Noctua Introduces NH-L9i Low-profile CPU Coolers for LGA1700 and NA-FD1 Fan Duct

[Dredi]

But you wanted to subtract 78 Watts from that.

 

[R0H1T]

those speeds/power consumption is where the chip(s) thermally throttled.

I'm pretty sure you can't measure that to such a perfect integer value, you'd have to poll data at much less that the regular 2sec intervals for hwinfo (or do actual hardware measurements) & even then do multiple runs to come to this value without altering anything in the BIOS. Also that number (160W) will vary each run if you're looking at peak cooling capacity of this cooler.

Speaking of ~ this reminds me of the way I got my post count so high! I just wish I'd chosen law now

Not to mention Valantar's probably already one

 

[Metroid]

But you wanted to subtract 78 Watts from that.

If we subtract 319 - 78 then it is 241 watts, so there is no 300 watts plus the cpu alone if system idle is indeed 78 watts.

 

[Dredi]

If we subtract 319 - 78 then it is 241 watts, so there is no 300 watts plus the cpu alone if system idle is indeed 78 watts.

Tom’s hardware managed to get their ddr5 12600k to a max. of 135 Watts. - package power.

Overcloked it can probably go to 500W with liquid helium. So this cooler is a dud.

 

[Valantar]

It was a Fractal Design Node 202 case so airflow was good since the cooler was right underneath the perforated part of the side panel.

The CPU reached 80 degrees or something like that. I don't remember the exact tests I performed but I do remember thinking to myself "Wow, so this is the limit of such a low profile cooler. Good thing I didn't go for a higher TDP CPU".

Hm. You were using a 39mm cooler in an up-to-52mm case. Sounds like you would have benefited from their foam ducts launched alongside this revision! Seriously though, you'd be surprised just how little air a fan pulls in through a grate that's spaced out from it by just a tad. Ducting is almost a necessity in cases like that - you need very unrestrictive panels for a fan to work as an intake fan if it's spaced more than 5mm from the vents. And the Node 202, especially with the included filter, is quite restrictive.

So your theory is if fans are 100% always and the radiator keeps heating up then it will increase performance in heat transfer but what about temperature?

No "theory" (at least in the common usage of the word, though I guess in the scientific one, yes), but fact. At some point the loop will reach steady state, assuming no other part of the thermal loop breaks down, i.e. the liquid temperature will stabilize at a point where thermal output at the radiator is equal to the input at the heat source. That is what liquid coolers generally do.

I have done a similar test here few days ago, steady 100% fans and then temperature going up and up and then it hit the thermal limit, yes probably you right on that but the temperature kept increasing and boom, system turned off.

Hey, a concrete statement of what "insufficient" means! Thank you. Only took, what, five posts? That does indeed sound like it's insufficient - though are you sure your cooler is mounted properly? What case are you using, and is it restricting the airflow for the radiator too much? Might it be recycling hot air through some sub-optimal airflow layout?

What that means in relation to the "theory" above: the steady-state point of that loop is too high to keep the CPU temperature within its operating parameters. I entirely agree that this is insufficient (and it really shouldn't be for a 240mm rad on a 5900X - something is off here). But please understand that for people to understand what you're saying, you need to provide enough information that it can actually be understood.

So my statement holds true, if fans are 100% and temperature keeps rising them gameover, maybe some people think a temperature will be hit and will not rise anymore than that, I thought that too and I was wrong. It kept rising and rising, steady, clearly it needed better fans or better heat transfer to keep from rising.

Is your chip OC'd in some way? It should thermal throttle to keep itself in check, and thermal shutdown should essentially not happen. Heck, I've run a Threadripper 1920X with a clogged liquid cooler (one of those notorious Enermax coolers) and it never shut down. It sat steady at tJmax and ~600MHz all core, but it never shut down.

That's stretching it, it's at best a CM 212x level cooler. Best case scenario ~ it'll come close to a CM 412 cooler. Of course a lot will depend on ambient temps, near the tropics like in Asia or Africa or even down under (summers) this is no good for a "workstation" period!

