PDA

View Full Version : Insane GPU & CPU Temps in G74S



TheDoci
11-20-2015, 12:41 AM
Well after using my G74S as my workstation ever since my main Worksation had a unfortunate run in with the force of Gravity, and the very impressive force of my cousins stupidity.
I had my G74S for about 4 years now, (if i recall correctly, i bought it sometime in early 2012. It had been my backup / mobile system for a while (also my lanparty system). But about 1 1/2 years ago my main rig was murdered, and i switched to the G74 for my work / study.

I have been using very CPU and GPU intensive tasks, not including gaming on it.
A few months ago, i noticed the fans where spinning a bit loudly, but i chalked it up to the summer temperatures. Well now, that the ambient temps are down, i though the fans would be quieter as well. But that was not the case. About 1/2 year ago the GPU temps where at 61C, now they are at a insane 85 to 90C at idle with ambient temps of 22C. The CPU is the same story.
The keyboard is also very hot.

The fun thing is, the temps where at about 40C or so when i got it. I am not sure what has happened to it. Could it be the Thermal Compound going bad, or could it be dust in the heat sinks?

I am currently tempted to just open it up, and check the heat sinks (and clean it if needed), but i dont feel like taking the laptop apart.

Since i have some CPU and GPU intensive work coming up soon, i would not want to have my CPU run at over 100C. Or use LN2 to make it work.

Has anyone else have this issue?
If so, what was the solution.

Also if its the thermal compound, do you guys have any recommendations for good compound to use, my favorite was ZALMAN ZM STG1, but i can no longer get it.

Or should i just send the laptop in to asus (even though i no longer have warenty on it) and have them deal with the problems?

Also, yes there is air comming out of the vents.

Clintlgm
11-20-2015, 04:38 AM
Yes it could be just dust accumulation, and you could try to clean it out. you'll have to disassemble to actually get to the heat sinks etc. and if they are not that clogged up. you almost there to go ahead and repaste. It seem that IC Diamond is the best past for notebooks these days. Be sure to read the instruction and follow them for a good pasting.
I think if I were to send it anywhere it might be to a custom vendor. or if you have a good local shop to do the repasting. If you do decide to do it yourself Goggle the disassembly there will be some video's that will show you how. Its not like a desktop things are more fragile and not straight forward. hidden screws etc.
You issue does sound like the pasting has failed.

TheDoci
11-20-2015, 11:38 AM
I have repasted a laptop CPU Before, but that was almost 4 years ago, and the laptop was not all to good / expensive (also i got it for free, since it was basically broken).
But back then i used the Zahlman stuff i mentioned earlier. (also the dis-assembly was a lot easier then what i have seen with the G74). I will however give it a try, and order the IC Diamond thermal compound (i actually had some a while ago, but it has dried up).


Also, fascinating observation, today, the GPU and CPU are only at 65C Idle (but i do have a lot less stuff going on, since i just booted it a hour or so ago).
I am unsure what caused the temps to soar as high as they did yesterday, but i will at least clean out the heat sinks and fans.

Clintlgm
11-20-2015, 03:32 PM
Yes on the IC Diamond it is very thick. A trick we use is to hold the tube in some hot water that will thin it out. It is also very important to use a Pea size drop in the center of the CPU and GPU do not spread it out. Tightening down evenly on the clamp screws. Look on there website for these directions very important. Your Idle temps should be no higher than low to mid 40's C @23C Ambient.
Temp will clime as you start using it. no mystery there. Cleaning may be enough if you find the heat sinks clogged with a blanket of dust, if you get it apart and don't see a lot of clogging probably would be best to repaste while you have it apart.

Almighty1
11-20-2015, 09:07 PM
Your issue might be related to this:
https://rog.asus.com/forum/showthread.php?3322-Sandybridge-Throttle-Issue-and-Workaround

IC Diamond causes corrosion. Gelid GC Extreme is supposedly the best. I had replaced TIM on Dell notebooks before with Arctic Silver 5 about 10 years ago and it was way better.

Korth
11-20-2015, 09:25 PM
Gelid GC Extreme is low-viscosity (it's thin and runny), best used when mating surfaces are interlocking or have very tight contact tolerances. It cooks/dries out faster and sometimes need reapplication (on very hot parts) every few months. It has excellent performance and seems to have the best (or "least worst") performance in subzero overclocking.

