Comments Locked

48 Comments

Back to Article

  • saratoga4 - Monday, September 26, 2016 - link

    >Note that a large percentage of this energy consumption is inserted as additional thermal load for the cooler to dissipate.

    All of it should show add to the air cooler load. Energy is conserved.
  • ImSpartacus - Monday, September 26, 2016 - link

    This is one of those times where you're reminded why coolers use the tech that they use. It seems to be tough to beat.

    Good review though. Always interesting to see new things.
  • LordOfTheBoired - Tuesday, September 27, 2016 - link

    Honestly, in this case it seems like poor design decisions.
    They went with an unusually small fin stack, and thus unusually limited heat dissipation area, for a cooler design that needed to dissipate more heat than the norm.
  • evilpaul666 - Wednesday, September 28, 2016 - link

    I was surprised to see this review when I popped in today. I'd recently remembered the TECs of the early 2000s and wondered, "Whatever happened to those?" Back in the day they were too expensive and poorly controlled so condensation was a serious risk.

    The latter seems to be addressed by this product, but having just picked up an AIO 480mm water cooling solution for slightly less money than the older ones I looked up that were still listed on Newegg/Amazon this one is still a bit more expensive although it solves the MASSIVE heatsink that won't fit into most cases problem that older TEC solutions had. Possibly to its detriment as the performance doesn't scale great. That might be fixed by different design decisions like you mentioned.
  • Demi9OD - Monday, September 26, 2016 - link

    I ran a TEC with a water cooler way back in the day on my Pentium 3. Condensation was awful, I needed a catch under my CPU for the dripping water.
  • BrokenCrayons - Monday, September 26, 2016 - link

    I had severe condensation problems with a TEC cooler I purchased at a computer show on an AMD K6-3. It didn't ruin the motherboard, but when I pulled the cooler off to check on things after running one for a couple of days the trench around the rim of the heat spreader was filled with water and the underside of the chip was damp. It didn't offer any additional overclocking headroom either, which was what I purchased it for to begin with. It was a 400MHz chip and I never got it to reliably hit 450. After discovering the condensation, I popped my cheap air cooler back on the chip and never looked back.
  • Samus - Monday, September 26, 2016 - link

    Alas, I had condensation issues with my peltier/liquid cooling on my Athlon slot-A which left some options to address over a socketed CPU, but still did something dramatically different to address it.

    I moved the peltier to the water reservoir, installing a heat sink/fan on the peltier. The water reservoir was made of copper so the chilling effect conducted well and this dropped the water temperatures to sell below ambient.

    This system worked awesome for a long time until one day my water pump failed. Yep, the water got so cold it froze. I didn't out antifreeze or anything in, opting for wetter water instead because of its better high temp properties.
  • Marlowe - Wednesday, September 28, 2016 - link

    Samus: That's actually really smart. Although in retrospect you should've applied some antifreeze :-) In what way do you think can a modern AIO water solution incorporate a peltier element?
  • Bigman397 - Friday, September 30, 2016 - link

    A new class of AIO cooler that has an external reservoir/fin stack with a peltier? Seems like an interesting idea. I always figured if I went to the lengths of making a big custom watercooler set I would use a multi-loop system with a heat exchanger, obviously ridiculous in a practical sense but fun to theorize.
  • Stuka87 - Sunday, October 2, 2016 - link

    Another option is to have the peltier on an adjustable voltage slope. When the CPU is idle, it doesn't do much. But as temps increase, it too increases. That way you kind of have the best of both worlds.
  • DanNeely - Monday, September 26, 2016 - link

    Do they have an plans to make a 120/140mm version and a larger TEC? The way this struggles above relatively low loads makes it a bust for the OC crowd who might otherwise be tempted to buy it.
  • zepi - Monday, September 26, 2016 - link

    Indeed, would be interesting to see this being bundled with NH-D15 level cooler.
  • LordConrad - Tuesday, September 27, 2016 - link

    I would also like to see the TEC with a better air cooler attached.
  • saratoga4 - Monday, September 26, 2016 - link

    Or one integrated into a watercooling setup. The main problem is that the TEC COP of performance is something like 1.0 to 1.5 watt of cooling per 1 watt of energy spent on the TEC, so you end up needing 1.5-2.0 watts of radiator for every watt of CPU heat generated. A dual 120 radiator ought to handle that easily.
  • MrSpadge - Tuesday, September 27, 2016 - link

    Chilled water with a phase change (refrigerator) cooler is way more efficient.
  • JesseKramer - Wednesday, September 28, 2016 - link

    Packaging would be more of an issue though I would think.

