My GPU was reaching temps > 90°C in games like the Witcher 3 in a matter of seconds. So I decided to repaste it with Conductonaut. My laptop came with LM on the GPU and delidded CPU, but I hadn't taken of the heatsink, so I was also curious how the reseller applied the LM.
First step: take out the battery and ground yourself.
![]()
Those brown squares are little feet to improve airflow. One is missing, I know.
Opening up the case.
![]()
Removing the heatsink.
![]()
Not a surprise: the LM on the GPU die has completely dried out. Also on the heatsink, where it bonded with the copper.
Big surprise: no delidded CPU, no LM on the cpu, just a thermal pad? I will repaste the IHS/heatsink contact with Kryonaut.
Let's look closer to the GPU die.
![]()
I don't know what tape that is, but is that TIM smeared around the borders of the die?
Removing the tape...
![]()
Yes it's some kind of TIM. Removing it carefully with a swab and paper cloth. Cleaning the die.
Also the dirty heatsink.
![]()
When all the dirt is removed, I tape along the GPU die with TESA tape to prevent damage from accidental LM spillage.
![]()
Time to paint the GPU die with a tiny drop of LM, and use a bigger drop to rub in the heatsink.
Then I follow Mr. Fox's suggestion: instead of making foam barriers I make tape 'straws' to stop any LM run off.
![]()
Then I paste the CPU IHS with Kryonaut and reattach the heatsink.
Temperatures are now at an absolute low. Running the Witcher 3 doesn't increase temps above 61°C.
![]()
The highest temps got with stress testing were 71°C. Without any tweaking I get a 21k Firstrike Graphics score and that's with a conservative undervolt of 1835 at 950V which results in no power or voltage limits. The CPU runs at 4.6 all core, -130 undervolt, at max 75°C, although I haven't been heavily stress testing it.
Of course the real challenge is to see within a few weeks if temps remain stable or start increasing again if the LM dries up / pressure is insufficient - or (hopefully not) if my tape 'dam' doesn't hold.
-
BrightSmith Notebook Evangelist
-
-
yrekabakery Notebook Virtuoso
That's a very interesting looking heatsink. I don't think I've seen it before. It's got a different heatpipe arrangement than both the P775DM2/DM3 unified heatsink and the P775TM (Pascal) non-unified heatsinks.
-
BrightSmith Notebook Evangelist
-
yrekabakery Notebook Virtuoso
-
jc_denton likes this.
-
BrightSmith Notebook Evangelist
This is the one I have:
This is the other one:
The main difference is the pipe that bridges the CPU and GPU. In the second type it's a big heat pipe that connects them, while in the first case it's a smaller one. I guess it will depend on the kind of load which type will be more beneficial.
In any case it does the job of removing heat from the GPU well enough. -
Meaker@Sager Company Representative
There is that third heatpipe being thicker to the gpu heatsink side too.
-
BrightSmith Notebook Evangelist
-
Meaker@Sager Company Representative
Looks that way yes.
-
yrekabakery Notebook Virtuoso
The P775TM Pascal version is supposed to have the non-unified heatsinks I linked. Not sure why your reseller installed this one instead.
-
BrightSmith Notebook Evangelist
On the other hand, the advantages of a non-unified heatsink depend on the type of workload, as I understand it. -
Meaker@Sager Company Representative
It gives a quieter idle and better cooling under individual load.
Repasting my 775TM GPU with LM - a brief report
Discussion in 'Sager and Clevo' started by BrightSmith, Feb 21, 2020.