Improving P170EM 7970M GPU cooling performance (foiled)

Would this or this work?

 

Those should be fine. If you want to get maximum benefit from thermal conductivity between components, 3M 1245 copper tape might be a good bet. It seems the corrugation cut through the adhesive and make metal-metal contact. While other tapes have an acrylic adhesive layer that is probaly not that good for thermal conductivity.

 

Heres my foil mod. How does it look? I feel like I used a bit much on the GPU fan. Also, are there any long term problems that could arise from this? Maybe the aluminum burning/melting and causing damage to the heatsink.

 

I'm not questioning the results people report, but the gaps were never that huge in any p170em/p150em and never amounted to more than say 5% of the length of the fins, the air is blowing in the right direction anyway, so it kinda doesn't make sense that closing a 3 mm gap would drop your temps by 10 celsius.

Also metal foil conducts heat, so this foil potentially transmits the heat back to the case through the fan's metal casing.

Placebo effect going on?

 

Even a small gap is going to allow a significant amount of hot air back in. It does also work with electrical tape. I haven't seen much difference between that and foil tape.

 

No it really works. Attached is a screenshot of Kombustor after 10 minutes. Before doing the foil method it would hit 88-89 at around the 9 min mark. Now it took 10 minutes and it barely broke 84 (which it hovered around for a few minutes). Not a 10 degree drop, but I don't think I even applied my foil as well as I could have so if others are doing this optimally I could believe it.

The foil transmitting the heat back into the case may be happening slightly, but sealing the gaps more than makes up for it. It's like plugging a hole in a leaky valve. It might be a small hole, but fixing it increases efficiency for the entire system.

 

I was looking at the inside of my old ASUS, and they did the same thing. It isn't a new solution. Everyone will see a different effect anyway, the gaps and alignment aren't consistent.

 

1. The problem is that if you have any leak between the fan output and radiator, you lose a lot of pressure, so the air velocity is greatly reduced through the fins. Not only do you have a leak, you have a direct and very short path right back to the intake which compounds the problem.

2. The goal of any heatsink is to transfer the heat to the airflow so it gets evacuated. The more surface area you have to achieve that, the better. Even if the heat is transfered upstream to the fan intake, then that's all good, as long heat is transfered and it's airborne by any means and gets evacuated out of the chassis (and far from the intakes).

 

You need to run Kombustor burn-in in D3D11 mode at 1920x1080 resolution. My temperature was also around there in OpenGL mode. In D3D11 mode at 1920x1080 I saw 93C, but it might be different for your GTX 675m.

 

Ok ill try this but regardless, I used OpenGL mode for all of my tests and saw a drop in temps. Max will change depending on the mode but there was nevertheless a decrease.

 

Any chance you could log your GPU fan speed and ambient temperature when your GPU settles at 84°C?

Sometimes the fan can settle at level lower or higher rpm. It's possible that your fan is spinning a little slower than before the foil mod (or faster). If this is the case, it's possible your foil mod is more effective than it appears (or has little effect <- unlikely though).

 

its more than 3mm
the gap is around radiators,big gap on each end
its not placebo,foil mod is working

 

The gap is huge in terms of cooling and a 10oC drop in temperatures does not lie. Just like an exhaust system without clean air flow from a blowing exhaust performance suffers.

£1 roll of alluminium tape for results better than any cooling mod. No argument.

 

Fair enough. I stand corrected. Just thought I'd play the devil's advocate

I've heard about "burning-in" your new components. Is there any sense behind that?

 

That depends..

 

Don't know if it's the foil mod or me repasting with Arctic Silver (or both) but after running Kombustor for ~15 minutes my 7970M is only at 81C

Used to get to 90C-95C after about ten minutes... nice!

 

It won't be the Arctic Silver

It is difficult for people to realise what a big difference this makes but the only cooling a laptop has is the fans through a radiator which draws the heat away from the chip. If say 20-30% of the fan air is escaping the cooling efficiency is crippled by 20-30% so in terms of temperature expect an increase of up to 10-15oC. It hurts the GPU more because the gap is more profound. This mod is a no brainer it is simple and easy to do everyone should be doing this. It does not excuse the laziness during production however if this was a wood joining factory the designer would be sacked on the spot.

