I'll be interested to see what they can achieve.

Kohl BaasTEC is dead tech.

Not really true.

The voyager I and Voyager II used by NASA, is still being driven by a combination of nuclear heat in combination with TEC's that power the damn space thing for over 40 years in space now sending valuable data back to earth / NASA. en.wikipedia.org/wiki/MHW-RTG

If used properly you can extract power from a TEC like for example on a stove / heater. You need alot of heat tho but it's free energy extracted from heat. I used tecs in the past too, both waterchillers and in direct cooling methods; they do their work, but they are very inefficient for 24/7 usage. It's more like a bench / OC record set. And yes a decent compressor goes well below -25 at load and -40 at idle. Derbauer made a vid on that a while ago:


If done correct; you can get a very well performing chiller / cooler.

X71200Look at the figures from an old fridge to a new one.

Better fridge ≠Better compressor. New fridges are more efficient mostly thanks to better control circuitry, better insulation, defrosters etc. Basically the compressor itself only improved a little thanks to slightly better and slightly more efficient motors and maybe some advancements in heatpipe&radiator design (but those are minor). Your new fridge uses less power only because it doesn't have to run as often, that's it. It doesn't matter how much money companies throw into TEC research, it'll only get you so far.

BlueberriesNot sure why everyone seems to think Peltier's don't work.

Of course TEC does work. In this situation they are used for the wrong reasons and with a wrong purpose.
I'm pretty sure that cherry-picked 10900K can do 5.7GHz for screenshot and CPU-Z validation, or run 5.4GHz all-core on a regular 360 rad (which they didn't show), so for me adding TEC modules just for the sake of having sub-ambient only when you are idle - is stupid. It's a giant extra point of failure in an already fragile system.

With a decent controller, you can actually set the desired idle or load temp. Thats just the beauty. No more old-skool insert-12-v into your tec or so.

I'd always wanted to pair a bunch of tecs and use them as a chiller for whatever OC project i want todo now; but mweh. You'll have one night of fun and the rest of the time it's in the corner.

Peltiers will make you a literal sweatlord, that's for sure. I see some of you are already sweating by giving examples that have nothing to do with computers on earth, like outer space projects.

silentbogoBetter fridge ≠Better compressor. New fridges are more efficient mostly thanks to better control circuitry, better insulation, defrosters etc. Basically the compressor itself only improved a little thanks to slightly better and slightly more efficient motors and maybe some advancements in heatpipe&radiator design (but those are minor). Your new fridge uses less power only because it doesn't have to run as often, that's it. It doesn't matter how much money companies throw into TEC research, it'll only get you so far.


Of course TEC does work. In this situation they are used for the wrong reasons and with a wrong purpose.
I'm pretty sure that cherry-picked 10900K can do 5.7GHz for screenshot and CPU-Z validation, or run 5.4GHz all-core on a regular 360 rad (which they didn't show), so for me adding TEC modules just for the sake of having sub-ambient only when you are idle - is stupid. It's a giant extra point of failure in an already fragile system.

the problem, the reason they don't work in a pc, is that there isn't enough working surface area for the TEC to deliver justifiable performance, all the added energy load for slightly better performance than an air cooler doesn't cut it. That's why any TEC that wasn't a standalone unit that was plumb outside the case has long since disappeared.

X71200and they have a curve that makes them slowly increase in temps unlike Intel chips that skyshoot right when you launch Cinebench.

lol, wat? My 3900X shoots straight up to whatever just as fast as my Intel.... stock, overclocked... you run a stress test on it (or cinebench) and boom, skyrocketing temps.



Anyway, this clearly isn't for anyone here, really. Few at TPU are able to 'think outside the box'... anything extreme cooling is met by a majority of users hating on it. Me, I think it's pretty cool. Is it a 24/7 solution? Not for 99.9% of users, but it can be (set post earlier about setting temps, etc). People say xx chip runs cool and blablabla, but the reality is people are thermally limited regardless so something like this can change the delta and help get that extra couple hundred Mhz out of it.

EarthDoglol, wat? My 3900X shoots straight up to whatever just as fast as my Intel.... stock, overclocked... you run a stress test on it (or cinebench) and boom, skyrocketing temps.

My 3900X both idles and increases temperatures slower than my 7900X, regardless of settings. I have spoken with other users in chats who have tested Zens against other Intels and they experienced the same. You don't know what you're talking about.

X71200My 3900X both idles and increases temperatures slower than my 7900X, regardless of settings. I have spoken with other users in chats who have tested Zens against other Intels and they experienced the same. You don't know what you're talking about.

Riiiiiiiiight... I just have one sitting here right next to an Intel... I have no idea what I'm talking about... lulz.

If you don't believe me, feel free to look at Tom's Hardware and some motherboard reviews. In it they test the VRMs and in the same chart, are CPU temps. You can easily see that both a 3900X and i9-10900K ramp up immediately.

EarthDogRiiiiiiiiight... I just have one sitting here right next to an Intel... I have no idea what I'm talking about... lulz.

