News Noctua's next-gen flagship CPU cooler finally arrives — Noctua NH-D15 G2 released at $150

[TheHerald]

The biggest problem with aios is, it completely destroys a cases airflow. That might not be a problem with your average gpu, but back when I had my aio with a 3090,oh boy. During gaming using the aio as exhaust, the cpu was literally cooking itself. Using it as intake the gpu was cooking itself. Switched to a U12A and holy cow, the drop in temperatures was massive.

 

[bit_user]

back when I had my aio with a 3090,oh boy. During gaming using the aio as exhaust, the cpu was literally cooking itself. Using it as intake the gpu was cooking itself.

IMO, the proper solution to this problem is to have a separate compartment for the radiator, with its own intake. You could also have an externally-mounted radiator, but I gather people don't like those.

 

[bit_user]

Peerless Assassin isn't a great comparison due to having a lot less thermal mass and there being better coolers from Thermalright on the market. There should be no question that if value enters into your purchasing decision Thermalright has cornered the market.

(nobody but GN has G2 reviews that I've seen)

FC is good, but there's certainly better options now:

Why do this and the TechPowerUp graph @Zerk2012 posted using the regular NH-D15? Isn't the newer NH-D15S supposed to be better?

 

[thestryker]

IMO, the proper solution to this problem is to have a separate compartment for the radiator, with its own intake. You could also have an externally-mounted radiator, but I gather people don't like those.

It's not actually a real problem though. If an AIO being used as exhaust has worse temperatures than an air cooler you've failed miserably at something along the way. The video card heat is the same either way and the majority of its heat goes through the CPU cooler for both. So if a CPU was "cooking itself" then it would happen with either type of cooler.

 

[thestryker]

Why do this and the TechPowerUp graph @Zerk2012 posted using the regular NH-D15? Isn't the newer NH-D15S supposed to be better?

Isn't the S just a single fan design?

Pretty sure it's rated lower, but it might be better with 2 fans? I'm not familiar enough with the revision.

 

[CmdrShepard]

Extra heatpipe for user is irrelevant. All that matters is the end result for somebody buying today or tomorrow. Thermalright offered a superior product for most use cases that was also cheaper, and a few other companies are offering similarly efficient products lately.

Yes, but they did it 10 years later hence my sarcasm.

If price is irrelevant (min-maxed rig + have leftover $), and you have brand loyalty/trust for Noctua only, and your needs requires a product that Noctua is selling, well, go ahead, your money and life. Do what makes you happy.

Price is never irrelevant -- I do consider it when buying like everyone else. It's just that I am not infected with "it must be cheap" virus and I don't have the need to brag about this or that "great deal" I snagged on the latest sale. I get that some people genuinely have to save every cent because they have it hard. However, looking to buy something as cheap as possible and then brag about it is just another form of attention seeking which I detest and that's what I had in mind in my previous comment.

For my current use case (Xeon W, 4677 socket), Noctua was the only air cooler available (if you don't count horrible sounding workstation / server air coolers) when I was building the rig and I think that's still the case. In the meantime, an AIO has showed up but availability is still poor. I am still debating whether it is worth upgrading to that.

 

[bit_user]

It's not actually a real problem though.

No, CPUs being thermally-limited clearly is a real problem.

The video card heat is the same either way and the majority of its heat goes through the CPU cooler for both. So if a CPU was "cooking itself" then it would happen with either type of cooler.

It's not necessarily the same. For a CPU air cooler, it really depends on the airflow of the case, which I think was @TheHerald 's point. Conversely, if you're exhausting most of the air through a radiator, then that would be including most of the heat from the GPU.

 

[bit_user]

Isn't the S just a single fan design?

Yes, I think it only comes with one fan, but you can still run 2. They also made some other changes to improve compatibility & performance, but it seems that even running it with 2 fans only lowered CPU temps by a degree or so.

 

[thestryker]

No, CPUs being thermally-limited clearly is a real problem.


It's not necessarily the same. For a CPU air cooler, it really depends on the airflow of the case, which I think was @TheHerald 's point. Conversely, if you're exhausting most of the air through a radiator, then that would be including most of the heat from the GPU.