You seem to be misunderstanding what I'm saying. This cooler can fit in a case like this, for example. That is the size of a large-ish book, at 4l in volume. Heck, a 212 would fill most of that case. And of course you can't run a 12900K or 5950X at anywhere near full performance in there, but with a cooler like this (i.e. a really small but very good for its size cooler) you can tune a chip like that for massive multi-core performance even in a size that you can fit in literally any backpack. Will it outperform, say, a stock 5800X or 12600K? Depends on the application. You clearly don't build something like this unless you need it. So yes, it can make for an extremely powerful, tiny workstation. And if you have a bit more space, something like the L12s will let you run everything a tad faster, and still have it portable.

 

[TheinsanegamerN]

lol, how can this thing can possible cool a 300 watts cpu. This can only be a joke.

I read and I dont believe that thing can cool a 300 watts cpu. The can say anything they want about testing and the bs, I still dont believe it. Not even my 240mm aio can cool a ryzen 150 watts well.

I think the REAL joke here is your cooling setup. How on earth can you not manage to cool a 150w ryzen CPU with a 240m AIO? I was cooling an OCed 2700x with a hyper 212 air cooler.

 

[looniam]

shots fired!

 

[FreedomEclipse]

Im curious about if we were to have a shroud that wrapped around where the fan sits on the heatsink, covering some of the exposed fins if it would improve cooling at all due to the increased static pressure??

 

[Metroid]

I think the REAL joke here is your cooling setup. How on earth can you not manage to cool a 150w ryzen CPU with a 240m AIO? I was cooling an OCed 2700x with a hyper 212 air cooler.

50 degrees outside and 35c inside is just one of the reasons eheh, when is minus 10c is easy to even using this low profile cooler on the 12900k ehhe, my point there are many factors as to why, so without having all the data is hard to know about things, for example why noctua used a 12900k on this low profile cooler hehe, when a said product is used with another product that means it's likely possible to use together which here was just a brag marketing bs. Anyway with delta fans and favorable ambient temperature, this low profile cooler rocks on a 12900k ehheeheheheheheh, will hold true for all users? no.

 

[TheinsanegamerN]

This should work. What isn't pictured is the air duct supplying cool air from outside the case. I personally think these are not good for clear panel cases, but for a small form factor like a Dell computer case, it will do the trick. Dell has been using these ugly air ducts for decades and they appear to work well.

The ducting will indeed make a big difference. Directing cold air to a cooler and forcing hot air to exit makes a huge difference in temperatures. I remember old dell systems that could cool 140w pentium Ds with a signle 80mm fan and heatsink combo thanks to ducting. Modern cases using 10 fans blowing ina nd out are really bute forcing cooling.

No different then cars really, take any modern car, remove the ducting aroudn and underneath the front, and watch it overheat within 10 minutes.

 

[Dredi]

I think the REAL joke here is your cooling setup. How on earth can you not manage to cool a 150w ryzen CPU with a 240m AIO? I was cooling an OCed 2700x with a hyper 212 air cooler.

Maybe the pump isn’t plugged in?

 

[Valantar]

Im curious about if we were to have a shroud that wrapped around where the fan sits on the heatsink, covering some of the exposed fins if it would improve cooling at all due to the increased static pressure??

You mean in that little gap between the fan and the top of the heatsink? Likely. Airflow goes the path of least resistance, so if you block that off you'll increase flow through the heatsink by some measure - but you'll also increase flow resistance (which will in turn reduce overall airflow somewhat) and change the acoustic patterns of the cooler, so whether it'll be better overall isn't quite cut-and-dried. Worth a try though! I have an L9i on my travel PC, I might try that out at some point

 

[FreedomEclipse]

You mean in that little gap between the fan and the top of the heatsink? Likely. Airflow goes the path of least resistance, so if you block that off you'll increase flow through the heatsink by some measure - but you'll also increase flow resistance (which will in turn reduce overall airflow somewhat) and change the acoustic patterns of the cooler, so whether it'll be better overall isn't quite cut-and-dried. Worth a try though! I have an L9i on my travel PC, I might try that out at some point

yeah. just add some gaffer tape around here:

on both sides.