Prolimatech PK-2 is mid-viscosity (sort of average), best overall TIM for most parts in my opinion. It adheres really well and can last many months or a few years before dryout. It also has excellent performance. PK-3 is supposedly even better (never tried it because my big jar of PK-2 is still half-full, lol).

Arctic Silver 5 is mid-/high-viscosity (kinda thick and pasty), it's a better choice on large-gap, planar, or slightly convex/concave surfaces (like LGA2011-3 processors) because it stays where you put it and maintains great performance pretty much forever (the stuff takes years to cook off). It's also really cheap. I sometimes even use it as a thermally-conductive glue, lol.

IC Diamond contains abrasive microparticles which promote metal corrosion and - after repeated applications - can actually scratch off part markings. Many people swear by it, I've never used it.

Phobya He-Grease and Tuniq TX-4 are about as good (some say better) than PK-3 or AS5, but again I've never tried them.

The reality is that any premium TIM will outperform any cheap TIM and the actual measurable differences are nearly insignificant among top performers - each brand has passionate diehards and vehement haters, I suspect conflicting benchmark results usually have more to do with variances in application method or quirky heat sources than with substantially differing TIM characteristics - my personal experience is that the main difference between TIMs is their viscosity/adhesion parameters, no single TIM can be absolutely perfect for every possible application because each is better suited for a different kind of interface geometry.

You can be assured that any properly-applied premium TIM product will outperform that cheap pink goop Asus put into your machine.

Clintlgm
11-20-2015, 11:36 PM
One comment virtually evey custom notebook vendor offers a repaste using IC Diamond on all there notebooks sold. notebooks are not desktops an have different requirements. My 4790K idles in the 20'' C I've never gotten it over 60C even with burn in test. IC Diamond it exactly that Ground diamond with a bonding agent very thick when cool, hot water Warming it up makes it more fluid for installation.

TheDoci
11-21-2015, 11:01 AM
Hm, very interesting,

I still have about 1/2 a tube of Artic MX-2 lying around, i don't know the dry out time of it, i use it on my main rig, and reapply it every year.
If the dry out time is decent, then i might use it, instead of arctic silver, or IC Diamond.

Almighty1
11-21-2015, 10:26 PM
I always thought it was the AS5 that was the one that dries out and requires application as that's what I read. I still have like 10 tubes of AS5 which are over 7 years old and I noticed that it's runny as it's stored with the tip down, do I just discard the thin part and use only the thick part? How does one use AS5 as a thermally conductive glue? So with IC Diamond, it's actually the diamond particles that causes the corrosion? Is IC Diamond similar to how thick Shin Etsu 751? What should one use for a LG G4 Smartphone as it has thermal grease and I heard some people replaced it with AS5 and it works better than stock.

I was looking at the chart here a week ago:
http://www.play3r.net/reviews/cooling/best-thermal-paste-2015-update/

They seem to only cover the PK1 but not the PK2 and PK3.

Korth
11-22-2015, 02:25 AM
Old TIMs can chemically separate over long storage times. And consistency can vary from batch to batch. It's a good idea to warm the TIM up (float the tube inside some heated water) for a few minutes then - if possible - to mix the contents of the entire container to an even consistency, like you would do with a can of old paint. The diluted runny part is just as valuable as the sludgy thick precipate, and the TIM will only perform as it was engineered to do if all of the substance that comes out of the bottle is the same as it was when it was when they put it into the bottle.

I actually tend to work with TIM out of small resealable containers, not squeezed out of tubes and syringes directly onto the parts. Most of my CPU and GPU applications involve putting the TIM onto a razor blade and scraping it across both mating surfaces. I want an even application with the minimal amount of TIM I can use to interface the gap (because the TIM is a far better thermal conductor than air/vacuum but it's a far, far worse thermal conductor than heatsink metals) so I scrape it down as fine as I can.

Actual application method and precise thermal properties of the TIM are not as critical as most people think unless you're shooting for elite extreme momentary overclock records. Proper quantity of TIM is far more important. Proper coverage of TIM is far more important.

