Updated: Thermal Paste does make a difference.

[Nardella]

EDIT: As pointed out by a commenter below, the test is invalid.

One of my favorite small youtubers did a video testing the importance of thermal paste. Turns out that, at least in this scenario, it is not very important! I want to see more tests!

[Skiiwee29]

Yes, thermal paste matters and DOES make a difference... this is a crock of hogwash if you ask me. 

[Supportsneedlove]

It only costs a few bucks a tube, just use the fuckin thermal taste. 

[Mira Yurizaki]

12 minutes ago, Nardella said:

One of my favorite small youtubers did a video testing the importance of thermal paste. Turns out that, at least in this scenario, it is not very important! I want to see more tests!

The person who uploaded it said this in the comments section (emphasis added)

Quote

I just noticed that the entire experiment may be rubbish... The fans on the right side (where the CPU without paste is) turn much faster than on the other side. This server can control all six fans individually. Today I powered it up with both CPUs properly "pasted" and the fans were spinning at the same speed on both sides. That means, the CPUs only have the same temperature because the server reacts to heat differences and spins up the fans that are close to the hotter CPU. And that means the two CPUs would not have the same temperature if all fans were spinning at the same speed... Conclusion: THERMAL PASTE MAKES MORE DIFFERENCE THAN THIS VIDEO SUGGESTS.

 

[Skiiwee29]

8 minutes ago, VegetableStu said:

a retraction of that scale needs a follow-up video IMO o_o

or just a removal of the video all together for spreading bad information.. most people arent going to actually read the comments for follow ups. 

[cat of all the melons]

for 45p on amazon it is not worth the risk

[Mira Yurizaki]

Depending on the part being cooled, its power consumption, and the cooler's surface, for short term testing you can get away without using thermal paste. Phil's Computer Lab admitted to not using thermal paste on CPUs because he constantly swaps them out and so applying paste would be annoying (and probably expensive) to do so every time. And the CPUs he tests tend to be at or under 45W and with bare dies for the higher power ones.

 

But for long term use, you should be using thermal paste. Unless somehow you have a bare die, a mirror finish on the heat sink where it'll contact the part, and the die and heat sink surface are insanely flat.