M17x R2 full internal cooling mod - huge improvement!

So to sum up this lively conversation

Mkelliny, your temps are awesome, especially at the rate that your 5870's are clocked at and that you managed to achieve this without the retention clip mod.

Fellow to-be-modders are more then welcome to try Mkelliny's method, however results do vary so it may be more or less successful (hopefully more ). As we all know, this is the case for any mod as different cards and conditions can cause cooler or warmer results.. So i think what Cygoris is trying to say is that more consistent results can be achieved via the retention clip mod, as is shown in the past. However, with any mod there are inherent dangers, and there IS a possibility that damage may occur to the GPU if done carelessly and incorrectly.

Congrats to Mkelliny on his successful outcome, and I wish all our temps were as good as Mkelliny, including me! But I have resigned to that fact that I have hot 5870's (I had very high temps, straight out of the box).

Don't flame me for getting involved, just trying to be the mediator

 

see my 5870's never ran over the high 80's before I did my repaste and repad. So I never really had hot running GPU's not like what others have said on here in the 90's and higher. Yes I do believe that he was trying to say that the retention mod is more consistent, and I would agree. I just am happy with the results I achieved and how I achieved it and wanted to share my findings with my fellow NBR members/ Alienware Owners.

 

Mkelliny is right, a risk, however small, does indeed exist. We all approach risk taking differently, it is prudent to err on the side of caution.

With that said, Cygoris is also correct that although a risk exists, it may be a risk worth taking as it appears only one with the mechanical dexterity of a Neanderthal falls into the high risk category and the reward-4 degrees Celsius depending on testing conditions and pads or paste-is a big reward.

The main problem is trying to determine whether one falls into the Neanderthal mechanical dexterity category, and some of us unfortunately WILL find out that we do.

I hope that I dont.

I am using Shin Etsu, can anyone describe the "credit card" application method in a bit more detail, preferably with pictures and video?

Am I applying it only to the GPU die and the top of the RAM chips and removing the thermal pads that lie elsewhere on the cards altogether or do I apply TIM there as well?

Thanks everyone, especially to those who are taking the time and energy to help the unexperienced such as myself complete this task.

 

I cant reply from experience....yet But I have Shin-Etsu on the shelf ready to go. I am just waiting on the thermal pad for the CPU and I am going to redo the whole lot.

I think the credit card application is straight forward, your looking for as thin as possible and as even as possible....while the Shin-Etsu is hot. I am using this method on the GPU die's and the CPU.

I wouldn't use the credit card trick on the ram, as don't forget some of the ram doesn't make full contact with the heat sink so you need some more TIM for bulk to fill gap. I am going to use the grain of rice method, but will probably use more than a grain as the ram chips are larger. For those where there is the 'gap' use a little more.

A lot of this information is thanks to Cygoris, so he deserves the Cudo's.

Also, in terms of the retention clip mod, the main danger is to apply pressure unevenly. I believe the GPU's themselves can handle a bit of pressure, just not unevenly, as that is when they crack. Use the numbering on the retention clip, start off by screwing each one with the same amount of turns. For example, 1 turn for each screw going around and around the clip. You will find that one side has less play, so I stopped tightening that side and continued on the side with the 'play' until I felt it was as roughly the same tension.

Just take it easy and evenly, and I think you will be alright

 

YouTube - ‪Tuniq-How to apply TX-3 thermal grease‬‏

there is a video choose which method of paste application you want to use is totally up to you. I use the horizontal line method in the video and drag down some use the rice grain and some use the pea.

With TUniq products you have to use more than normal so don't use as much as in the video, follow each brands application amount directions.

 

Hi all,

Just giving an update on my latest shenanigans!

I have re-done my 5870's using Shin-Etsu X23. I had previously used AS5 on the GPU's and TX-3 on the memory.

I have also done the re-paste on the i7-740qm using Shin-Etsu X23, as well as a Fujipoly thermal pad for the voltage regulators(?) and the results are most pleasing.

All tests where done with my NZXT Cryo LX at full capacity, as this is how it is all the time anyway.

There wasn't a great improvement for the 5870's. I wasn't expecting much though, as I had already done the re-paste and the retention clip mod. BUT! any improvement is a good thing and my second GPU is running significantly cooler then before. However I would not put that to the Shin-Etsu but rather to a better application of the TIM.

