Repasting GPU, No change w/ ICD7

So I did a little repaste on my 460m hoping there would be like a big drop but nope.. there wasnt..

Before repasting I ran furmark for 15 mins and maxed out my temp at 93 degrees C.

So I bought the IC diamond 7, repasted then I tried FurMark again. STILL it maxed out at 93C. And no drops at idle temps.

The only difference is, before the repasting, it only took 2 mins in Furmark to reach 93. Now that I switched to the ICdiamond it took about 10 mins to reach that high.

Any ideas on what I did wrong? or was the repasting unnecessary to begin with?



EDIT: Btw I've repasted my CPU too.. max temps before was 95-97 degrees C. Now it never goes above 84.. I was happy with the results on my CPU but i wonder why it never mimicked the same results on my GPU. No change at all.

 

IC Diamond does usually help by 2-5C it could need some time to "set" before you see the lower temps.
Refer to the the Application instructions as well to make sure you use enough or not too much

 

Clean out the vents in your laptop if you haven't yet. Also, let the laptop run for 24 hours before testing temps because the thermal paste needs time to spread and produce optimal conductivity.

 

Aren't those temps still a bit too high for temps after repasting?

 

yeah i know right? its pretty high... i had some idea of how to repaste the GPU.. i had some success with my CPU but i dunno what happened with this one..

and also the vents are clean.. i made it a habit to do some cleaning atleast every 1-2 months.. and i cleaned my vents before i put everything back together


EDIT: the link above ive already seen it and i did the same method.. the same 5mm peanut sized method spread evenly with pressure.

 

What are you using to measure the temperatures?

 

furmark has its own GPU sensors.. plus i also have hwinfo64 running in the background.. they show same results..

 

I haven't repasted my GPU or CPU ever. Idle temps are 48C and 50C respectively. Max temps while gaming are 83C and 82C respectively (playing GTA4 maxed out). I have never cleaned the innards of my laptop nor have I ever blown out the vents. Neither is my laptop overclocked/underclocked/undervolted. What were your temps (idle and load) before you repasted and what are your idle temps now?

 

Remember repasting is an arguable myth.

I swear by it, others swear by it but its not the only factor involved. I always say proper application and conductivity beats any type of thermal compound as you have compound on the Vram you should not have the problem we have with the JH in that the pads are too thick.

I always say ICD because it allows you to apply it so easily but the differences in the actual compound themself for overall loss of temperature is minimal. One thing you normally see with ICD is it may take a while to reach the max temp but it also takes a while to come back down to idle once the load drops.

How did you apply the compound? Did you use the pea method? Did you warm the ICD before applying it?

If you answer yes to all three then there must be other factors involved for instance the heatsink is not flush enough, dust and dirt clogging the fans/vents or overall the Die/heatsink can just run hot from not being evenly flat which can only be fixed by lapping, however until you hit the 100's most tech's will say those temps are fine.

If you answer No and you are doing the credit card method or likewise I recommend going back and trying again because first attempt does not always reap results. Firstly look at the imprint on the die when you remove the heatsink and see if you have too little paste there to cover the entire area. Place a 3-5mm pea shaped blob dead centre of the die and Vram modules and lower the heatsink as carefully as possible onto the screw locks and then tighten in numerical order without moving the heatsink too much while doing so.

 

Well, 93 C max temp for your video card is acceptable, albeit a bit high. I currently get 87C max without repasting (which I will do tonight with ICD7). If you don't have any heat related issues, can you be happy with it the way it is?

 

i'd think so. they seem to align with a fair/poor stock cooling on the gtx460m in the 15.6 incher. he's got the jw not the sw, but on the pure gpu temp side he's sitting about where i am after four minutes.(caveat: defective gpu here pending replacement climbs to 96C probably would hit higher if allowed to continue running).

the change in ramp up time to that temp says the tim has had an effect and is moving the heat away more efficiently, and the thermal bottleneck is further up the cooling system. could be ambients too high, fan isn't getting up to full speed, poor heat pipe fill level, airflow blockage( could be dust, something blocking exit vents, something blocking inlet vents- note some large single fan cooling pads can actually create an airflow dead zone over the air intake and increase temperatures when running thier fan at mid or high speeds), or gods forbid did you bend or kink the heatpipe even slightly?.

did you let it sit as a 5mmcube worth of icdiamond before assembling on the gpu side? you know you are supposed to let it exhaust some of the alcohol before pressure clamping the heatsink on, and as noted it also takes a bit to cure before first use, and full cure after. if memory serves it was something like 15m>heatsink>hour>startup/operate>24 hour cure. lot shorter than the silver based t.i.m.s 200 some hour full cure time.