Please explain how you think the airflow of a case can benefit a tower cooler over an AIO.

 

[35below0]

Yes, I think it only comes with one fan, but you can still run 2. They also made some other changes to improve compatibility & performance, but it seems that even running it with 2 fans only lowered CPU temps by a degree or so.

The D15 comes with an optional fan. the D15S does not. That's the only difference.

Using the optional fan on a D15 is... optional. I've tried with and without and i don't want to say there's no difference, but it probably comes down to the individual components and use cases.


As for the wider topic, it seems the price is the main headline. The cooler performs well. No amount of slightly beating another cooler is going to make up the difference in price.
But it's a good quality cooler and anyone not on a budget and not wishing to cheap out on a cooler will consider it. Noctua probably know they will not sell the most but they hope to win awards and maintain a premium position.

Could they lower the price? Probably yes, but it will not increase their revenue enough and they're not bothered about market share.
Same people will buy them anyway. Or avoid buying them.

 

[bit_user]

Please explain how you think the airflow of a case can benefit a tower cooler over an AIO.

A tower cooler, in a tower case, is sitting in line with airflow from the intake fans. Some portion of its intake is going to be directly from that stream.

In contrast, a top-mounted radiator (I assume we're talking about like a 360 mm) is going to have 3 fans. That's probably more than doubling your number of exhaust fans and they'll be drawing air from wherever they can get it, including the GPU-heated air. So, if we hold the number of intake and dedicated exhaust fans constant (or replace some of the exhaust fans with the radiator), you just end up with a more heavily negative-pressure setup that will tend to pull air up from the GPU more.

It's very hard to argue hypotheticals, without going into specifics of different airflow setups, but I think @TheHerald has a point - and one backed up by personal experience. I'm not saying you can't do things to try and improve either configuration, but I think simply switching between a tower cooler and a top-mounted radiator, without changing anything else, will tend to work out as described.

 

[Phaaze88]

My point was that FC140 doesn't beat the D15 on Intel so there's no way it would beat the G2 there. As for AMD it might still be ahead there, but based on GN's results comparing D15 to G2 I doubt it.

Understood, but what I was getting at was for the other person to prove their statement, or they were spewing air. Not just on value(everyone can see that), but the other factors that aren't as obvious. This is one of those products whose focus was the engineering that went into it. Not everyone cares about that, nor is value treated the same way, and either's fine.

They haven't shown anything, so... I'm going to move on.

Isn't the newer NH-D15S supposed to be better?

D15S is a hidden gem. Performance barely behind the original, superior case and ram compatibility.
If you were specifically shopping for a Noctua cooler, and if the S model could fit in the case, it was their best darn one overall. But it kept getting skipped over, because:
-Single fan... on a dual tower???
-"D15 is the better performer anyway, just get that one, bro."
The point behind the S was the compatibility anyway. The narrow performance gap may as well be icing.

The biggest problem with aios is, it completely destroys a cases airflow.

Chimney style cases. They're rather niche, unfortunately - mostly Silverstone these days, but the O11 Dynamics can be set up to do similar.
Too bad gpu AIOs or custom loops aren't more common too, 'cause then the largest heater could be positioned to exhaust right out the case.
Blowers aren't popular for other reasons.

 

[thestryker]

A tower cooler, in a tower case, is sitting in line with airflow from the intake fans. Some portion of its intake is going to be directly from that stream.

In contrast, a top-mounted radiator (I assume we're talking about like a 360 mm) is going to have 3 fans. That's probably more than doubling your number of exhaust fans and they'll be drawing air from wherever they can get it, including the GPU-heated air. So, if we hold the number of intake and dedicated exhaust fans constant (or replace some of the exhaust fans with the radiator), you just end up with a more heavily negative-pressure setup that will tend to pull air up from the GPU more.

Where do you think most of the video card heated air goes when using a tower cooler?

Video card cooler designs blow air out in every direction (or have flow through, but those dump hot air right into the CPU socket area) so the majority of the hot air will follow the flow of air in the case. That means most of the heat goes through the CPU socket area as that's where most of the airflow will be with a tower cooler.

