an investigation into repasting laptops (or desktops) and invitation for you to do the same.

[manikyath]

this one has been long coming, and since i had a severe lack of internet this morning, i decided to make the plunge and think of a format.

 

kind of as an introduction to why, i just want to mention that the only reason i'm making this thread is because i get endlessly annoyed by people sticking "repaste CPU/GPU/your grandma" all over threads about issues that in some cases arent even related to thermals, and often from people who have no idea about the improvements (or lack thereof) that a repaste may have.

 

comes to something i've promised to do for a long time: repasting my trusty old HP compaq 6820s.

for those unfamiliar, they look about something like this:

 

mine has been with me for almost 10 years now, and has seen the worst a laptop can deal with.

- it's been in some seriously bad sea weather, at one point i had it held down to the table i was working at to avoid it flying off from the waves.

- probably a good 4 years of that 10 year age were spent slammed at 95°c when powered on

- this thing has seen long workdays, games it was never meant to run, very cold enviroments, very hot eviroments, agression, careless transport, etc.

- in those 10 years, the heatsink has never come off the cpu.

 

all that made this the ideal specimen for my test, because if any laptop is gonna see improvement, this thing has gotta be it.

 

my "aim for disappointment" was 5°c improvement, anything less was a dead on disappointment, i was hoping to see at least a 10°c difference to call any validity to repasting as a remedy to problems.

 

something else that made this thing a good specimen, is that dust cleans out relatively easy if you have half an hour and a screwdriver, meaning after the warranty on this thing ran out, i gave it a propper dusting, and it ran dust-free ever since. i couldnt be bothered to document the entire disassembly/reassembly process, but i got some glam shots for the fanboys:

Spoiler

found a date code that was rather helpful with deciding age, when i got this thing it was pretty much the latest and greatest, so this is probably a *pretty* good indicator:

the money shot of the removed motherboard:

and something that may amuse some, and help me out with arguments with others, the air intake.

when i discovered this, i also realised why it was so bad at handling first person games, because on the other side, is where WASD (or ZQSD in my case) is located:

that aside, in came the benchmarking process:

 

first, i loaded it up with hwinfo, and 7zip as my tools of choice. hwinfo has proven a very reliable sensor readout tool for me, and 7zip remains my "real world scenario cpu benchmark" of choice.

 

then, i let it run with 7zip's built in benchmark for a few hours until temperatures stabilized, took note, and then let it cool down to take note of idle temps.

 

after that, i shut down, took out the screwdriver, and replaced the original thermal paste, which was about the consistency of a brick house after 10 years, with some fresh artic MX-4. because smart people use non-conductive thermal paste.

 

bolt it back together, with the appropriate "one screw left over" , and repeat the same benchmarking process...

 

with as a result.. a mere 25% disappointment;

idle temps went an acceptable 5°c down, mind you it's winter, and a >30°c delta between ambient and system temperatures isnt very great either way

load temps.. were a strangely different story.. where core 0 went down an acceptable 6 degrees, but core 1 only went down 4°c, in essence disappointing me.

 

either HP used some god grade thermal paste on this thing, or the whole repasting stuff is all.. mediocre at best. that's where the invitation for you to do the same comes in:

 

if you have an old banger laying around that's running a little hot, i invite you to do the following:

 

1: if it's dusty, benchmark while it's dusty, and include those benchmarks in the process

2: clean it out as well as you can, bonus points for finding hidden pockets of dust.

3: benchmark while clean, take note of results

4: repaste, if you encounter any dust while disassembly for repasting, that means you didnt clean very well, and most likely will also change your result.

based on nothing at all, i feel like this is the main reason for people attesting a huge temperature drop to repasting.

5: run benchmarks again with the repaste, take note of results.

 

i really want to hear of everyone's experiences in this regard, make sure to include a brief note of age, history, and procedure. i'll leave it to your creativity to prove validity of results, there's nothing to be won by faking your results other than sounding like donald trump talking about terrorist attacks in sweden not a single swede knew about.

[You_are_a_cunt]

I'm in, as soon as I get replacement hinges for my old K50, which has been conveniently gathering dust for the past couple of years. Cya in about 2-6 months! (which is how long I expect the shipping will take, based on my current waiting time for 6 orders to arrive from various parts of Asia)

[zMeul]

you only repasted? not cleaned the rad and fan?

[manikyath]

Just now, zMeul said:

you only repasted? not cleaned the rad?

this laptop is shocking levels of clean to be 10 years old.

 

the thermal paste however.. kinda "crumbled" off instead of "wiped off"

[manikyath]

4 minutes ago, huilun02 said:

Last time I repasted my 290X Lightning it dropped 10c

Either MSI put bad paste or it got deep fried by the Hawaii over time.

I recommend repasting, even for new products.

or the heat sink wasnt screwed down properly, or got twisted a bit loose over time from sag.

 

that said, i could jump at it and repaste my GTX970 strix for shits and giggles

i'll need to find a reliable temperature test for that tho, i feel like valley wont really cut it.

[manikyath]

Just now, huilun02 said:

justdoit.gif

my main rig isnt as accessible as preferable, downsides of having your rig under your desk include wrecking your back each time you do maintenance

[manikyath]

so.. i did a derp..

 

i repasted my GTX970, having totally forgotten to do thermal testig beforehand, because i regularly do valley runs either way, thing is, valley doesnt track thermals

 

that said.. memory is pretty well for me, because i do kinda have rainmeter skins all over the place. off memory, it cant be more than a few degrees difference, and valley is within the same margin of error it usually is, so GPU boost isnt doing anything different either.

[Mira Yurizaki]

1 hour ago, manikyath said:

this laptop is shocking levels of clean to be 10 years old.

 

the thermal paste however.. kinda "crumbled" off instead of "wiped off"

That style of thermal interface material is more like parafin wax than a "paste". It's meant to "melt" and fill in the gaps, only to solidify again after cooling down.