I use AS5 as a "thermal adhesive" in the sense that it's thick and sticky (and cheap) enough to mechanically and thermally secure little parts together, especially if the parts are then also bolted or fastened down some other way. It's not a strong glue, lol, it won't hold things together on its own. Sometimes a normal TIM is insufficient to fill the gap while a foamy thermal pad is too bulky and a graphite strip isn't workable and a heat application of a phase-alloy or HOPG would damage components.

The microparticles in IC Diamond do not directly corrode metals. They abrade the surface (like a fine sandpaper) and the exposed metal is then exposed to atmospheric oxygen. Corrosion is then often accelerated by the eutectic and galvanic action of dissimilar metals held in a nonequilibrium state of sustained heat and pressure.

Shin Etsu formulas are, in my experience, remarkably similar to Gelic GC Extreme. I haven't used them a lot but I would expect them to work best in similar applications and have similar general properties.

You'll find lots of comprehensive TIM testing and comparisons online if you look hard enough. You'll also find the same handful of products are always listed as the top performers. Some of the testing "laboratories" do an excellent job and introduce proper controls into every detail - they'll carefully describe their testing components, testing conditions, application methods, etc. But none measure long-term performance, very few bother to even wait out full cure durations to obtain peak performance. My interest in TIMs is not about building a machine that'll break elite records for one day - it's about installing a TIM that'll perform for months or years at a time. Reviewers get to rip apart their hardware and play with new toys every week, I prefer to service computers maybe a couple times per year and not have to remount heatsinks until I install hardware upgrades.

Factories use cheap TIM, they probably buy big buckets of it and squirt it out of robotic nozzles, they're always interested in minimizing defects but chances are that defective units will pass QC diagnostics well enough then suffer thermal breakdown after a few months real use. As stated above, any premium TIM will easily outperform whatever cheap gunk the factory installed - in your ROG laptop, smartphone, TV box, whatever.

Almighty1
11-22-2015, 07:25 PM
Thanks for the pointers Korth. I was asking about redoing the TIM in the LG G4 phone because the phone overheats and people have replaced the stock TIM in the LG G3 and G4 which gave better temps as the phone runs extremely hot on a Snapdragon 808 underclocked CPU.

As for IC Diamond, does it only abrade the surface at initial application since I couldn't see it doing that if the HS is mated permanently to the CPU as I thought some form of movement would be needed.

There was a TIM I used called Nanotherm PCM+ which was a phase changing TIM that solids when cold and liquids when hot that was later acquired by Arctic Silver. I remember that caused corrosion on the bottom of the heatsink in my experience.

As for Shin Etsu, I remember a friend had a Dell XPS Pentium 4 Northwood CPU and taking out the HSF would pull the entire CPU out as it acted like the CPU was permanently glued to the Intel stock heatsink fan, I believe that was the Shin-Etsu G751 as I still have a tube of the stuff here from Corsair when they made the first Water Cooling system with Delco in a external solution. I thought Shin-Etsu would be more of the high-viscosity as it is very thick compared to AS5.

You're right that tests online only talks about days and not long-term which is why you can't trust it too much.

Speaking about TIM's and IC Diamond, IC Diamond mentions that they are better than Silver which is what's used in AS5 but what does Prolimatech use for PK-1, PK-2 and PK-3 that makes it perform better than AS5 as I don't think it's silver and didn't think there was anything better than silver other than diamond.

As far as the AS5 as a "thermal adhesive", I did use AS1 or AS2 in a Motorola Flip Cordless home phone, the ones they made in the early 1990's when the flip was not working correctly for the on-hook/off-hook function and it did fix the problem. I remember in the 1990s with the Intel Pentium CPU, the HSF was glued onto the CPU with some type of thermal adhesive, just not sure what they had used except it was hard as epoxy.

Korth
11-22-2015, 08:31 PM
lol, I'm trying to say the specific brand and properties of specific TIMs aren't that critical unless you're trying to engineer something for ultimate peak performance. The premium TIMs are all consistently superior, the cheap TIMs are all consistently adequate. I understand that you only want to use the best of the best of everything in your machinery but if you truly want only the most extreme edge in thermal performance you'll have to clearly define your expectations to narrow down choices, you'll probably also even use multiple TIMs on any given motherboard, laptop, or mobile device because each is best at particular types of part coverage.

