Test Setup and Test
An Intel Core i9-13900K was used, which has been outfitted with paste and coolers so many times that it doesn’t bend any further. Meaning, it has surely reached the final stage of its spondylosis, ensuring reproducible mounting. I throttled the unit to exactly 200 watts to prevent the CPU with the old paste from hitting the thermal limit, which would significantly distort the delta to the current paste. I used a Be Quiet! Pure Loop II 360 with a fixed fan speed of 66%. It handles this easily and has only two mounting points with a fixed stop. Reproducibility trumps beauty.
For the comparison, I read several different temperatures, as the deltas also vary. Starting with the CPU temperature, the average of everything. The delta between the pastes is about 8 to 9 degrees, which is already significant. Mind you, a durability-focused reference paste from the industry with a guaranteed 4 W/mK performs very similarly (poorly) and ends up around 85 °C.
In the comparison of the computing units, the delta is only about 6 degrees, but that’s still a lot. With modern, good pastes, we measure just 1 to 2 degrees difference, if any.
Finally, comparing the so-called socket temperature, there’s about a 5-degree difference. These are still significant differences, but not quite as extreme.
We see that the old paste, even if it had been new, never stood a chance. Although the real question was whether it still works today. And yes, it does – within its somewhat limited capabilities.
Conclusion
Lies have short legs, or in this case, small tubes. The fact that as a reviewer today, you almost can’t avoid material testing, is somewhat sobering. Also, the performance of the then “high-performance thermal paste” can be depressing. Compared to a TCTG-4.0 from MCT, which as an industrial paste has a similar composition to the Arctic Paste, the old Silicone Paste performs very similarly and thus can be classified in the range of about 4 W/mK. It shouldn’t have aged significantly, even though such pastes are typically given a maximum shelf life of 12 months (ideally 3 to 6 months). The false claim of “20% carbon” back then was simply bold, but no one could prove it. Once again, this is a reason to look more closely in the future.
The counter-test with the Apex shows what’s technically possible today. However, the performance differences in the upper price range have become smaller over the years, as long as they are pastes with more than 70% corundum. Mixtures of corundum and ZnO also perform really well, and often the application skills make more of a difference than the minimal differences in actual thermal conductivity. Because that is also difficult to verify, so it remains, unfortunately, a sad playground for marketing.
I’ll put it this way: anyone who spends 10 euros or more for a gram of thermal paste today has something to spare and is a marketing victim (often misled by influencers), as it can be done cheaper and in larger tubes for much less money. But you shouldn’t go too cheap either. But that wasn’t really the topic today. The fact is that the Arctic Cooling Silicone Paste still performs as it did a whopping 20 years ago. Compared to today’s pastes, however, it doesn’t stand a chance. And that was also to be proven.
50 Antworten
Kommentar
Lade neue Kommentare
Urgestein
1
Veteran
1
Urgestein
1
1
Urgestein
1
Urgestein
Urgestein
Urgestein
Urgestein
1
Urgestein
Moderator
Urgestein
Mitglied
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →