Why does everyone, including the manufacturers themselves, advocate for an inefficient method of applying thermal paste?
>>53575071
Because broken cpu = more money for intel/amd
to get you to buy more thermal paste
it's not about what works best it's about what works but how can we get the consumer to buy more
at the crux of it they're trying to sell you more stuff, some of which is snake oil, some of which isn't.
Heat distribution is more efficient when the whole processor is covered. Basic physics.
>>53575071
Line/X method is the best. Google the core layout, and put the line or X accordingly.
>>53575071
Because the temperature differences are so miniscule it doesn't even matter.
>>53575071
because nobody care you sperg, the difference doesn't matter
>>53575071
>inefficient
Thermals are not affected by the spreading method used at all. All that matters is that enough paste was used.
the actual cpu die is not the size of the spreader retard.
its more efficient to have perfect local exchange than bad total contact
>>53575071
If anyone can link the chart with CPU temps and different styles of spreading paste, that would be great
X - style proved to be the best
Spreading the paste causes air bubbles when the heat sink is applied.
https://www.youtube.com/watch?v=EyXLu1Ms-q4&t=1m53s
>>53575071
Card method may leave some a air bubble
Dot method doesn't
And thermal past is not a electric conductor so there is no electric risk if it leak on the border
>>53575267
>thermal past is not a electric conductor
Some of them are and some of them aren't
Corners don't matter as much as air bubbles. Pre-spreading the paste basically guarantees trapped air in the center.
>>53575071
Shit doesn't seem to really matter you guys
https://www.youtube.com/watch?v=r2MEAnZ3swQ
Do what feels right just not to much or to little.
>>53575209
But the difference is probably negligible.
With the card method, you risk contaminating the paste, introducing air-bubbles, risk bald spots on key areas, over-applicating, etc.
Also, the card method is fine if you can mount it square on the first try, but 4/5 the cooler is going to slide on you.
>>53575353
True. You'd have to leave a perfect uniform layer of paste before putting the heatsink.
>>53575071
Because it's more important to have the thermal paste covering the area right over the actual CPU die rather than the entirety of the heat-spreader.
Pentium-Quad Core i7s have a rectangular die that's positioned N-S on the substrate. A line method applied roughly at the center will suffice.
AMD and Core i7/Xeons with more than 4 cores have a boxy rectangle shape that fills out most of the substrate. A simple X or asterix in the center will do.
Oh no, my cpu may or may not be 0.5c higher.
>>53575071
Illustrated answer to your question OP.
>>53575512
Basic physics laws state that the area of the heatspreader around the CPU die can't get hotter than the area directly in contact with the CPU die, so the thermal contact in that area is the most important.
If you are relying on the area around that for thermal dissipation then you're basically doing it wrong.
>>53575071
>>53575789
>>53575512
why can't AMD and Intel sell processors without IHS, just like their laptop models?
>>53575789
>I have no idea how a heat spreader works
I usually tint, then do dot and spread. Works bretty good for me.
>>53575828
Because dies are fairly fragile and you couldn't mount your massive tower cooler on them.
If you have ever taken apart a laptop or a GPU where they use bare dies you would notice that they hardly use any mounting pressure at all compared to a normal desktop cooler.
>>53575353
this + various simple tests.
in the end it doesn't even matter.
>>53575828
The heat spreader does two things:
>carries away heat onto a much larger surface area
>distributes load on the CPU die around the die itself
The only non-IHS CPU Intel ships are found on SoC motherboards, but those have a larger passive or active heatsink/heatpipes on top of those as standard. Their mounts are incompatible with LGA115X sockets.
>>53575871
>i have no idea how physics works
>>53575071
Are you trying to square the circle?
It doesn't fucking matter
https://www.youtube.com/watch?v=r2MEAnZ3swQ