Just buy yourself any premium TIM and you won't be disappointed, they're all superior to the factory-installed gunk. Of course you'll see people reporting superior thermal efficiency after they've (properly) re-pasted their devices. Small profile devices like mobile phones have comparably low heat output but also have constraints on their ability to exhaust/radiate heat accumulation - you'd probably cool them off better with higher-efficiency battery chemistry or display panels than with a superior TIM, but every little bit helps.

I don't know the exact chemical formula for AS5, PK-3, etc - they're proprietary secrets. You'll find a few laboratory analysis reports online about various TIMs, but they're mostly concerned with qualitatively confirming/denying the presence of silver/copper/aluminum/gold/etc.

CPU and GPU heatspreaders ("IHS" plates) usually have steel or aluminum surfaces, their heatsinks/coolers usually have zinc or copper (or zinc-/copper-plated aluminum) base surfaces. Motherboard parts usually have bare metal or carbon/polymer chip packaging. Memory chips are usually sealed in blackened polymer or ceramic. In short - all these already have a layer of somewhat inefficient thermal conductor between the hot silicon and the TIM, so a TIM which is 99.98% thermally conductive is not going to really outperform one which is 98.5% thermally conductive. You can always de-lid parts and run bare silicon, but chances are you would then go for direct liquid cooling instead of a TIM/heatsink approach.

The abrasive particles in IC Diamond are measured in micron scales. Like a jeweller's rouge or optical lens polishing agent. But they're harder than the metals they're being pushed against and will scratch it. Thousands of tiny scratches normally invisible to the eye but more than large enough to promote oxidizing chemistry. I wouldn't recommend it on finely-honed parts. While pure carbon allotropes like diamond and graphene are the ultimate thermal conductors they are also highly dependent on molecular alignment, and I kinda think a statistically random distribution of diamond particles will not be uniformly-aligned enough to have measurable impact on bulk thermal conductivity - still, a lot of people swear by their IC Diamond products and it's a popular choice, give it a shot if you like. It may or may not be the ultimate TIM but it'll definitely outperform cheap Asus metal-oxide goop.

I'm not truly expert with home computers, they are just a subset of the piles of component-level electronics I work with every day. But there's nothing particularly special with gaming PCs, they follow all the same engineering rules as any other high-performance logic device.

Almighty1
11-24-2015, 03:41 PM
You're right that it's not going to be critical except one wants something that is long lasting and reliable and ofcourse performs as well. It's obvious that premium TIMs would be consistently superior except it seems AS5 seems to not perform as well as the things that came out in the last 5 years or so. You made good points that different parts have different requirements but there should be one that works well for everything.
As I mentioned before, I already have a large stock of AS5 so will use that unless there is something that is better in performance in the long run. The temps on the Qualcomm Snapdragon 808 and 810 processors seems to be higher than desktop/notebook CPUs as the temps can reach up to 105C. The battery and the display really has nothing to do with it as it's the CPU that is generating all the heat. I actually started a thread at xda-developers about the LG G4 and the discussion is pretty interesting:
http://forum.xda-developers.com/g4/help/lg-g4-thermal-compound-mod-t3255074/post63981823

No one would know the exact chemical formula for the premium TIMs but AS5 tells you they use silver particles while IC Diamond tells you they are using Diamond and even says the Diamond is superior to silver which is the next best thing after diamond, maybe they are just trying to say they are better than AS5 without mentioning the name. PK-3 doesn't seem to mention what they use that makes it better than the competition but looking around, it's actually aluminum.

I've never had problems with the IHS corroding but the copper on the Zalman Heat sink fans, I know the Nanotherm PCM+ which is a phase changing TIM did cause corrosion on the copper. Nanotherm got acquired by Arctic Silver awhile back. PCM+ performed better than AS5, probably because when the system is on, it's a liquid and when the system is off, it's a solid. It does seem like some TIMs perform better for air cooling while others perform better for liquid cooling. Was looking at this chart:
http://overclocking.guide/thermal-paste-roundup-2015-47-products-tested-with-air-cooling-and-liquid-nitrogen-ln2/6/

I know diamond is harder than metals but what I meant was is the scratches actually caused by the vibrations since you really only mount the HS to the IHS only once which is usually physically tight so there is no further movement unless it's from the vibrations. I always thought it was the liquid metal TIMs that caused corrosion.