 

Heres a few screens (I can do a full log i you want but thought you might be able to see from these). It started to fluctuate from 83 to 84 around the 8 min mark. Did that for another minute or two then settled at 84 around the 10 minute mark. The fans were mostly around 3250 until the last 30ish seconds, then jumped to around 3400, as you can see in the screenshot of Kombustor and the sensors. It even went back to 83 at the 12 min mark (but i believe this is because of the load anomaly mentioned below)

Note: for some reason my GPU load was jumping around usually in the 95-95% range, but would occasionally dip to 86% for half a second then go right back up. I believe this is because of the Clevo drivers I just installed.


Edit: Log of a lengthy BF3 session. Column AX is GPU temps. Hits 75 maximum, usually sitting in the 73-74 range. Never breaks 3k RPM either.

 

Try Dx11 mode + 304.79, your temp should increase a lot because it's always running at 98% @1280x720.

 

Yeah I had 304.79 a few days ago and it still wouldn't break 84, but I'll run it for your sake

 

yeah the only problem is when you need to clean the dust out, you will have issues.

 

What was your approximate ambient room temperature?

Thanks for the log. It's interesting to see how being on the edge a 75°C trip point could really cause some funky fan behavior. 150 RPM difference is not negligible noise. If you can keep your cooling below those trip points, it result in a more enjoyable experience. Looks like 150rpm causes a good 3°C drop in temp.



It's a good idea to log your performance like that once in a while. When your fan bearings start to get tired it'll really show.

 

Yeah maybe I can put a bit more tape to really cover the gaps. Overall I'm pretty happy though. Lowered max temps in BF3 by 5 degrees. My ambient room temp is ~78 degrees F. Also, is that graph doable right in google docs?

 

I was just making a broad comment, that some people's cooling might be short of 1°C of what you have, so their GPU fan may oscillate between 2850 and 3000 rpm constantly instead of holding steady at one speed. It seems you have good cooling. Not sure if you can achieve much more by adding more tape. If anything, you can consider increasing intake grills apertures. Is the laptop on a cooler?

It's interesting to see also that there can be a good spread at each fan levels. I'm sure the min and max fan speeds within each fan "level" can make a difference at least 1-2°C sometimes in results, but it may not be noticeable by ear that the fan is running at a different speed. I'm not sure yet if the spread at each fan level is random or related to temperature.



It would be interesting to see how these factors vary from one system to another.

I haven't used google docs, but it looks like it should have tools to plot graphs etc:
Create a chart or graph - Google Docs Help

 

It isn't on a cooler, although I do have one of those crystallized gel ones that sit under the laptop and basically, when heated, the little crystals turn into a gel and that is supposed to help dissipate heat from the bottom. Considering the intake is on the bottom, I figured it might just hinder the cooling but I'll try it just for fun.

 

I'm getting the rig in the sig in another 3 days. My plan of the mods will be the following:-

1- Stock paste + no cooler benching.
2- Stock paste + cooler without fans.
3- Stock paste + cooler (U3) with fans.
4- MX-4 + no cooler.
5- MX-4 + cooler without fans.
6- MX-4 + cooler with fans.
7- MX-4 + cooler/fans + foil mod.
8- MX-4 + cooler/fans + foil mod + widened intake grill.

I wish I could do the extra aluminum heatsink/fins mod with minifans but I simply lack the DIY know-how and I think it's a bit excessive since I don't plan on any heavy OC'ing (international customer, can't RMA if fried without shelling out +350 USD for two-way shipping) and just want a stable cool peformance to last me 2-3 years.

I have a fairly powerful 32 000 BTU AC in the room that I plan on hooking to some improvised funnel to direct the 14 Celsius air directly into the laptop at one point just to see how cool it can run

So I take it that kombustor is the stressor of choice for most people? I heard that AMD/Nvidia warn against it because it can damage components ..etc. What is this "burn-in" I hear about by the way? I never thought it wise to stress a component for 48 hours straight just to ensure stability. Instability issues are bound to arise soon if they're meant to. Isn't that like buying a brand new car and then driving it 2000 miles non-stop at high speeds?

Also, what are the diameters of the GPU/CPU fans in the Sager NP9170? I'm looking to order some intake grills to replace the pathetic ones Sager has, perhaps from PC Case Mods Store, so I wanna order accordingly. Although I'm assuming 80mm will be overkill as it is.

 

Kombustor stresses the GPU basically by utilizing it to its maximum potential, much like a very high end GPU hungry game would. You don't run it for 48 hours either. Most I have done is 12 minutes just to see how the temps go. If you left it going for 2 days problems might arise, so I don't recommend that.

