Run away train! 100C temps... on 7600k

[EarthDog]

Ditto. As we pointed out earlier, the hyper 212 is not actually "the best" air cooler for that CPU. Not by a long shot.

Yep... it is the bang for the buck cooler at $30. It does not compete with Noctua NHD-15, etc......which does come within a C or two to most 2x120mm AIOs.

 

[SPL Tech]

It seems the largest issue is getting the heat from the CPU and into the heatsink very quickly because temperatures in a CPU change by massive amounts in very short periods of time. That makes big spikes and the heatsink cannot flatten out those spikes fast enough. For example, if the heat was being transferred to the heatsink quickly and efficiently, I would expect the heatsink to be hot. My GPU runs at around 80C and it gets extremely hot. However, if my CPU is running at 90C, still it does not get very warm. Whatever heat does come from it is actually mostly from the GPU radiating up onto the CPU heatsink. So in reality the CPU is at 90C but the heatsink is barely warm to the touch. That tells me that CPU heatsinks in general just are not that effective at transferring heat and I think the main issue is the contact point between the die, IHS and heat sink. There is too much of an inefficiency there and not enough heat is being drawn from the CPU cores which is why they get insanely hot while the heatsink is basically cold.


I envision the future is something where the CPU itself is submerged in water. Like a water cooling block that claps directly to the mobo and makes a watertight seal around the CPU, and then floods the entire CPU socket with water. Distilled water does not conduct electricity so it is a plausible concept. Either that or CPU manufacturers need to start making larger CPU dies so the heat is spread out over more surface area.

 

[EarthDog]

Sort of... you are correct the trouble comes from getting the heat out of the die. A smaller die is the big issue. That said, temperatures skyrocketing quickly is normal...cpus...gpus... etc. So, it isnt the heatsink as you said, so much as it is the die size.

Also my heatsinks are surely warm to the touch... hell my 3x120mm rad is warm to the touch after a while...

 

[trents]

It seems the largest issue is getting the heat from the CPU and into the heatsink very quickly because temperatures in a CPU change by massive amounts in very short periods of time. That makes big spikes and the heatsink cannot flatten out those spikes fast enough. For example, if the heat was being transferred to the heatsink quickly and efficiently, I would expect the heatsink to be hot. My GPU runs at around 80C and it gets extremely hot. However, if my CPU is running at 90C, still it does not get very warm. Whatever heat does come from it is actually mostly from the GPU radiating up onto the CPU heatsink. So in reality the CPU is at 90C but the heatsink is barely warm to the touch. That tells me that CPU heatsinks in general just are not that effective at transferring heat and I think the main issue is the contact point between the die, IHS and heat sink. There is too much of an inefficiency there and not enough heat is being drawn from the CPU cores which is why they get insanely hot while the heatsink is basically cold.


I envision the future is something where the CPU itself is submerged in water. Like a water cooling block that claps directly to the mobo and makes a watertight seal around the CPU, and then floods the entire CPU socket with water. Distilled water does not conduct electricity so it is a plausible concept. Either that or CPU manufacturers need to start making larger CPU dies so the heat is spread out over more surface area.

Actually, they're trying to shrink them or at least stuff more transistors into the same space. That's what going with smaller and smaller nano fabrication is all about and it allows the CPU to run at a lower voltage which in turn results in lower temps, all other things being equal of course - which they aren't.

 

[EarthDog]

More stuff in a smaller space. Process shrink... physical die size shrink.. overall more dense. It's easier to get xxxW out of something size of a quarter than a dime.

 

[SPL Tech]

Actually, they're trying to shrink them or at least stuff more transistors into the same space. That's what going with smaller and smaller nano fabrication is all about and it allows the CPU to run at a lower voltage which in turn results in lower temps, all other things being equal of course - which they aren't.

But they arnt making CPUs for phones here. They are for desktop computers which has a massive amount of room. There is no need for the CPU to be physically smaller. The goal here should be performance, not size. That's the entire idea behind a desktop computer otherwise we would all just use phones and nothing else. The CPU die takes up basically absolutely no space relative the sum total of available space in a standard ATX case. It doesent even take up that much space relative to the motherboard.

 

[freeagent]

32nm is easy to cool..

22nm getting kinda tough..

Under 22nm I would imagine would be a bit of a bear.

212 sucks, never was a good cooler, just better than stock.

Edit:

And it was cheap. Now it seems like you pay for its legendary status. For literally 8 dollars more you can get a True Spirit 140 Power which will destroy it in every possible way..

 

[EarthDog]

But they arnt making CPUs for phones here. They are for desktop computers which has a massive amount of room. There is no need for the CPU to be physically smaller. The goal here should be performance, not size. That's the entire idea behind a desktop computer otherwise we would all just use phones and nothing else. The CPU die takes up basically absolutely no space relative the sum total of available space in a standard ATX case. It doesent even take up that much space relative to the motherboard.

If only it worked that way...

Larger dies means less profit off the same wafer. The smaller the process, the more inherent improvements, but thermal mitigation becomes an issue. Smaller is progress in many ways.