    There is lots of infrastructure set up around radiators inside cases.

    Adding a TEC into an AIO could be a very interesting product
  • Lolimaster - Monday, September 26, 2016 - link

    Nothing beast using air cooler and undervolting your cpu and maybe drop 100-300Mhz to maximize the gain. Do you really notice the tiny extra performance of OC while wasting tons of watts and heat?
  • damianrobertjones - Monday, September 26, 2016 - link

    Yes! I, as a child, often heard scary stories of the 'Nothing Beast'!
  • AndrewJacksonZA - Monday, September 26, 2016 - link

    *chuckle*
  • Death666Angel - Tuesday, September 27, 2016 - link

    I do. Running my 4770k @ ~3.4GHz like you suggest vs. the 4.5GHz it currently runs at would not save much energy on normal tasks (I OC with an adaptive offset, so 800MHz use ~ 0.71V). Even in many modern games (The Witcher 3, Battlefield 4), I would get lower FPS with such a CPU clock and my R9 290X, which is more CPU dependant than typical Nvidia graphics cards. And doing video encoding of my Blu Ray rips would be painfully slow, scaling nearly linearly with clockspeed. The ~50 to 90W I spend more on intensive tasks is worth it for me, since I can saturate 110FPS with my monitor @ 110Hz and save time during encoding (which make total power consumption per rip not as bad as it would seem).
  • Communism - Monday, September 26, 2016 - link

    The best way to do long term below ambient CPU or GPU cooling is to attach a mini-split (google it if you don't know what the term means) to a water cooling loop attached with a copper to copper [imagine something similar to interface between the IHS and the contact plate of a cooler] connection (To interface the mini-split to the water cooling loop without any fuss).

    This will simultaneously cool your CPU/GPU below ambient for extended periods of time while piping the heat directly outside your home.
  • hybrid2d4x4 - Monday, September 26, 2016 - link

    This is more of a comment on your testing platform than this review, but it seems to me that your current setup has a few shortcomings.
    1) The testing platform isn't close enough to simulating the sockets of mobos to give any meaningful comment on how good the mounting mechanism is on the cooler. This is especially true of AMD socket.
    2) The operating range for your thermal load isn't well suited to represent CPUs. IMO, it would be infinitely more useful to have something that ranges from 5-10W (~the idle state of a modern CPU if not less) to 100W (or whatever actual power draw an enthusiast OC setup would be). While interesting in an academic sense, testing up to 340W is completely irrelevant and beyond the design targets that these coolers were aiming for.
  • BurntMyBacon - Monday, September 26, 2016 - link

    AMD FX-9000 series processors are rated for 220W. Processors consume more power at an exponential rate as more voltage is applied. Perhaps 340W is a bit much, but not as far off as you seem to think.

    That said, I would like to put in another vote for representing more lower power gradients. Common TDPs are something like 5W, 15W, 25W, 45W, 65W, 95W, 125W, and 140W. Higher TDPs like 180W and 220W are also present, though less common. You don't need to hit all of these TDPs, but just 60W probably isn't a good representation for everything under 100W.
  • DanNeely - Monday, September 26, 2016 - link

    The crazy high TDPs are also reachable with aggressive overclocking and high end cooling.
  • Sushisamurai - Monday, September 26, 2016 - link

    I too second this opinion. I would like to see more temp gradients around the 60, 80, 100, 120, 140W of usage... A graph for these things would be nice if time was permitted, as when the author mentions thermal resistance/performance falling off, he's not very specific at what point does it "fall off" (eg: is it competitive at 60-100 but falls off at >120?). I imagine a lot of people use this site's data and opinions to shop for products, and having more gradients to align with purchase decisions would be nice. I know I have a 80W stock load, 120W Oc, 140W stock and 220-240W loads. Extrapolating your data is doable, but I think not everyone can. My re-verification numbers on my home hardware #'s are similar to the extrapolations and your test results, so I thank you.
  • eldakka - Tuesday, September 27, 2016 - link

    I'd also like to see lower TDP measurements. I'm not thinking of overclocking situations, but more HTPC-type situations, where with a good, quiet cooler you might be able to put a more powerful processor in the HTPC while still remaining quiet.