As for the intake grills I don't have a 9170 so I can't give you a definite on that. I would say though that with the MX-4, the foil, and the cooler, your notebook should be running at very comfortable temperatures. Probably isn't even worth it to replace the intake grills, especially if you are going to be pointing AC at it as well. Just my 2 cents though.

 

I'm familiar with kombustor but was wondering if there were other tools (apart from benchmarks obviously) and if anybody practiced the "burn-in". Yes some people stress their components for hours on end to "break them in" whatever that means. I'm not going to point the AC at it. The entire house is pretty much ACed 24/7 (Middle East) but I thought that just for the heck of it, I'd funnel that AC's output directly into the laptop for kombustor for 10 minutes and see how cool it runs. Not on a permanent basis.

Tbh I think that the intake grills (and possibly the foil) are the most important mods. The stock back cover grills covers roughly 30% of the fan areas and the grill itself is so fine I doubt enough air makes it. It requires cutting and soldering no doubt so not as straight forward as foiling, but it's also aesthetic if you get something that looks nice

I'm kind of guessing that the intake grills will have the biggest effect on temp.

 

Prime95 is another stress tool I've seen people use.

 

No news yet from any owner of the NP9170 on the size diameter of the stock grills on the back cover?

 

I've just done this mod to my sager np8150 (7970m upgrade).

Without this mod (but fresh IC diamond application) I had a max temp of 77C.

After this mod, max temp was 74C.

Room temp was around 20C for all tests.

This is a very good mod. So simple.

 

noobish question, but.. what did you do? never did this before so i don't see a difference

 

I was also seeing a 3C decrease on mine with the same mod.

 

You need to read the thread. There is plenty of discussion about how to do the mods.

 

well i actually did read the thread, i just thought he was posting something new in the pictures and i didn't see it
thx anyway

 

You don`t see the tape over the copper and the fans?

 

Guys, room temperature makes a HUGE difference. After the tape mod when idling (with windows and sager in balanced power mode) I would see 2 cores around 49C and the other two in the low 40s or high 30s. Now that I have moved to my dorm which has AC, two cores idle in the low to mid 40s and the other two in the mid to high 30s. I didn't know how big of a difference room temperature would make on idle temps, but it does make a large difference.

 

as i said, i thought he was talking about an easy mod he did which hadn't been mentioned before and i didn't see any difference in the pictures. misunderstanding, thats all

 

Anyone with the 680m applied this mod???? I know the 680m runs cooler, and I'm curious if this mod help to further lower the temperature.

 

with this easy mod i gained roughly 15/16°C on my VGA, previously i could reach 102°C under kombustor (@1031/1500), now i don t go above 87/88°C..amazing. Tomorrow ill order some nice cooling paste from amazon.com and repaste too.

PS which 1 is the best cooling paste?

 

ic diamond 24
I use that and its fantastic

 

hmmm today i ordered the Prolimatech PK-1 after reading some reviews, i hope i don t have to repent of this choice......any experience with such a thermal paste?

 

Fanto... U do realize ICD24 and ICD7 are the exact same thing correct? 24 just comes in a bigger tube

 

@james_het, I'm sure whatever paste you use will be better than stock, it's the application that matters, not the paste.

 

I found it to be working better than most of the paste I tried, however PK-3 is the best I've found.

 

Thanks guys, so far im happy with the temps i reached for the VGA, now i hope to improve those and lower the CPU Temp as well...

 

thx for the guide! +rep!
i've been following it on my p150em, used Prolimatech PK-1 to repaste cpu and gpu and then the foiltape.


Results:
15 minutes kombustor run (burn in DX11) before: max 103°C with Fan@Max

15 minutes kombustor run (burn in DX11) after: max 94°C with Fan @ 1 below max


now i only have to figure out to OC this card... been following all the guides, still showing 0/0 in afterburner, even with kombustor putting load on the card :/

 

Try going to program files > msi afterburner > MSIAfterburner.cfg
And change:
UnofficialOverclockingMode = 1.

 

yeah that doesn't work... i've included the ati files, tried different driver versions (modded/nonmodded), registry ulps changed, put load on the dGPU prior to starting afterburner and so on... as soon as i change the .cfg settings from 0 -> 1 afterburner shows me 0/0 instead of 850/1200 and the sliders are stuck

 

Two things:

UnofficialOverclockingEULA = I confirm that I am aware of unofficial overclocking limitations and fully understand that MSI will not provide me any support on it
UnofficialOverclockingMode = 1

Then just save (admin permissions needed)