ASUS GeForce RTX 4080 Noctua OC

[Chrispy_]

Highly recommended award eh? I would love to see a noise normalized test between this and a 4090 A 4090 profiled right to match this in noise levels and within acceptable thermals, say 80°C for the hot spot, might still be higher performing, and for 50$ cheaper, and whilst being a smaller card. I am all in for replaceable, off-the-shelf fan setups on graphics cards but this is just absurdly priced. No cooler is worth 400 USD over the same PCB.

Given how absolutely enormous some of the 4090 cards are, how much power can you give them before their fans even need to start?
4090 and 4080-compatible cases need to be huge and well-ventilated anyway, so a passively-cooled 4090 might be able to dissipate ~250W of power at 0dB, and I'd wager a 4090 at 250W is faster than a 4080 at 350W.

 

[W1zzard]

Highly recommended award eh? I would love to see a noise normalized test between this and a 4090 A 4090 profiled right to match this in noise levels and within acceptable thermals, say 80°C for the hot spot, might still be higher performing, and for 50$ cheaper, and whilst being a smaller card.

Interesting idea .. Use a 4090 FE?

 

[DLAV]

Interesting idea .. Use a 4090 FE?

W1zzard, can the cooler profile graph be updated in future reviews to include fan % and RPM next to the GPU temp?

Something like this.

-------

It would also be nice to map the RPM range of the fans. You are already there in afterburner or whatever is used to dial in that 35 dBA. Just set the fans at 100%, note the RPM and dBA. Do the same for 75, 50, 25%. Then set the fans at zero and find the % they need to move. If they need a high % to start, go backwards and find the % they stop at. This way a complete image of the fans is provided and the people who don't like idle-fan-stop know if very low speeds are possible in a custom curve.

 

[Daven]

TPU needs an over engineered award.

 

[izy]

Its just ugly in my opinion

 

[watzupken]

In short, Asus made an overpriced GPU even more overpriced over a couple of Noctua fans that cost below 100 bucks. It will be a niche product that sensible people will not buy since no matter how quiet it is, it still depends on how quiet your entire system is. Plus, it is clearly pricing way over its worth by going into the RTX 4090 price bracket. I feel the overall rating by TPU should serious consider factoring price as a major consideration. How can it be highly recommended when a RTX 4090 makes more sense given the same price point? It is highly recommended if it cost a little over the general RTX 4080 price and if you have the space to house the thick cooling solution.

TPU needs an over engineered award.

They don’t need any more awards. They need to review the factors that determines the score/ awards instead. To “highly recommend” a product that is way overpriced don’t make any sense, even if the product is decent/ good. And really, we are talking about RTX 4080 that cost as much as a RTX 4090, just because of a couple of fans which in actual market value does not cost near 100 bucks. If this is highly recommended, then the RTX 4090 is going to be ”product of the year or 2 years” because it is clearly a superior product.

 

[Vayra86]

Highly recommended award eh? I would love to see a noise normalized test between this and a 4090 A 4090 profiled right to match this in noise levels and within acceptable thermals, say 80°C for the hot spot, might still be higher performing, and for 50$ cheaper, and whilst being a smaller card. I am all in for replaceable, off-the-shelf fan setups on graphics cards but this is just absurdly priced. No cooler is worth 400 USD over the same PCB.

This.
Silly product

 

[hineighbor]

I don't like it, but dammit do I respect it.

 

[megaclite]

DOA
overpriced garbage

 

[madshi]

@W1zzard,

looking at the 4080 reviews, there's one thing confusing me: Basically the "Temperature & Noise Comparison" table seems to contradict the "Cooler Performance Comparison".

To be more specific, looking at the "Cooler Performance Comparison", the MSI Supreme X seems to have the best cooler of all 4080 cards. However, looking at the "Temperature & Noise Comparison", the PNY 4080 seems to be significantly better during gaming, having both better temperatures *and* lower noise at the same time. Isn't that a contradiction?

 

[W1zzard]

Isn't that a contradiction?

At stock the cards run at vastly different power levels, a variable that's eliminated by the normalized cooler testing

I think it's worth adding a note about that. Any thoughts for wording?

 

[madshi]

Ah, that makes a lot of sense, I wondered if it would be that, thanks. Maybe you could simply add the power levels to the "Temperature & Noise Comparison" table, so that it's obvious that the GPUs were measured at different power levels?

