The gap between the GDDR5 and the heatsink is too wide to use thermal paste, hence the reason for pads. You don't want to burn the memory after all...
-
Ok I'm considering getting this one, Thermal pad he claimed that the thermal conductivity is 12W/mK, if I'm not wrong, ICD7's thermal conductivity is around 4.5W/mK according to ICD's official site. 12W/mK is some crazy number I know, what do you guys think?
-
Do it with ICD7, if you do it right, you'll have a thickness way below 1 mm which will more than outset the lower thermal conductivity.
The conduction resistance to heat transfer is as follow R = t/(k*A) where t is the thickness, k the thermal conductivity and A the surface area perpendicular to the direction of the heat transfer. Considering that the surface area won't change, if your paste thickness is three times lower or more than the pad thickness and it will be WAY below that with good paste application, the overall resistance to heat transfer will be lower, resulting in overall lower temperatures. Thermal conductivity isn't everything... -
I see, so my best bet here is probably ICD7 on the die and this 12W/mK thermal pad on the VRAM. Refering to your signature Tijo, is 1.0mm thermal pad too thick for G73JW's VRAM?
-
Don't know for the JW, but honestly, you shouldn't have to touch the thermal pads for the vRAM, the stock ones do the job quite nicely. I got maybe 1C lower by changing the pads so it's not worth the trouble, it just gave me more bragging rights for my sig
.
If you go for a pad change nonetheless, you should buy a larger piece than 30mmx30mm. I bought 2 pads that were 10mmx150mm. I had to cut a small border around each pad to make them the same size as the original pads, but i still ended up needing a pad longer than 30mm. A 30mmx30mm pad wouldn't have done the job.
You have a JW i suppose, what are your temps anyways? I haven't heard of JWs needing a repaste yet. So i'll ask this, if your temps went up, did you clean your vents first? -
Max temp is around 84C clocked at 800/1600/1600 after 30min of Furmark 1.9.0, the temp is ok, after reading some articles of thermal pad it just gives me the idea that would replacing the pads make a difference versus stock pads if I want to maximize the thermal conductivity.
Thanks for your replies though, I'll definitely take your advices into consideration if I happen to replace them. -
Your temps are way more than fine, if it ain't broken, don't fix it
. I repasted because i had to, didn't repaste even after seeing Chastity's results until i ended up with degraded paste due to bad luck.
-
i repasted with IC7. i think ive used too much paste because my temps increased by 2° C...
ill do it again... and report....
i have used "a pea size" of IC7. that was too much. I was concered about the pressure of the heatsink on the die. it seems not tight enough. so i decided to put enough IC7 on it.. stupid that i did not try and check the paste once, before reassembing... -
Once the paste is there, you must not remove the Heatsink again. Like Chastity i used a 3mm pea, anything in the 3-5mm range should be good. Make sure the heatsink screws are tight in place but don't tighten them like a maniac either and tight them in their numbered order 1 to 4.
-
Size of a very small pea.
-
yes. my pea was like the green peas.
but something went wrong. ill check it today.. at the evening. i think it was just too much.
any more advices? hehe...
this time ill mount and check and redo. just to see if my pea size was enough/too much... -
A little bit bigger than the blob of compound in the picture I posted but if the heatsink sits flush as it should do if replaced correctly and firmly bolted then a small amount should be fine.
-
i just worry about the small round pea applied in the middle of the die may not get to the corners...
-
If you are worried about that, use a 5mm pea and trust me with the pressure from the heatsink, it will spread even outside of the die area.
-
By putting on too much it may seem wise but you are just lowering the conductivity between die and heatsink and causing higher temps. I have the thinnest of paste on mine and I dont see the 80 degrees very often nowadays. IC7 or a thick TIM is best when using the pea method because it literally does all the work for you. -
ok, fine.
I did it again. And i did not need the video this time
It was too much paste....
Now I have a drop of 2°C in idle and max. on furmark. That does not sound much but my stock-temps were registered at a much lower ambient temp. i would say that makes a difference. so im happy with the new temps...
if i tell you the stock temps you would kill me for even thinking of a repastebut i want the G73 to run as silent as possible. I did the cpu as well.
CPU dropped about 5°C. (38-43° on idle)
How can i check the Graphics-Mem-Temp? Im using GPU-Z... It gives me 4 GPU-Temps that are nearly the same... -
nearly all the same temps is very good actually. means you did it correctly.
-
GPU-z is fine for checking the temps I like to use HWINFO sensors personally. -
Just to kind of recap... It's been 10 months since the first post
and I'd like to show my current performance. As you can see, even after almost a year, the TIM has not degraded, and I still get my 47C idle. The image shows a 820/1100 overclock, and a max of 80C.
Damn fine paste.
On an off-topic, TeamROG is doing an event tomorrow, and Chew is prepping for the event... by doing live LN2 overclocking now. (He's like this... he's done liquid nitrogen overclocking in a hotel room for 3 days straight) If you want to catch the feed, you can see it here. You can even make a nick and chat online with him, the team, and other OC wackos.
(I know how much everyone here <3 overclocking)
Anyhoo, be cool and GAME ON! -
Will the MemIO temps also drop with core repasting?
-
-
Yep, the memIO sensor is for the memory controller which is on the silicon die. There is no sensor for the vRAM chips.