Sure you do not. I have tested both chips lots of times on same cooler configurations.

The best approach using a tec would be a addition to your existing watercooling setup, a chiller for example. Downside? Extra power, and extra radiators in order to remove the heat. There where some very lit build on xtremesystems.org back then (2007 and so on) > www.xtremesystems.org/forums/forumdisplay.php?94-T-E-C-Cooling

I miss that era / time of people hacking up xtreme tec setups in their own basements.

X71200Sure you do not. I have tested both chips lots of times on same cooler configurations.

And I'm supposed to believe some rando on a forum? lol

Here you go bud.

www.tomshardware.com/reviews/gigabyte-b550-aorus-master/4
www.tomshardware.com/reviews/nzxt-n7-z490/3

Looks like a remarkably similar spike in CPU temps..........

Anyway, this is a bit OT... I digress. The info has been posted. ;)

EarthDogAnyway, this is a bit OT... I digress. The info has been posted. ;)

Seemingly you are unable to understand that a 12 core CPU does not get fully loaded in real world applications most the time, and it finishes the Cinebench which is spoken about in couple seconds. If you actually for once went ahead and checked your temps right after Cinebench loads on both the Intel and the AMD setups you talked of, you will realize the AMD gets there later. This is because it functions at lower temperatures. That some random on a forum understands more than you, it seems. Maybe try actually doing what I said rather than talking nonsense.

X71200Seemingly you are unable to understand that a 12 core CPU does not get fully loaded in real world applications most the time, and it finishes the Cinebench which is spoken about in couple seconds. If you actually for once went ahead and checked your temps right after Cinebench loads on both the Intel and the AMD setups you talked of, you will realize the AMD gets there later. This is because it functions at lower temperatures. That some random on a forum understands more than you, it seems. Maybe try actually doing what I said rather than talking nonsense.

Why are you insulting me and getting personal? WTF is your problem, man? :(

My guy, both CPUs 'skyshoot' when you start the render or run a stress test. Cinebench (rendering) uses all c/t it can get... even if you use one thread (think Super Pi or something) that one core will 'skyshoot' too.

PS - Here is ~30 seconds worth of data (from Hwinfo) showing CPU temps... please pick out the Intel over the AMD. it should be clear to anyone looking at these values that both CPUs ramp up almost immediately.
Done here... :)

@EarthDog
@X71200

You're both done here. To everyone else, please follow the forum guidelines, stay on-topic, respect each other, and lastly, avoid trolling and insulting each other and let's all try to have a better experience in this topic. Thanks! :toast:

TheUn4seenWell, damn physics didn't change in the last few years, just as it didn't since the universe came to existence (at least as far as we know).

I meant more materially. We've had some improvements in other areas; including near-room temperature superconductors, but still no alloy or composite that slightly improves Peltiers by even 1-2%.

That aside, the only major CPU cooler that is relatively modern and uses a TEC is the Phononic HEX 2.0 CPU cooler. They were pretty ambitious for the period, and it works in helping to take some heat edge off the CPU, but they never did expand into 120mm coolers despite originally mentioning they would.

Curious what EK are doing about condensation. Regardless of what tech they're using, Linus posted temperatures at or just above freezing, so condensate is by far the biggest concern I have (and my concerns over the ridiculous energy cost of TEC or the complixity/environmental concerns of pressurised refrigerant loops aren't exactly minor concerns in the first place).

Actually Hardcore Overclocking -- Rambling about Peltier cooling computer parts.

silentbogoTEC and TEG are inherently inefficient. You can't change physics.
Maybe a century later, when we get some new super-duper-magnets and cheap room-temperature superconductors, we'll be able to achieve something more adequate, like being half as efficient as phase-change... For now, let's leave it where it belongs.

Nope. It's still as inefficient as its components separately. The only thing that'll maybe improve is heat transfer, because the hot side of TEC is gonna be hotter than CPU, hence warmer water, and faster dissipation on the rad side (assuming it's not gonna get saturated)

Running at TEC at less voltage is more efficient. it will transfer more wattage then it uses.

I've got a 40mm x 40mm 225 watt TEC That uses 24 volts and 15 amps.
I also Have at 24 volt power supply with 18 amps. At max with the T delta of 64C it uses 360 watts
Running the TEC at 12 volts turn's the TEC 112.5 watt TEC It will pull 7.5 amps and use 90 watts with T delta of 32C
Running the TEC at 7 volts makes the TEC 65.6 watts it will put 4.375 amps and use 30 watts with T Delta of 18.6
running the TEC at 5 volts makes the TEC 46.87 watt it will pull 3.125 amps and use 15 watts with T delta of 13.3C

you need to add more and add the component's you're cooling wattage into it also.

Tec's biggest problem is the amount of TEC's you need vs the cost for running decent where it run into the $1,000's that's the least cost.
This is the reason that phase change is cheaper, cause you can turn an old used air conditioner into a phase change unit. Phase change cooling has a wattage rating too. For the a temperature held. Exceed the wattage and it will not hold that temperature. Let's not forget that phase change systems also have horse power ratings too. 1Hp is around 750 watts
How ever they are fairly loud for daily use.