    This thing being chromed might look pretty cool in a HTPC sitting under a telly...a slimline case with a hole cut in the top for so the fins stick out the top, like a blower sticking out of the bonnet (hood) of a car ;)
  • hybrid2d4x4 - Tuesday, September 27, 2016 - link

    Wow, thanks for that eye opener! I assumed the most power-hungry CPU you can buy today that's not AMD from a few gens ago was 95W. 220W! What a beast...
  • Vayra - Monday, September 26, 2016 - link

    Well, I for one am completely uninterested in idle temps, because given a large enough heatsink you can likely even passively cool that.

    It is much more interesting and informative to have 'over the top' TDPs rather than a slew of low TDPs because that is when cooling starts to struggle. For any non-OC'd CPU, you can suffice with the regular boxed cooler and it will keep it safe from throttling 99% of the time, or at least close enough to not matter at all.

    Typical mid range consumer TDP is 65w up to 95w for a quad core. So the 60w and 100w very clearly represent the majority of use cases and non-OC situations. Lower TDP is irrelevant - all coolers will perform as good, or better, at lower TDP - this enters the region of 'who cares' because there are literally zero benefits to running very low temp at low. Temperatures matter when they pass the 50-60 C barrier because they then *might* start influencing current leakage. Below that temperature, it's basically a non issue on all counts.
  • DanNeely - Tuesday, September 27, 2016 - link

    While its true that all large coolers will do well at <60W, low wattage desktop CPUs (eg Intel's 35W series) are often used in slimline mITX cases where even something like Intel's stock cooler is too large to fit. Temperature/noise tradeoffs there become an important consideration again; as do idle core temperatures. That would be a different set of testing ranges than the one that E. Fylladitakis currently runs; and more inline with what Silent PC Review authors test.
  • Vayra - Monday, September 26, 2016 - link

    Well, I for one am completely uninterested in idle temps, because given a large enough heatsink you can likely even passively cool that.

    It is much more interesting and informative to have 'over the top' TDPs rather than a slew of low TDPs because that is when cooling starts to struggle. For any non-OC'd CPU, you can suffice with the regular boxed cooler and it will keep it safe from throttling 99% of the time, or at least close enough to not matter at all.

    Typical mid range consumer TDP is 65w up to 95w for a quad core. So the 60w and 100w very clearly represent the majority of use cases and non-OC situations. Lower TDP is irrelevant - all coolers will perform as good, or better, at lower TDP - this enters the region of 'who cares' because there are literally zero benefits to running very low temp at low. Temperatures matter when they pass the 50-60 C barrier because they then *might* start influencing current leakage. Below that temperature, it's basically a non issue on all counts.
  • MrSpadge - Tuesday, September 27, 2016 - link

    Cooling at low power consumption levels (up to ~30 W) is very easy with modern coolers, unless you are restricting the hot air exchange with the environment (those HTPCs which for some reason have to be super small). If you're interested in such cases it would make more sense to test the case & envirnonment rather than the cooler itself.

    60 - 100 - 150 W covers typical CPUs under load very well, with 200 - 250 W representing overclocked socket 2011 chips and 340 W the extreme.
  • BurntMyBacon - Monday, September 26, 2016 - link

    It would be interesting to see the crossover point between half on and full on. The aircooler isn't particularly large and doesn't seem to dissipate heat fast enough to keep up with the TEC. While providing a larger delta between the hot and cold plates, adding an extra 20W for the air cooler to deal with can, in some circumstances, actually increase the steady state temperature.

    A good test (that I'm not sure this kit would let you run) would be if you could test several TDP points at both half (20W) and full (40W) TEC power. I imagine you would see full power come in cooler up to the point that the air cooler starts failing to keep up. Switching to half TEC power would then be more efficient as it would dip back under that threshold. Once you increase thermal output to reach that threshold again, it would be interesting to see which mode works better with an overburdened air cooler. In the past, I've was told to use TECs with water cooling systems as air coolers at the time always got overburdened too early.
  • Avenger762 - Monday, September 26, 2016 - link

    Once again, this thing is no match for my old CM V10. BTW the V10 in my tower has been running since 2006 with the exception of a couple of motherboard and fan replacements.
  • VeauX - Monday, September 26, 2016 - link

    it would be nice to see it compared to :
    1 - same size air cooler (comparable cu.cm or length / height / width)
    2 - same price coolers including AIO Liquid Coolers
  • HomeworldFound - Monday, September 26, 2016 - link