 

[Max(IT)]

Anyone who would buy this over a 4090 deserves to get milked by Asus.... Awesome reviews as always just a silly product even vs their own Strix Card.

Well, i wouldn’t buy a 4090 anyway, due to its ridiculous power consumption , but you are right: you can get any other 4080 for much less

 

[oxrufiioxo]

Well, i wouldn’t buy a 4090 anyway, due to its ridiculous power consumption , but you are right: you can get any other 4080 for much less

The 4090 just like the 4080 can be power limited by about 15-20% and lose around 5% performance. Even out of the box in most games it's well under 400w. I have mine set to 85% and It's impossible to notice the difference in performance. power sits around 300-325W in most games. I do the same with my 3080ti comically they use about the same amount of power set up like this the major difference is I get around 80% more performance in heavy RT games sometimes more.

Who ever at Nvidia decided on the default power limits was on drugs.

I did contemplate going with a 4080 as well but it just isn't fast enough for what they are asking for it.

 

[seanmack]

Overkill for a 4080 and unsure why they dont use the chromax black fans with a nicer looking shroud

 

[Yttersta]

Interesting idea .. Use a 4090 FE?

This is based on 4080 TUF as far as I know, so let's be bold and even use a 4090 TUF :-D Still similarly priced and it'd provide a great comparison. But picking the worst 4090 we can find is another option just to rub it in Asus' face on this for their insane pricing. I cannot imagine this cooler over the TUF cooler costing 400$ premium. 100-150? Sure.

 

[Thierry]

Europe
- Noctua 4080 = 1599 euros
- FE 4090 +220 euros
+ not the most expensive 4080
+ the noise - 3db = -50% noise
+ cooling Unique
+ power consumption
YES
About cooling & silence this is the best of his kind.
Excellent choice for a workstation
Work & Play

Nb. Thanks to TechPowerUp for a test with so many details. It could be useful to remind a decibel is exponentiel and not linear, only -3db represents 50% less sound.
Notice The bracket is in 304 Stainless Steel.

Noctua.at - Premium cooling components designed in Austria

Designed in Austria, Noctua's premium cooling components are renowned for their superb quietness, exceptional performance and thoroughgoing quality.

noctua.at

"Noctua Edition of ASUS’ GeForce RTX 4080 is the third graphics card to feature Noctua fans and a heatsink that has been co-engineered by Noctua. By switching from the stock three slim 110mm fans to two of Noctua’s award-winning NF-A12x25 PWM 120mm units and to a more elaborate, tailored heatsink with vapour chamber and five instead of two 8mm diameter heatpipes, thermals and acoustics of ASUS’ original TUF Gaming GeForce RTX 4080 could be improved significantly for the Noctua Edition, making it the quietest card in its class."

 

[Space Lynx]

more I look at these numbers, super impressed that the XTX holds it own

 

[TheDeeGee]

From the fan graph it looks like they still have the 30% fan curve limitation in place, which is stupid since the A12x25 fans can spin at 225 RPM at 10%. There is no reason to instantly spin to 30%.

 

[Chrispy_]

From the fan graph it looks like they still have the 30% fan curve limitation in place, which is stupid since the A12x25 fans can spin at 225 RPM at 10%. There is no reason to instantly spin to 30%.

Heat transfer isn't linear with fan speed though, and neither are noise levels.

You might think you want 225rpm but that's doing very little of value, other than adding more hours of wear to the bearings. For an A12x25 I would expect speeds under 450rpm (20%) to be basically irrelevant in the context of scrubbing the boundary layer from heatsink fins. The fans aren't even audible until about 500-600rpm (to me) so any speed that starts under 600rpm is a perfectly reasonable starting speed.

Low speeds are useful if you're just trying to exchange air in a fixed volume (ie as an unrestricted intake/exhaust fan on a case) but for sinks and radiators where the purpose of the fan is to disrupt the boundary layer and stir air rapidly through turbulence, you really aren't getting much value from the fan until a much higher speed. I'd need to model it to simulate it to even get an approximate estimate of what speed that is, but a very loose rule of thumb is that the whoosh of airflow is usually the turbulence you need from a heatsink. If you can't hear anything, even when you put your ear to it, then the fan's probably not doing very much to a heatsink.

In an ideal world, there's just the right geometry and fin surface area to extract maximum heat exchange from minimum turbulence - which is what will keep your cooling-per-decibel at the best value.