With a proper airflow setup the disadvantage here is equal between an AIO and tower cooler. When using an AIO for exhaust you still want to try to have an overall front to back airflow path in the case rather than front to top (just talking standard case design front/top/back). That means setting up a fan curve that matches your workload or setting a low fixed RPM. The whole point of using liquid cooling (with reference to cooling advantage not aesthetic) is that you don't need to brute force anything you just need enough airflow that hot air doesn't stagnate.

 

[bit_user]

Where do you think most of the video card heated air goes when using a tower cooler?

I said why I thought @TheHerald 's observation made sense to me, and it's also backed up by a n=1 sample. I'm not going to get into a long, drawn out debate over the topic, because it's not the easiest thing to discuss, there are lots of variables, and there's not much in the way of high-quality data that we can draw from. So, I acknowledge the difference of opinions, but respectfully withdraw myself from this debate.

BTW, if you are using a tower CPU cooler and a high-power GPU, then I think you should probably make sure to have ample rear exhaust fans, specifically to draw off heat from the GPU so it doesn't tend to enter the intake stream of the CPU cooler.

 

[umeng2002_2]

I agree with the D-15S. It give you more clearance from the GPU and if you add another fan, it performs just like the non-S version.

 

[Zerk2012]

Why do this and the TechPowerUp graph @Zerk2012 posted using the regular NH-D15? Isn't the newer NH-D15S supposed to be better?

No the D15S is not better about a 2C difference, if you add the extra fan then it's the same. So why would they use it in the graph.

I think NG review sums everything up on his review of the G2 you can just start at the 17 min mark and watch about 30 seconds of it. If you go further and look at the 250 watt the G2 beats the only one in the test by about 4C

View: https://www.youtube.com/watch?v=heriTDWIU2g

 

[Conor Stewart]

$150 for a lifetime heatsink that will be supported on many sockets. Buy new fans after 10 years or something, IDK - they freaking last.
Buy a more affordable or expensive(your call) AIO, the lifetime of which both pump and fans is finite. They are leak resistant, not leak-proof.

We don't share the same value of the buck either, which affects purchasing decisions, and probably never will. Things like, "how much does cost matter if you can set it and forget it?", or "so what if something needs replacing every so often, if you have a steady income?"

I could see myself buying one for my overdue platform upgrade, but not without seeing some reviews first.


Support, R&D costs?
They gotta be giving their attention - perhaps too much - to certain details that other brands are skipping, or not focusing as much on. Things we might not be seeing as a big deal.
Like, why have we yet to see any white fans or heatsinks from them, but black has been fine? Is there more to black and white paints besides colors?
There's sound profiles too:

View: https://www.youtube.com/watch?v=F7ia_FZcthQ

Yes the colour does matter, it isn't just paint it is pigment mixed with the plastic. Different colours require different pigments and adding pigments is essentially adding impurities into the plastic. Carbon black is common for black pigment and titanium dioxide is common for white, these two pigments have very different properties and the amount of them that needs added will be different too.

Looking at 3D printing, white filament is generally more of a pain to work with than black and there can even be slightly different mechanical properties based on colour. White being a pain is due to the pigment itself and how much you need to add. There is a reason that recycled filament is usually turned black rather than other colours or white. In general it is much easier to turn things black than white.

Then looking at high end engineering materials, they come in natural or carbon fibre filled, not really any other colours because if you need to use those materials you want the best results possible to justify the cost and colour is an afterthought. Interestingly most really high end plastics are a sort of honey colour, I don't know what colour LCP would naturally be but it could just be that their standard colours are close to it's natural colour and adding pigment to turn it black is much easier than turning it white.

 

[Conor Stewart]

Which size are you talking about? I've been pretty satisfied with their non-PWM 120 mm and 140 mm fans. Also 150 mm PWM.

Recently, I've dabbled with some of their 92 mm PWM fans and I was disappointed with the high-RPM noise profile, as you say. However, I'm not actually sure if my fan controller might be at least partially to blame. Fan controllers sometimes support both voltage modulation and PWM. The BIOS on that board doesn't provide an option, so I'm not sure which it's using. However, it came with a 3-wire CPU fan, that clearly was ramping up & down its RPMs, so that's not a good sign that it's using PWM.