    I used to own a Thermaltake Subzero 4g Thermoelectric Cooler, used on a Pentium 4. This does seem like an evolved version of that. The Thermaltake used up a PCI slot and was oversized by 3/4 of an inch. There wasn't any condensation to deal with but I can't really say that it worked well.
  • andychow - Monday, September 26, 2016 - link

    Ask any physicist, and they will explain to you why thermoelectric cooling is a terrible idea. It's only if you need a specific solution, such as bellow ambient. Otherwise, it's super inefficient.
  • powerarmour - Monday, September 26, 2016 - link

    Yo dawg, I heard you need a cooler to cool your cooler?
  • RaistlinZ - Monday, September 26, 2016 - link

    I was interested in this until I saw the performance.

    Then I saw the price, and became even less interested.
  • benzosaurus - Monday, September 26, 2016 - link

    In which we learn that "refrigeration" and "heat dissipation" are distinct and incompatible concepts.
  • jlk440 - Monday, September 26, 2016 - link

    Thanks for an interesting review. I'm puzzled by your reversal of the meaning of thermal resistance. Lower number = lower resistance = fewer degrees difference per watt of heat. For heatsinks, you want thermal resistance to be LOW, which means heat conductivity is HIGH. For some reason your terminology incorrectly calls lower resistance numbers "higher resistance", so maybe you mean higher conductivity (the opposite of resistance).
  • MrSpadge - Tuesday, September 27, 2016 - link

    "As the load increases, the thermal resistance of the HEX 2.0 decreases, particularly with thermal loads above 200W. With a thermal load of 340W, the thermal resistance of the HEX 2.0 drops down to 0.1685 °C/W"

    Yep.. this should be:

    "As the load increases, the thermal resistance of the HEX 2.0 INCREASES, particularly with thermal loads above 200W. With a thermal load of 340W, the thermal resistance of the HEX 2.0 INCREASES UP to 0.1685 °C/W"
  • bug77 - Tuesday, September 27, 2016 - link

    Oh goodie. A cooler with Windows-only software. Where do I sign up? /s
  • BulkSlash - Tuesday, September 27, 2016 - link

    It's great to see someone experimenting with TECs again, I think the last time I used one was with the ATI Radeon 800 XT! It's a pity this doesn't really work for really high temperatures as that's really where a TEC can come into its own.
  • Haravikk - Wednesday, September 28, 2016 - link

    Seems like some odd choices. For me it always seemed like a TEC made most sense as an external unit for liquid cooling, since the liquid can transfer the heat out and the TEC then cools it somewhere that the condensation problem can be controlled. Plus externally you can have a massive heatsink or even a huge plate for dissipating passively.

    Trying to squeeze it inside a case like this just doesn't seem all that practical, and it doesn't really seem to offer much in the way of advantages; I'm not that fussy about how cool my processor is when it's idle so long as my fans are quiet at that point. Most good coolers will keep a CPU from exceeding 40-50ºC under light load without making too much noise (some will even do it passively in the right case) so I just don't see what this unit offers for the price and complexity.

    Also, while aesthetically I like the fan in the middle, practically I don't see the point; horizontal space usually isn't so tight that you can't fit the fan on the front or back of the cooler, plus with that design it's easy to provide mounts for a second fan. Sandwiching in the middle just seems like it limits the heatsink unnecessarily.
  • dave_the_nerd - Wednesday, September 28, 2016 - link

    On one hand, this is a compact, nice looking cooler that performs very well with the 50-65w TDP CPUs that Intel makes and sells.

    OTOH, I really want a Reeven Okeanos now.
  • zodiacfml - Saturday, October 1, 2016 - link

    Other reviews were not able to find that advantage at low loads.
    The question now is, which chips are suitable for overclocking to 65W. If it did, will it improve highest possible overlclock?
  • Sunburn74 - Tuesday, October 4, 2016 - link

    Was I the only one who thought the conclusion was a bit off? Why didn't the reviewer just say it's a bad product at a terrible price point? I mean, it appears to be that there is always a smaller, quieter, cheaper and equally if more efficient cooling option in almost every situation tested
  • Sadler2010 - Monday, August 22, 2022 - link

    I used to run an Ultra Chilltec Black until the Peltier died it successfully cooled a Phenom II 970T and my FX-8350 until it flatlined. So my experience with TEC's was good, but now I'm looking at AIO's to avoid the issues brought by TEC's.

Log in

Don't have an account? Sign up now