 

[Ruru]

For non-K and possible future T SKUs this could be fine especially when the power limits are on.

 

[Valantar]

For non-K and possible future T SKUs this could be fine especially when the power limits are on.

Or a manually tuned K (I mean, you could always leave it stock and just let it bounce off tJmax, but that won't be a pleasant experience). Clearly not worth it if the non-K is less expensive (and available) though.

 

[ShurikN]

They shouldn't have bothered with a 12900k at all. That cpu running at 4.2 GHz is the equivalent of cutting one leg on Usain Bolt and telling him to sprint... Might as well buy something else at that point.
And don't get me even started on the price.

 

[mechtech]

115+ comments on a low profile cooler!! That has got to be a record!!!

cheap cooling 8 months of the year

 

[Xuper]

this is bs marketing. Who does uv 12900K ? Nobody. You don't buy CPU to reduce power and then temperature ! even SFF , like you tell consumer :

Please reduce CPU power to 140w so it ensures low tier cooler can handle temperature .

Noctua shouldn't mention 12900K , only 12600K.

 

[Valantar]

this is bs marketing. Who does uv 12900K ? Nobody. You don't buy CPU to reduce power and then temperature ! even SFF , like you tell consumer :

Please reduce CPU power to 140w so it ensures low tier cooler can handle temperature .

Noctua shouldn't mention 12900K , only 12600K.

You should check out the forums over on smallformfactor.net. People do some crazy stuff. Though perhaps the most crazy is that guy running a full-power 5950X and 3090 in an NFC S4 mini with a single 140mm radiator, I've seen quite a few undervolted workstations geared towards crazy MT performance with high-end CPUs. If you need portability + have a workload that does better with more threads than faster threads, UVing a 12900K or 5950X are perfectly viable options.

 

[First Strike]

You don't buy CPU to reduce power and then temperature ! even SFF

300 watts cpu

How can a single thread attract so many cranks who doesn't even bother to fact check on Google before coining up obvious things.

 

[londiste]

Not many commenting on it here, but that foam duct kit is a brilliant addition. Should be really helpful for this cooler in many relevant situations. Good on Noctua for making it, and making it relatively cheap and flexible.

I was looking at the picture and thinking wtf. But after looking at the site what it is intended for - this is actually quite brilliant. Flexible fan duct, foam will probably help a little with suppression as well.

 

[Valantar]

I was looking at the picture and thinking wtf. But after looking at the site what it is intended for - this is actually quite brilliant. Flexible fan duct, foam will probably help a little with suppression as well.

Yep, plus its softness likely makes installation easier (less chance of it being knocked loose by a ridge on a slide-on side panel or something), and lets it compress slightly if you aren't able to tune the thickness just right, avoiding gaps. Plus, with rigid plastic it would be a nightmare to stack layers like that (or it could just get very rattly).

I just looked up their product page and they have a perfectly SFF-geeky installation video:

IMO, this is pretty much perfect for its target market. The video is clear, concise but still detailed (a smart way of measuring clearance, remembering to take panel thickness into account, etc.), the provided thermal data is exactly how it should be (mentions the case, the CPU, the clock speeds) - it's perfectly geared towards the types of people and builds this is targeting. Well done, Noctua.

The kit is only listed as compatible with the L9 series, as the mounting screws are threaded for that heatsink and only long enough for a 14mm fan. But it should definitely be doable to jerry-rig that onto other 92mm coolers as well.

 

[claes]

I don’t know why this debate is happening… “4.2ghz @160W” isn’t the stock boost of 4.9ghz. Clearly Noctua is telling us that this can’t cool the i9 at full speeds, and certainly not overclocked. They probably just redesigned the heatplate for the new socket size… If anyone’s being disingenuous it’s Metroid claiming that an underclocked i9 pulls the same power as a stock one. :shrug:

 

[Turmania]

Testing methodology is quite different for alder lake cpu's.

 

[Valantar]

Testing methodology is quite different for alder lake cpu's.

Nah, testing methodology for ultra-small SFF PCs has been "different" since CPUs started boosting aggressively. Noctua has used this rating system for at least a couple of years.