Testing reveals that some Nvidia AIB partners are using cheap thermal paste resulting in high GPU temperatures

Stozzy

Posts: 13   +1
Too hot to handle Igor's Lab investigated community feedback that some owners of GeForce RTX 40 series AIB cards are hitting hotspot temperatures of 100° C after a few months of operating within acceptable ranges. GPU hotspot temperature refers to the maximum temperature recorded in a specific area of a GPU, usually either on the GPU die itself or the voltage regulators (VRM).

Igor Wallossek conducted his test using the thermal paste from a brand new Manli RTX 4080 Gallardo. Wallossek used the ASTM D5470-17 standard for conducting thermal conductivity and thermal resistance with his thermal analyzer system, TIMA5. The thermal paste was tested at 60°C with various thickness and pressures, to mimic variability in thermal paste application and heatsink to GPU contact force.

The results show that at very thin layers, the thermal conductivity is very competitive with other thermal solutions, such as Thermal Grizzly Paste X and Thermalright TF8. However, as the application layer becomes thicker, the thermal conductivity worsens more rapidly compared to Thermal Grizzly Paste X. It is also important to note that Wallossek had to apply significantly more pressure to achieve the best thermal conductivity.

The Gallardo paste composition was analyzed under a microscope. Wallossek found that the paste's oily texture did not sustain itself and dried rather quickly. There were large particles of aluminum oxide (up to 16 µm), which perform thermally well at first but cause large gaps in the paste matrix, allowing the oil and smaller particles to bleed out, causing the paste to dissolve and dry quickly.

Wallossek concluded the following regarding the paste's composition: "In addition to the vast quantities of mostly coarsely ground aluminum oxide (Al2O3) and a lot of silicone oil (silicon, hydrogen, oxygen), we also have some finer zinc oxide (ZnO) as a cheap filler material. I can't find any carbon that might have suggested a more sophisticated silicone (side chains). So, it really is just the simplest oil."

Owners who are experiencing high temps on their GPUs could attempt to fix it by replacing the thermal paste on their card. However, this is not as easy as replacing the paste on your CPU, as it requires a full teardown of the GPU board. It's not impossible, but it is not practical for most.

It is unconfirmed if this thermal paste tested is affecting only higher-end Nvidia GPUs or if it spans all AIB GPU manufacturers. It is not uncommon for companies to cut corners to increase margins at the expense of the customer. It would be interesting to see additional testing across a larger range of AIB cards from Nvidia, AMD, and Intel to determine if this is an exception or the norm.

Permalink to story:

 
I’m surprised AIBs aren’t just advertising “pasted by Arctic” or whatever on their products. Seems like a wasted marketing opportunity unless it really was significantly more expensive or difficult to use consumer pastes compared to bulk pastes.
 
Vendors it's more cost effective to save a dollar on thermal paste then RMA the graphics cards that cost upwards of 2k usd. 🤪.
Luckily I have 3d mark score from 2022 and will rerun the test to see if anything was affected on my 4090 suprim liquid. The card barely reaches 60 degrees on 3ghz and +200 hz on the vram ( 450 watt) and often hovers around low to mid 50s in rt titles.
I suggest everyone rerun a previous benchmark and compare the results. The key is significant performance loss of 5% or greater. Things to consider; In the summer time ambient room temperature is definitely often as high as 10 to 15 degrees warmer ( in Fahrenheit) then in November when I purchased it so that might affect the results. 69*f vs 83*f now. Air conditioning will improve this to 75ish but might still affect results. I'll post my findings after work.
 
I want to bring something to your attention . Igor s Lab sometimes issue fake news . They claimed RTX 4060 was drawing 51W in idle , apparently using software tools , instead of measuring the power through PCI-E slot and PCI-E connector . They are not guys to be trusted . As for RTX 4060 , idle pwer of the aforementioned card is 11-14W . Just the card doesnt have sensor for power measuring but it s not a deal breaker .
 
Inno3D , Zotac and Manli are owned by a Hong Kong company with a headquarter based in Taiwan . I guess , the other 2 brands might be affected .
 
Last edited:
The new arctic paste is really good. I wonder why they even try to save money lol. It was like 4 bucks to buy here. It gave me much better performance than my older paste, I honestly couldn't believe it. As much as going from D15 to D15 part 2, from Noctua.
 
I want to condemn them, but I recall Nvidia leaves them so little margin that I can't be too harsh to them.\
Leave them a bit more money, green monster.
 
Vendors it's more cost effective to save a dollar on thermal paste then RMA the graphics cards that cost upwards of 2k usd. 🤪.
Luckily I have 3d mark score from 2022 and will rerun the test to see if anything was affected on my 4090 suprim liquid. The card barely reaches 60 degrees on 3ghz and +200 hz on the vram ( 450 watt) and often hovers around low to mid 50s in rt titles.
I suggest everyone rerun a previous benchmark and compare the results. The key is significant performance loss of 5% or greater. Things to consider; In the summer time ambient room temperature is definitely often as high as 10 to 15 degrees warmer ( in Fahrenheit) then in November when I purchased it so that might affect the results. 69*f vs 83*f now. Air conditioning will improve this to 75ish but might still affect results. I'll post my findings after work.

This. We should promote a culture of saving those early benchmark results on our new PC parts, then occasionally re-run them. Try to recreate the conditions as much as reasonable. This is good, professional practice and can alert us to hidden issues. It will also eventually put pressure on manufacturers to make stuff last. Right now, manufacturers are rewarded for making stuff shiny and fancy when you first get a thing.

I also love the comment that manufacturers are missing an opportunity to use high end paste, then marketing that fact.
 
Sorry for the delay it seems there is no significant loss since 11/22 to present and estimating 1500 hours of use.
 

Attachments

  • results 4090 over 2 years.png
    results 4090 over 2 years.png
    471.9 KB · Views: 3
Back