Plus, is the PWM carrier frequency even standardized? Could you have two different fan controllers that use different frequencies, thus exhibiting different noise profiles?


I think everyone is referring to this:

​

https://www.tomshardware.com/pc-components/air-cooling/thermalright-phantom-spirit-120-evo-review​

​

If you have a multimeter or ideally oscilloscope then you can check what it is using.

Also the PWM does not directly drive the motor. Most fans use brush less motors and have built in controllers, the PWM is just a way of telling the controller what speed it should spin at. Switching to other digital protocols is generally better. For example drone Electronic Speed Controllers (ESCs) used to use PWM but have switched to basically a digital serial protocol which gets better results.

 

[CmdrShepard]

@bit_user In my case the Xeon socket is rotated 90 degrees so that the tower cooler fans are facing up and the air is pushed vertically (i.e. upwards), not horizontally which makes much more sense to me because it relies on natural rising of hot air.

Case arrived with 3 fans and I put one at the bottom as cold air intake, and two at the top as the exhaust above the CPU fan. It works just fine for me, but if I were to switch to AIO for the CPU I'd have to reorient case fans again.

 

[Conor Stewart]

Well, one review is out:

View: https://www.youtube.com/watch?v=heriTDWIU2g


It's those fans. That's what most of the money you pay towards this went into.
Improved performance to noise profile, and the fans will age better. I know they wear and tear over time, but it never crossed my mind performance went down too.

Want to see a few more reviews, but so far, it's still got my interest - at least the standard model. I'll pass on the HBC/LBC.


EDIT: I think I found a reason in the video comments about why the lack of color options, or why it takes so long.
"Yes, because it's not just as easy as coloring plastic differently. Different dyes change the plastics properties. And when you want to release a product that's optimized for performance, changing the color would not give the same results as the brown/beige versions."
So paint isn't just paint after all, and Noctua has much stricter standards than other vendors.

Paint isn't paint at all in a lot of cases. They don't just paint on top of the part they mix pigment in with the plastic itself. Cut a coloured plastic piece in half (most coloured plastic objects) and you will see the colour all the way through.

Adding pigment into the plastic is essentially adding impurities. Pigments can be very different too, carbon black is often used for black, titanium dioxide is often used for white, very different chemicals. The amount of pigment needed depends heavily on colour too, it is generally a lot easier to make things dark colours than light with white being the worst, especially if you want pure white and not some off white.

If I had to guess then noctuas standard colours may not be far off from the plastics natural colour and hence doesn't need much pigment. Making them black likely doesn't need much pigment either but making it white probably takes loads of pigment or paint on top, both of which would likely lower the mechanical properties or the plastic or fan performance.

 

[bit_user]

If you have a multimeter or ideally oscilloscope then you can check what it is using.

It's not that simple. I'd have to make a custom cable or ruin a fan extension cable by cutting the jacket and stripping the wires, but yes. TBH, I don't really care that much, because whatever the motherboard's fan controller is doing, I can't really change it. So, the only benefit would be to satisfy my curiosity.

the PWM is just a way of telling the controller what speed it should spin at.

Uh, maybe. From what I've read, PWM fans used to be 3 wire, with the PWM signal on the wire providing power. The 3rd wire was for the tachometer. Then, the design was improved to give the tachometer continuous power, hence the 4th +12 VDC wire was added. I suppose the motor controller could now do as you say and use the continuous 12V for power.

Switching to other digital protocols is generally better.

Yeah, if the motor controller is indeed just reading the PWM signal and not actually using it for drive current, then they might as well use a mechanism like I2C for sending control signals. I suppose you could also then multiplex tachometer over the same wire.

 

[bit_user]

@bit_user In my case the Xeon socket is rotated 90 degrees so that the tower cooler fans are facing up and the air is pushed vertically (i.e. upwards), not horizontally which makes much more sense to me because it relies on natural rising of hot air.

Case arrived with 3 fans and I put one at the bottom as cold air intake, and two at the top as the exhaust above the CPU fan. It works just fine for me, but if I were to switch to AIO for the CPU I'd have to reorient case fans again.

I'm just curious how well that setup is working for anything that simultaneously stresses the CPU and GPU, such as gaming. Have you run anything on it, like that?