While you might not be a expert with home computers, your knowledge of TIMs is way better than mines even though I am a expert on computers and also a Astrophysicist by training. Seems like Thermal Grizzly Kryonaut is supposedly the latest top performer.

risingTide
11-27-2015, 04:28 AM
Has anyone else have this issue?
If so, what was the solution.

Also if its the thermal compound, do you guys have any recommendations for good compound to use, my favorite was ZALMAN ZM STG1, but i can no longer get it.

Or should i just send the laptop in to asus (even though i no longer have warenty on it) and have them deal with the problems?

Also, yes there is air comming out of the vents.

Hey there. I recently repasted my G75VW that was almost 2 years old because I was facing similar issues. Both the GPU and CPU were running crazy high. I'd recommend doing it yourself if you've done it before; make sure you do some solid research first to get some good videos or pictures on the tear-down/reassembly of your specific model. Personally I found two great videos on my model, and after using some Arctic Silver 5 my temps are down 20-30C on both the CPU and GPU. The fans rarely even kick on anymore...my only regret is not repasting about 6 months earlier when it started!

I hope the best for you.

Korth
11-27-2015, 05:11 AM
Note that you can also mix TIMs together. I routinely mix some of my thin Gelid with some thicker PK-2 or AS5, this allows me to create a "custom" TIM as needed for particular parts, mostly because I want the staying power of thicker TIMs (on parts I don't want to service again for a long time) but also want the easier application of thinner TIMs (on awkward parts with teeny gaps or interlocked mating surfaces). Again, I maintain that proper quantity and coverage of the TIM far outweighs exact properties of the TIM.

Thermal stress and temperature changes affect the chemistry of semiconductors profoundly enough to alter their electrical parameters, so engineering better thermal interfaces is driven by multi-million-dollar industry - performance, reliability, and costs are all considered exhaustively. The sad secret is that companies like Arctic Silver cannot market a "very best" TIM for all computers, they market a "best overall" TIM for most computers - they could engineer product variants for each possible processor and cooler pairing, but they probably couldn't mass produce it cost-effectively or profit from selling an unwieldly confusing array of products to a somewhat non-technical niche enthusiast market.

Almighty1
11-27-2015, 08:34 PM
Interesting about mixing TIMs, never thought about that one. Never knew it was a multi-million dollar industry though. It just seems none of the other companies has the same name as Arctic Silver, maybe they have better marketing or something or more resources. Just strange they didn't create new products after the AS5.

Korth
11-27-2015, 11:36 PM
Arctic Silver is basically the most famous because they were the first company which offered a premium TIM to computer enthusiasts. It turned into a huge market - now filled with plenty of passionately argumentive zealots who will champion their chosen TIM with unfaltering fervour - so now there's a bunch of companies offering a selection of awesome TIMs. Now there's even some polyalloy fusion TIMs - stuff like Coollaboratory Ultra and Indigo Xtreme - which were unheard of outside of electrical engineering a few years ago. I expect that HOPGs and nano-layered graphenes will be available to the most crazed PC enthusiasts in upcoming years.

(Incidentally, a graphene-based product called Thermene advertised impressive specs and attracted a lot of attention this past year ... but it turned out to be vaporware, or perhaps a credit card scam.)

Back in the dark ages, many gamers unlocked and voltmodded and overclocked without ever considering that a superior TIM grade existed or that it could raise performance thresholds, that cheap no-name silicon goop from Radio Shack was good enough, lol. Then Arctic Silver arrived - and demonstrated superior performance, and pointed out that silver is a great thermal conductor. The TIM industry has always been around, but a lucrative new TIM market was suddenly born.

Almighty1
11-28-2015, 06:40 PM
Forgot about them being the first... I remember AS1 used to be sold direct... Then they came out with 2, 3 and 5. 4 was skipped because it's considered a bad luck number in Asian countries as it sounds like death. AS is sold retail even at Radio Shack. As for the polyalloy fusion TIMs which are supposedly the best thing, it's basically permanent or hard to remove. I remember during the Celeron days when it used the slot instead of the socket, people did use the white silicon goop.