As desktop processors were first crossing the Gigahertz level, it seemed for a while that there was nowhere to go but up. But clock speed progress eventually ground to a halt, not because of anything to do with the speed itself but rather because of the power requirements and the heat all that power generated. Even with the now-common fans and massive heatsinks, along with some sporadic water cooling, heat remains a limiting factor that often throttles current processors.
Part of the problem with liquid cooling solutions is that they’re limited by having to get the heat out of the chip and into the water in the first place. That has led some researchers to consider running the liquid through the chip itself. Now, some researchers from Switzerland have designed the chip and cooling system as a single unit, with on-chip liquid channels placed next to the hottest parts of the chip. The results are an impressive boost in heat-limited performance.
This seems like a very logical next step for watercooling and processor cooling in general, but this is far from easy. This article highlights that we are getting closer, though.
It’s still a niche thing though.
There’s only so much you can do if the heat’s still getting generated and I doubt this will help things like smartphones, or PCs like mine that are designed for a room with no air conditioning and a specific thermal budget. (I always pick CPUs with a 65W TDP for that reason.)
You’ll never get a chip that runs without making heat. Heat is produced when the transistors switch (as well as resistance from traces etc), so having a chip that runs cold is impossible.
And people who need a decent amount of processing power (or many cores), you’ll still need a chip that puts out a fair amount of heat. That’s why so many chips today have TDP’s in the hundreds of watts
> You’ll never get a chip that runs without making heat
What if someone managed (“insert magic smoke here”) to create a room temperature superconductor? (well, over room temp)
This is unlikely to be seen at the consumer level due to some basic physics. Water in contact with metal for a few years becomes acidic, and thus needs replaced before that happens. Chips with these systems must never be allowed to freeze due to thermal expansion of water breaking it apart from the inside. I’ve already heard of water cooling AIO systems breaking from being left in the car during freezing weather. Laptops you can’t carry around in cold weather is certainly a major deal breaker.
Water is only one type of liquid, and each liquid has different properties. Even just adding a common additive to water will change its properties. One example is adding ethylene glycol, which is used in “anti-freeze” to prevent a liquid cooled car engine’s coolant from freezing during freezing weather.
I don’t know much about chemistry, but does this apply to any metal?
> chip must never be allowed to freeze due to thermal expansion of water breaking it apart from the inside
Future headline: “Intel announced revolutionary anti-freezeing “Liquid-Ice 9 Inside” cooling solution… Computing Revolution expected.
Day 2 headline: Technician accidentally drops prototype phone in lab. Apocalypse forecast 4 weeks. (Buying) run on SpaceX stock.
Day 3 headline: (Physical) run on SpaceX launch facilities. Some rioting, 5 injured, 1 dead. Musk quick-planning mass planetary evacuation.
Day 28 headline (from the Moon): Earth-Moon-Lagrange-point British/EU backed fusion reaction construction begun. Mars colonization to start in six months. Titan expected by 2032. Earth abandoned to its fate..
Don’t care, since my AMD Athlon XP I’m now constantly seeking for the most power efficiency (VIA C7, AMD APU, …) not the most power hungry beast that might deliver huge performance… some days. AMD Ryzen 4000 seems pretty neat these days, a 15W TDP able to outperform Intel counterparts eating 3 times that much. Which allows me to even lower the bar of my expectations. Good, good…
Why water?
We know we can use alternative liquids. For example PCs can be fully submerged into mineral oil:
https://www.pugetsystems.com/submerged.php
https://www.youtube.com/watch?v=tUBvWXH1hLs&feature=emb_logo
Water comes with all the issues affecting electrical systems. It is corrosive, has a low boiling point, and of course conducts electricity.
You can use distilled water, which does not conduct electricity…
Briefly. If it touches any tracks with a DC voltage across them it will be full of ions within seconds.
javiercero1,
I found this “Resistivity of some typical waters”
https://corrosion-doctors.org/Corrosion-Kinetics/Ohmic-drop-water.htm
https://www.labmanager.com/white-papers-and-application-notes/resistivity-conductivity-measurement-of-purified-water-19691
Distilled water probably wouldn’t cause much damage via electrical short. Many low voltage circuits will run ok underwater, the currents are generally too low to cause direct damage. However the electric currents will start an electrolysis processes, which immediately starts to corrode proportional to the current, which is bad.