The CPU on the other hand has shown very promising results. Strangely, more so at higher temps (single core, turbo mode) then at lower temps (all 4 cores).

All screenshots that you see are with my GPU's clocked at 800/1100 and my CPU FSB set to 151Mhz, from 133Mhz.

So the pictures below, in order, are:

1) Temps of my 5870's with the AS5, TX-3 and Retention clip mod clocked at 800/1100 & 1.1v

2) CPU overclocked at 151Mhz on a single thread Prime95 run with the stock Dell paste\pad for ~10mins.

3) After the Shin-Etsu X23 application. CPU clocked at 151Mhz, 5870's at same clock as above. Running Prime 95 in a single thread torture test as well as Furmark for over an hour.

I am happy with the results and tbh, the CPU re-paste wasn't actually that hard, just gotta disconnect a lot more stuff

I have some more before and after pics of the CPU running Prime95 across 8 threads if anyone is interested, however the differences are only about 2-5 degrees.

PS: I am going to keep stress testing the CPU as I realise that those temps are quite low and I want to be sure they are accurate...

 

Was the Shin Etsu x23 you used soft and easy to spread or did it crumble, dry quickly? Reason Im asking is that I purchased some Shin Etsu X23 off of ebay and had soaked the tube in hot water for maybe 20 minutes or so while I disassembled the R2 and contrary to what others had said it was very easy to apply, it was like toothpaste, not crumbling, cracking or drying immediately.

So I am wondering I purchased some fake Shin Etsu. I ran out of Shin Etsu and had to use Tuniq TX-3 on the CPU which in comparison was vastly more difficult to apply, much thicker and requiring much more patience, I actually had to redo it several times via the credit card method before I had a satisfactory application.

My temps a few days after application on both GPU's were 75 C on one card and 81 C on another about 10 minutes into a stability run on Furmark 1.8 with 1920x1200 resolution, AAx2 and after effects (heatwave). The cards were not cold to begin with, and I playing around with SetFSB trying to find an ideal OC figure.

I have done the C-Clip mod properly as well.

 

Ran Prime95 for another 40 mins, and the results appear to be consistent.

They are a bit higher cause it is pretty cold today and the heater has been on. TBH, these look a little more accurate as to what I was expecting.

I placed my tubes of X23 on a lamp for about 30 mins and through the entire process of dismantling the laptop. It was soft, but still thick and wasn't the easiest to spread. Especially if you tried to go for a realllllly thin application. It did not dry or crumble during the process.

Actually, i used alternate application methods on my 5870's. On one I used a card to spread the TIM across the GPU. I wasn't overly impressed with the spread so i decided to do the pea size application on the other. I did the GPU last (after the memory) so it would be warm when I applied the heat sink. I did the same for the CPU.

Both methods had the same outcome as you can see by my temps.

 

I just do the pea method, simple and easy.

 

I agree!!!

 

Good read, thanks man!

 

No problem!

If i can give another piece of advice. When i removed the heatsink from the 5870's after the first application i noticed that i hadn't used NEARLY enough TIM on the memory, especially those where the heatsink doesn't make full contact. I am actually surprised it didn't cause any issues....

Make sure you use a decent amount, and i wouldn't worry to much about it spilling over as the memory chips themselves are quite large....i would almost be inclined to say to use literally a blob the size of a pea, not a grain of rice... Maybe more. If you are unsure, apply to one or two of the chips, put on the heatsink, take it off and look at the spread. Then give it a clean and increase or decrease the amount until you feel it will provide enough contact.

 

For anyone still looking at this thread, here's some interesting information on a review of 25 different thermal compounds. LINK

 

Hate to necro this thread, but has anyone seen this?

M17x-R2 CFX 5870 modding for extreme oc

 

Nice cooling mods, Mkelliny. The plastic triangles on the corners of the die is pretty smart. Thanks for the link.

 

I'm sorry to be a bother, but is it at all possible for someone to create a video in which they perform Ash's three mods in detail? I'm having some trouble understanding the descriptions, especially for the fan/duct mod.

No disrespect meant toward Ash whatsoever. I need this mod to work because I've peaked at about 111 C according to GPU-Z. The fans work fine, so I suspect it to be the thermal compound. While I'm in the computer, I might as well mod it so that it runs cooler.