DemonicRyzen666Running at TEC at less voltage is more efficient.

It's not because of voltage. It's heat pumping capacity highly depends on delta. The closer hot side is to the cool side, the more heat it can pump. Running it at lower voltages simply makes the hot side less "hot", hence better relative performance, but also higher temps on CPU(which defeats the purpose of having TEC in the first place). You can achieve the same result if you control the current to run the CPU a bit hotter, which also defeats the purpose. I'm sure they have some highly-sophisticated PID control for both cases to keep it within an optimal range all the time, but as you've probably guessed optimal ≠best. As I said before, by the time you find that "sweetspot" it will perform no better than straigh-up LC loop, and the only "benefit" from having TEC in that compicated setup, is bragging about subambient temps on idle and suffering the consequences at high load.

DemonicRyzen666it will transfer more wattage then it uses.

Overunity? :banghead:

EarthDogAnyway, this clearly isn't for anyone here, really. Few at TPU are able to 'think outside the box'... anything extreme cooling is met by a majority of users hating on it.

Extreme cooling with TEC? In order to get some practical results you'll have to mount a big-ass heatspreader plate on the CPU with at least a pair of expensive high-power TEC modules, add some sort of exotic cooling solution that can dissipate ~600W or more from said modules all while keeping delta as low as possible, and only then you can overclock that CPU, give it a decent load and proudly say "TEC did it". With a single module(even the most expensive and powerful on the market) it won't make any difference in the grand scheme of things, especially for 10900k running 5.4GHz at over 1.4V as in Linuses video.

silentbogoOverunity?

Pretty sure they just mean that if the TEC consumes X watts to do its thing, it's capable of cooling a heat output of >X watts. No overunity there.

Still, I've yet to see even a single instance of a TEC actually being a reasonable way of cooling anything PC related. And that definition of "reasonable" includes compressors, LN2, and so on.

Heck, I still remember when my brother had one of those Asetek VapoChill cases. Really cool (both literally and figuratively), but rather impractical when your cooling system mandates a several minute pre-boot cooldown time.

ValantarPretty sure they just mean that if the TEC consumes X watts to do its thing, it's capable of cooling a heat output of >X watts. No overunity there.

That's the definition of overunity. TEC cannot pump more energy than it uses ('cause you know... laws of physics :laugh:).

ValantarHeck, I still remember when my brother had one of those Asetek VapoChill cases.

That's another cool craze of mid-2000s. Back in college I had a friend who copied a DIY phase change design from modding.ru by tearing apart his grandma's old fridge :D :D :D
Can't find a link to this project... it was almost 15 years ago...
Never played with phase change, but it's never too late ))))
Did some DICE back in a day, which killed at least two motherboards, including my precious BFG NF4 Ultra...

Just Some NoiseLets hope they do not......

Well, they're doing it.. if it's really TEC.

Just Some NoiseIf you need more power to cool your CPU, than the CPU needs for itself, then there is something really wrong.

Quantum Processors do that since the beginning. It' how they work actually :D

Just saying..

Just Some NoiseTEC's are only useful for some rare cases, but for the most users, even for Enthusiasts, it is dumb.

I totally agree with you, but some people don't care about power or total heat generated, before 10-15 years, a lot of people used Water cooled TEC to cool the CPU down, these lowered the idle temps below ambient, but have very limited effect on full load, you either needed a much powerful TEC (like 300~500W) or you will loose any benefit on loads as TEC's are not efficient in transferring heat away from the CPU. Not to mention the moisture issues with sub-ambient temps requiring extra insulation to prevent condensation around the socket.

silentbogoThat's the definition of overunity. TEC cannot pump more energy than it uses ('cause you know... laws of physics :laugh:).

Doesn't overunity just mean getting more energy out of a device than what is put into it? Any system with a TEC has two sources of energy after all, their power input and the heat energy of whatever is on the cold side, so for that to be overunity wouldn't you then need to get more heat on the hot side than the sum of those two, not just the power consumed by the TEC? I mean, the heat energy from the CPU has to go somewhere.

silentbogoThat's another cool craze of mid-2000s. Back in college I had a friend who copied a DIY phase change design from modding.ru by tearing apart his grandma's old fridge :D :D :D
Can't find a link to this project... it was almost 15 years ago...
Never played with phase change, but it's never too late ))))
Did some DICE back in a day, which killed at least two motherboards, including my precious BFG NF4 Ultra...

IMO it's frankly a bit weird that ready-made compressor setups like the VapoChill have gone away ... I mean, no, they're not practical whatsoever, but they sure are more practical than a DIY solution, or any other sub-zero cooling setup. I don't think condensation was much of an issue with them either, though I guess YMMV there.

I used TEC when I super-overclocked my Q6600 years ago.

Good times when we had Abit and DFI around.