHz so if this gets the temps down I'll be extremely happy.

Sent from my HTC One_M8 using Tapatalk

 

Any of you guys have experience with the Shin Etsu pastes?

 

Having to slide the heat sink across the die instead of being able to lower the heat sink directly onto the die is a flawed design no matter what kind of paste is involved. That being the case, I think the chances of it working better are greater with Liquid Ultra, as it is probably less likely to interfere with the proper spreading of the paste. I could see a problem with a pea-sized glob or thin line application getting messed up a lot easier having to slide it across the die like that.

I have an old (probably dried out) tube of it somewhere. I used it several times several years ago with good results with 2720QM and 2760QM processors, but I have never used it in an overclocked application. A good number of thermal pastes do a fantastic job in an ordinary use scenario, so I have no idea how well it would hold up to extreme conditions.

 

Its hard to explain... You don't *have* to slide it across the die but it's very easy to do because to get it in, you have to start inserting it down and then there is a point in the plastic that lets you move it into place before setting it down. It's a pain.

I just discovered my 4940MX will do 3.9GHz on 1024M 8 threads with only 1.12v (static voltage, I don't mess with dynamic it always gives too much juice) so this chip may be a bit better than I thought, still tweaking it down. Bored and have nothing else to do.

Sent from my HTC One_M8 using Tapatalk

 

Well, I repasted with ICD7 a week ago and things seem to look good...sort of.
Ran 3dmark today and temps look decent. 70-73 degrees for CPU and 84 degrees for GPU. I hope I will be able to do a real world test this weekend aka. a BF4 multiplayer marathon
I am pleased with the CPU temps, especially because they can be improved by undervolting, but also the fact that they are on the same level as CPU temps in G750JZ reviews.
The GPU temps worry me abit since they are considerably higher. They are considered normal by my reseller so there is nothing I can do. Liquid Ultra is out of the question since risk/reward benefit is small for non-overclockers. Sadly, undervolting the GPU is not an option since there is no custom VBIOS for the 880M 4GB version.

 

You wouldn't likely be able to undervolt the 880M anyway, it's pretty close to its limits with stock voltage.

Sent from my HTC One_M8 using Tapatalk