 

If anyone with a retention mod is reading this, could you please share your thoughts on how it can affect the GPU/CPU long term? I did something similar to my laptop - added a 0.4mm copper shim between my heatsink and GPU/CPU (for an Elitebook 8530 heatsink mod - check it out, maybe it's useful for other laptops, too), then screwed it down, and the pressure definitely increased because of the copper shim.
It works fine, the temperature is lower, too, but I decided to slightly loosen the screws just in case. I wonder if it can damage the GPU or CPU over time, especially since I'm overclocking both.

 

I have done absolutely everything you can do to these heatsinks, and I can tell you that the retention mod is not your best option. However, you've got to be a bit more hard core than the average bear.

If you take a look at your GPU heatsinks, you will see that there are two main areas. You have the copper square, which obviously sits on the GPU die. There is not a whole lot you can do here aside of using aftermarket thermal paste. AS5 works just fine. You don't need anything fancy.

The second area consists of the two rectangular sections that remove heat from the RAM chips. You more than likely had the el-cheapo Dell thermal pads here.

You will then see the 'rest' of the heatsink, which extends out over most of the other electronics on the card. This part is negligible, and actually tilts the rest of the heatsink upward so it doesn't get complete contact the GPU die or the RAM.

I took a pair of tin snips and cut the part of the heatsink that extends past the RAM chip cooler completely off. If it was doing anything important, it would have thermal paste or pads.

After cutting this part of the heatsink off is when I noticed my largest drop in temperature. This allows all of the compression from the four screws to be exerted squarely on the GPU die and RAM chips.

You can do one better by gently bending the silver tabs that hold the screws upward just a bit. This will make the heatsink press even a bit harder on the GPU die and RAM chips, but you won't have the PCB bending like with the full retention mod.

My GPU heatsink

Don't mind all of the Microsoft Paint writing on here, but this is one of my heatsinks. There is literally nothing left but the RAM cooler and the GPU die cooler.

 

Thank you for this post! Here are my results.

System specs.
i7 72QM
Single HD5780

Temps Pre-MOD at Idle



Temps Post-MOD at Idol



As you can see that's an average of 30 degrees difference running at idle!

Temps Pre-MOD stressed with FurMark



Temps Post-MOD Stressed with FurMark



And a >40 deg difference in temps while under stress! I'm very pleased.

FTW I did not perform the ventilation tape MOD.
I also used Arctic MX-4 for both the GPU and RAM.

My fans are super quiet now and they rarely kick on anymore unless I'm doing some serious gaming.

Thanks for the tutorial OP! +rep

 

I'm seriously considering modding my heatsink by cutting off the overlapping parts as you have done.
Just a couple of ?'s
Have you experienced any long term adverse effects from this mod?
How many degrees C drop did you gain?
Seems like this should be the best solution by far.

 

Is there an up-to-date description of safe changes one can make to the R2 with 5870's anywhere? I'm thinking about paste and thermal pads, mostly.

 

Nice modding

 

Thank you, Ashtefere, for this post! I just did this using AS5 and AS Ceramic and my temperatures have dropped by 10-15 degrees Celsius!

 

Does it works with m17x R1? (too lazy to dissasemble it just to check if its the same)
cuz mine runs too hot @.@

and if it doesnt...is there any mod for r1?

 

What about more recent M17x's?

 

Well see, that's your problem.
Why look into modifying the stock cooling system if your temperature problem can be fixed by simply cleaning the heatsinks and probably changing the thermal paste?

In time, the heatsink fins get clogged with dust which affects airflow.

 

Your criticism isnt requested nor useful.

I disassembled it 2 months ago, cleaned it, and changed the thermal paste for Artic Silver 5 paste

Im not gonna disassemble it again just to CHECK if the design is the same as in the R2, which is why im asking here, to save some time.

So please if you dont know, dont waste the time of the readers in this forum by replying questions that werent asked.

Do somebody knows if this mod works for the R1 model?

 

Wasn't criticizing, just stating a frequent situation. Figured, since you're too lazy to disassemble as you pointed out, that this might be the cause of your high temperatures.

Now to answer your question, the M17x's design is very similar to the M17x R2.

 

Thankseh, Ill try it then ~