Noob here - how to tell if I did new thermal paste OK

mrshnatter

n00b
Joined
Mar 23, 2020
Messages
2
took the heat sink off a 5? year old nVidia video card (GeForce GTX550 Ti 1GB) that came with an HP desktop. the card has 2 DVI ports. I only have 1 DVI cable connected (does that not work it at hard?)

I had some arctic silver heat conducting paste that's a few years old that I used after cleaning the surfaces.

What's a good way to know if I did a good job with the thermal paste?

I ran 1 stress test app - OCCT v5.5.5 and after 45 minutes it says the GPU temp is 75 deg C, while it says other temps in the box are in the 30s.

Anyone know a good number for temp to tell if I did a good job on the thermal paste? With OCCT still running after 45 minutes,the GPU has gotten to steady state at 78 deg.

But that's 1 monitor connected to it. It has 2 DVI and 1 HDMI port. If it's actually feeding 2 - 3 monitors, would it be working harder?

Here's a couple pics of 2 apps I ran (concurrently) and their results

https://www.dropbox.com/s/7y0702mu5lv5883/1.PNG?dl=0

https://www.dropbox.com/s/ma1hyl96elh0fhu/2.PNG?dl=0

I also put furmark on the machine. it says GPU is at 100%, mem at 40% and fan 49%. letting that run a couple hours, I get to about 79 deg. C.

a) Am I working it as much as possible with that 1 monitor?
b) does it seem that 80C is ok for the new paste I put on it?

Interesting - if I stop the test and start it back up in a minute or so, it starts at mid 50s temperature (ie, it cools off fast?).

thanks!
 
Arctic Cooling says Arctic Silver 5 has a shelf life of 12 months. While those temperatures are fine I'd buy a new small tube and redo it with fresh stuff. MX-4 would actually be preferable on a GPU over AS-5.

How did you apply it? I like to first spread a thin layer on the GPU surface first and then apply a drop in the center of varying size depending on GPU (2080 Ti is a big chip so I would use about a "pea" on it). MX-4 is non-conductive and non-capacitive unlike AS-5, so you wouldn't have to worry about spillover.
 
thanks. Yeah, rather than try to describe the amount I used, I was hoping to tell if the system is working OK or not. I got a 2nd monitor hooked up using DVI and running furmark. it says gpu 100% mem 45%, fan 53%. Interesting though, the video of furmark is on 1 screen only. The other window stays the windows desktop / furmark control panel is there. Is there a way to get it to be doing graphics on both screens?

Steady state after 30 minutes is 79C.
 
NVIDIA builds protections into its driver for Furmark. You'd get a more realistic temperature if you run something like the Heaven benchmark on a loop. But like I said, those temperatures are fine. You wouldn't need to worry about it unless you were pushing 90 Celsius in a stress test.
 
Arctic Cooling says Arctic Silver 5 has a shelf life of 12 months. While those temperatures are fine I'd buy a new small tube and redo it with fresh stuff. MX-4 would actually be preferable on a GPU over AS-5.
Where did you get that info? When I contacted their support team about my 5y/o tube of AS5 they said it didnt have a shelf life.
They did note it might separate after years of sitting in the same position but wouldn’t affect its properties if remixed.
 
GTX 500 series uses a CPU like- Heatspreader shim, to "protect" the GPU die, you have to remove it to apply the new paste, much like a CPU-Delid method nowadays, you will be just wasting your time if you only change the exterior paste and not the internal from the GPU die..
 
i've used old AS5 and its worked fine. my old go-to for attaching ram heatsinks to a video card was equal parts of AS5 and arctic alumina. the alumina was a ceramic epoxy so there was no way to get the heatsinks back. mix in some AS5 and it becomes slightly rubbery. it'll still hold, but you can peel the stuff off. you can imagine that i don't use that stuff often.

as far as your temps go, you certainly weren't wrong to pull the HS and reapply some TIM. however you can't expect miracles with the stock cooler. improving the TIM usually only gets you around 2ºC. of course there are cases where its far more but that is because of a very nasty cheap TIM used from factory or something wrong with the mounting. intel comes to mind here. doing the delid on my skylake and some liquid metal TIM dropped my temps by over 30ºC. that's a design or manufacturing defect.

so what i'm getting at is see if you can look up what the normal temps are for that card. if you're within the ballpark, don't worry about it.
 
Hard to say for a lot of reasons the card isn’t some kinda power hungry monster like the GTX580 so temps shouldn’t be too bad. I’ve never heard of multiple monitors mattering over GPU performance unless you are running them all in some surround display setup. Also what is your ambient room temp? Case flow? Fan speed? Clock speed? Memory speed? All of these variables change and will effect your temps. If you want some good weigh in here I would recommend you take down all of this info also I would say just a 40min burst of anything isn’t really steady state temps but that might just be personal preference.
 
Where did you get that info? When I contacted their support team about my 5y/o tube of AS5 they said it didnt have a shelf life.
They did note it might separate after years of sitting in the same position but wouldn’t affect its properties if remixed.
It was from my own contact with their support team around... Holy crap, 15 years ago. I don't know how one would go about remixing it without making a mess unless you got a container of the stuff instead of a syringe like most people.
 
You squeeze out the amount and swirl it is what I did, it rested horizontally so the grease/oil separation came out semi evenly.
My contact with them was probably 5y ago, I used that big tube up years ago though, I'm on MX-4 now as I got annoyed with the burnin time and wanted final results faster.
 
Back
Top