Manufacturers could add detection circuits to shut down the computer before damage occurs, but it would need to be done on the motherboard and maybe other components, so unless it were a normal common failure mode, it’s not clear that it’s worth doing.
Check Linus Tech Tip videos on their mineral oil pc. Acrylic doesn’t hold up, the components don’t hold up, and a patent troll stopped Puget from selling any more kits in 2014.
> patent troll stopped Puget from selling any more kits in 2014
Any hope that all the major CPU makers could sue their behinds for inhibiting innovation? Something in an EU copyright directive maybe?
Yes, I heard about the patent issues.
They mention running a system over five years:
https://youtu.be/yAZRPXWy_nM?t=673
So it should be pretty safe.
sukru,
I was curious and watched the video you linked.
The guy linus was with briefly mentioned that he burnt out a GPU by taking it out of oil. Apparently it had eaten away at the thermal compound.
There are new videos where linus reviews the system and he gets less optimistic over time. He tested the components again and while it worked initially it appears his GPU became damaged as well (jump towards the end to see).
“Parts from the Oil-cooled PC – Do they still work???”
http://www.youtube.com/watch?v=sSnGmAqQaFs
The also had a maintenance video:
http://www.youtube.com/watch?v=0ZrXwq5MHB0
A few times they talk about how it makes rubber brittle, including wires. All and all, it sounds like while it can work, it’s not really worth the trouble and hassle in the long run unless you really want to be different. Mineral oil might make an ok thermal fluid, but from what I’ve seen I’m not tempted to submerge my PC in it.
Alfman,
You are right. I think we got sidetracked. My initial point was using oil instead of water “in” the pipes, and this was just an extreme example. Even if it leaks, parts would still function fine.
Isn’t this problem already solved with PC water cooling anyways? I would guess whatever solutions are used in normal water blocks applications might already fit the bill…? As I understand it some of the solutions you get for these applications are not corrosive for this very reason.
If you really wanted to go extreme, you could use a refrigerant to achieve below ambient pressure with a compressor such as in a refrigerator.
These guys claim to have overclocked to 7Ghz in 2017 using cryogenic liquids and apparently the CPU was so cold they had to use a blow torch to heat it up. It’ll make you laugh if you haven’t seen it.
http://www.youtube.com/watch?v=WwJvHJ1hyto
It shows that you can get a lot more performance with active cooling. Granted you’d want a properly regulated rig rather than manually switching between nitrogen and a blow torch, haha.
(Intel ad in 2024) “TorchTech Inside”
Technology like this won’t hit the consumer market for years, if ever. This is more a technology that will find it’s way into mainframes and supercomputers.
One of the biggest cooling bottlenecks in the consumer water-cooling space nowadays is the metal between the silicon, and the water. Many people get around that by removing the IHS from chips, and/or lapping the cooling blocks. By bringing the water-cooling inside the chip, you negate, or at least mitigate, that lump of metal between the water and silicon, allowing the chip to run faster and cooler.
This could be a game changer in the high performance computing space, and i’d expect IBM to use this in their z/Architecture mainframes in the next 5 years
The123king,
It’s a bottleneck primarily because we’re using an ambient temperature coolant, which is really lousy especially in the summer. When you chill the coolant, suddenly that bottleneck nearly goes away so that you can move on to the next bottleneck. While the video used frozen nitrogen, that’s more extreme than necessary. and enthusiasts can and do get higher performance just by using refrigeration.
To be clear, this adds a new source of inefficiency and mainstream manufacturers will probably try to stick to ambient cooling as long as possible, but with hot enough chips refrigeration is going to become inevitable and I won’t be surprise to see the custom water-cooling market embrace it. Although…the noise could be a con, Peltier cooling?
The concept from the title scares me. water expands with heat. What’s to prevent it from cracking and seeping into other components of the chip at a micro level? or change to a gaseous state upon overheating and burn the chip itself entirely? This concept needs to mature a lot before I can put money on it. LOL