 

[Conor Stewart]

D15S is a hidden gem. Performance barely behind the original, superior case and ram compatibility.
If you were specifically shopping for a Noctua cooler, and if the S model could fit in the case, it was their best darn one overall. But it kept getting skipped over, because:
-Single fan... on a dual tower???
-"D15 is the better performer anyway, just get that one, bro."
The point behind the S was the compatibility anyway. The narrow performance gap may as well be icing.

Looking at Noctua's website the original NH-D15 was symmetric, the NH-D15S is asymmetric to allow more GPU clearance and is missing the front fan for RAM clearance. Looking at the NH-D15 G2, it is asymmetric just like the NH-D15S, so it should share the same compatibility benefits and has two fans, the front one can be raised up for RAM clearance but it makes the cooler taller or it can be removed completely.

So the NH-D15 G2 is an improved version of the original NH-D15 whilst having the extra compatibility options of the NH-D15S and can be used as a single fan cooler if required.

 

[Conor Stewart]

It's not that simple. I'd have to make a custom cable or ruin a fan extension cable by cutting the jacket and stripping the wires, but yes. TBH, I don't really care that much, because whatever the motherboard's fan controller is doing, I can't really change it. So, the only benefit would be to satisfy my curiosity.


Uh, maybe. From what I've read, PWM fans used to be 3 wire, with the PWM signal on the wire providing power. The 3rd wire was for the tachometer. Then, the design was improved to give the tachometer continuous power, hence the 4th +12 VDC wire was added. I suppose the motor controller could now do as you say and use the continuous 12V for power.


Yeah, if the motor controller is indeed just reading the PWM signal and not actually using it for drive current, then they might as well use a mechanism like I2C for sending control signals. I suppose you could also then multiplex tachometer over the same wire.

Or you just need steady hands or some jumper wires but I do understand not wanting to poke around a live motherboard with metal pins.

4 pin fans now do use the 12 V for power, the PWM is just for speed, you can tell because if you plug in just ground and 12 V the fan spins at max speed. The PWM pin tells the controller which speed to work at and if there is no PWM signal they generally default to max speed. Another way you can tell is to run a fan whilst using an Arduino to control the PWM pin, Arduino pins are rated at maximum 20 mA, nowhere near enough to run a PC fan, all the power for driving the fan comes from the 12 V pin. If you have a controller in the fan anyway (required for brush less fans) then you can get rid of a lot of complexity on the motherboard side. If the motor is driven directly or gets power from the PWM pin then the pin needs to be able to deliver high power meaning the motherboard needs drive circuitry, like a MOSFET, if however the PWM pin is just signalling then it can be driven straight from low power electronics which will reduce complexity and heat generation on the motherboard.

This is also why noctua include the Low Noise Adapter (LNA) for when you can't control the fan speed through PWM, it just lowers the fan supply voltage making it run slower, it doesn't do anything to the PWM line.

As for your last paragraph, exactly, it could also allow driving multiple fans from a single header but at different speeds, like how addressable RGB LEDs do. I wouldn't be surprised if this is similar to what LianLi is doing with their highest tech fans. In the uni fan TL, you can connect 10 fans in a row or 16 fans in total and seems to say you can individually control the fan speed but you are required to use their controller, the fans don't have a normal 4 pin header. So they will be using some kind of digital protocol and have more complex controllers in the fans, especially since you can control the LCD fans through the same connection.

LianLi seems to be the ones really adding new features to fans, like the interlocking daisy chaining and now moving to a better protocol. What we need though is a standard that motherboard manufacturers can use too, so you can use any brand of fan without separate controllers. This shouldn't be too difficult either, the existing 4 pin fan layout should be usable still since the PWM pin is just a digital pin, just drive it like a PWM signal for PWM fans but then use it for a different protocol for other fans, like how you can change between DC and PWM on some motherboards.

 

[TheHerald]

Please explain how you think the airflow of a case can benefit a tower cooler over an AIO.

Really now? You call other people failure and you don't understand what a freaking radiator does? A radiator placed into your exhaust reduces your airflow by as much as 1/3. Go check some fan reviews and see how much less air they push when put in front of a radiator.