-
If done properly your MemIO should be only +/- 3C from core and shader, if not less. Mine is about +1.5C
-
Been running 835/1099 stabel now for 3 months with IC7. Might be the weather as its getting warmer but the GPU idle temps have gone up 4-5c. With a room temp of 25c am running at 52-53 and full load 70-74c. The fans speed up more often now but just for a few sec.
Still looking for better thermal pads to do a repast. Strange is my mem is hotter then the core an about 10-14c. and when gaming can be even more then 20.
So if core is 70 mem is near 90. I think i need to replace the thermal pads. Might be one of the reason i can not set clocks to 1100.
Why is on pad half transparent and the other is a kind of blue color looking? -
High MemIO like that usually means unsatisfactory pasting on the outer edge where the MemIO part is. You may have to redo it and use a little more paste to get a better cover.
-
Did not repaste the mem just added new 5.0w/mk thermal pads. But now found some 7.0w/mk pads for sale. Also i got the parts for my heatsink mode so i will have to repast anyway
-
Only put thermal paste on the GPU Chip and use 1mm or 0.5mm thermal pads for the VRAM and cut 3 small pieces from the pads for the memory controllers for best conductivity and the pads should be unsoiled and clean.
How are your temps now after repasting? What compound did you use? -
No i did not use it on the mem chipsi used the 5.0W/mk pads (something similar to the stock ones. After i repaste the GPU and add a bigger heatsink with fins (not like the so called fins on the stock heatsink) then i will replace the pads with 7.0W/mK pads.
before IC7 the GPU temp was ok Idle at 51c and and load 76. But i dont know how the card calulates the avarage as at full load core was at 67 and mem was near 88. At sometimes even over 90c.
After i repasted with IC7 and added the thermal pads (5.0W/mK) idle temp was 46-49c at 21c room temp (no cooling pad). Now i got temps that i said a few posts back.
For a short time the mem and core were almost at same temp. Now am getting sometimes almost a 20c dif.But everything work ok if i do not go over 835/1099 i am ok. 1100 will make it crash.
I might have done something wrong with the pads. Will find out on the weekend when i take it appart again. -
Again, the memIO sensor is right on the edge of the chip - look up the chip diagram on google R770, I THINK, could be wrong (just working off memory). Anyways, the memIO reads temps from the edge, which means that even a mm of paste will make a 10 degree difference.
I've got a 5C spread between core and memIO. Fine by me, and I max out around 95 with furmark, and 86 or so when gaming. It just isn't going any lower for me. -
As DCX mentions there will be a difference between the temps and I saw 9oC difference with AS5 even though I repasted correctly its just the way it is but with ICD I only see 2oC difference max so I got away with a jolly good repaste.
How did you apply the paste did you use the pea method? Did you warm up the ICD before applying it? 20oC is too much go back and try again and make sure you reseat the heatsink correctly and apply the screws corner to corner 1-2-3-4, clean off all old residue with alcoholic swabs.
3mm pea method. -
Ou well better luck next time -
As soon as you apply the compound to the GPU you must reseat the heatsink 'immediately' as ICD will set within seconds once it makes contact with the cold surface.
That could be your problem you might have a hardened blob that has not reached the outside of the chip. Once applied ICD turns into a putty type substance very quickly. -
Sry wrong way lol after i added the heatsink waited for 3 or 5 min. -
''Do the one thing you think you cannot do, Fail at it. Learn from it. Do it better the second time'' -
It is not critical to warm the paste for application. I just did a CPU paste, and never even warmed the tube. It's also so super conductive of heat that once you place it on the CPU, it transfers its warmth to it. I watched it cool right up during application.
The warming is to just make it easier to squeeze out and hopefully shape into a ball. -
Yes its much better to apply when warm. First notebook i repasted with IC7 i did not heat the paste and man was it hard to get it out of the syringe.
Was reading the alienware posts and found this http://forum.notebookreview.com/alienware-m17x/507962-m17x-r2-full-internal-cooling-mod-huge-improvement.html
How healthy is it to add paste to the vram ? I know Dallers said not to but just asking as they seems to have done it with alienware. With IC7 there should be no problem of shorting something out. Also the heatsink will cover the GPU better? With my first JH repaste i replaced my pads with 1mm 5W/mk before but it seems to force the heatsink up abit leaving me to belive that its not pressing on the GPU as it should.
i think someone on this forum used 0,5mm thermal pads or am i mistaken?
A nother thing that one of my lecturers said was why not use 0.5mm copper spacers on the vram like the GPU has on it. -
People have used Thermal compound and copper shims on the VRAM but do so at your own risk I wont recommend either because Thermal pads are much more reliable when covering several sections evenly, you may see better results by using the alternate methods but I would only do them if you are confident you know what you are doing. -
Hey so i repasted my GPU about a year ago using IC Diamond. At first, the temps were held steady at like 60 and it was phenomenal. Now, however, the temps can get back up to 100 during 100% load and the GPU and CPU fans are pretty much always revving. Could this be just due to dust buildup in the heatsinks? Or do i have to repaste my card
-
Hello guys!
Can anybody tell me how big thermal pads I should get for my JW's 460M?
I really couldn't find any info besides they should be 1mm thick and since I
need to order them online, taking laptop apart just to measure currently installed
pads in not an option.
Thanks!
Chastity's Repaste Experience
Discussion in 'ASUS Gaming Notebook Forum' started by Chastity, Aug 2, 2010.