eh shrug. if it was a g73_w i'd expect a stable point at 85C as the middle of the bell curve for occt/furmark burn ins. for the g53_w it seems like 90 is expected, so might be worth retrying it after a few more days if there's no improvement.

after all the old adage is each degree below rated operating temp is twice the expected life for the component.

edit: of course it takes longer to drop temp with icd. efficient thermal transfer works both ways. if the heatpipe/radiator/fan assembly is hotter then the chip it makes it move heat just as readily into the chip as it does out of the chip when the situations reversed it's thermal paste/thermal interface material not a peltier device. always see pads as the weak link in these designs, but understand the cost effectiveness/mass production reasons to use them. biggest reason to like icd over artic silver/other. diamond is carbon. pure carbon. granted these notebooks seem sensitive to static charges building up in tape glue in places even, the carbon is far more neutral/safe than silver if it oozes someplace it shouldn't. then again technically the antifungal cream/zinc oxide+aluminum pastes of yesteryear (yes the same active mix in the old white thermal grease as antifungal foot stuff over the counter- i find it ironic and amusing) was electrically safer then silver. lower electrical conductivity/more resistance.

 

ICD is a TIM it does not need to cure it is set in around 10 minutes if that.

Also as soon as it is applied to the die you should reseat the heatsink because it sets very quickly.

 

In fact, unless if you totally change the cooling system, the G53/G73 will always reach 93°C with furmark, mine does with stock paste whatever if it is overclocked or not.
When you reach 93°C, you will see that the performance lower a bit and temps will jump between 92 and 93°C: the card reduce the performance to avoid overheat. The fact that it takes longer to reach this temp prove that your repasting is effective.
Unless if you reach those temp in games ( should not happens ), this is not an issue.

 

Are you saying that the GPU is being throttled? I know that the gtx580m GPU in the Alienware m18x is intentionally designed to throttle when running Furmark. I don't think this is the case with the G73JW but it would explain things.

 

Yes it is throttled as soon it reaches 93°C (fps drop and the temperature stop rising without acceleration of the fan), but it should only happens programs like furmark.

 

checked current application instruction on their website, apparently i was reading the back of the packaging instructions for a tube i've apparently had for a LOOOOONG time mostly unused. kind of makes me wonder A) when they changed that, and B) if that particular container is still usable. /shrugs. cheap enough to pick up more.

either way, point to you on that one dallers. apparently only the tubes from more than four years ago had alcohol with a 92% pure diamond by weight mix(current mix claims 94% diamond) and required curing out the alchohol added. newer stuff doesn't.

 

i tried the peanut sized.. the same way i did it on my CPU.. the 5mm pea sized ball method.. i did some research on proper application and did basically everything i learned about the ICD7 TIM. I blow out the dust once in a while so i dont think its a dust blockage. And no i would never bend the grills of the heat sink.

still same 93C results.. i guess throttling starts there thats why it doesnt go any higher.. but this time it takes a while to reach that high.. as to the other guy who was saying application was successful.. i dunno.. maybe?

furmark does run your CPU at a 100% full blast all the time.. so maybe it was set to really stress the GPU out and produce way too much heat the cooling system cant dissipate heat fast enough to cool it down.

Like what Megaltariak said, i guess maybe he's right(again . . maybe).

the only difference this TIM would offer would be the time it would take to reach max load temps. (Maybe) theres an increase of heat dissipation. or maybe the cooling system itself on a g53jw isnt enough to properly dissipate the all the heat produced.

what im saying is...maybe the cooling system on the GPU cant expell heat FASTER than it is produced..

Any thoughts guys?

 

I did a repaste on the gpu aawhile ago and it had very lil affect on my problems. It took getting the new V bios to get it to eliminate all gsod issues and then a clean install to get rid of a lock up issues. I didnt have any of this issues till all of a sudden I started getting the overheat shut downs. After this I ve been stable for prolly 2 weeks solid

 

I have the similar temperatures for my G53SW. I did repaste with MX 4.
93-92 in furmark, get hot very quickly.
85, my highest temperatures playing games.

My clocks are the same of 560M. 775-1500-1550

 

As stated, the G53 is not as efficient as it's larger G7 cousins. The G53 I had here would ramp up to 91C in Furry Donuts.

 

Lol Furry Donuts? Is that a game? Sounds cute