In my opinion, the coolers aren't necessarily "better"; they are simply larger and so can cool more with more volume - better (i.e. denser, more "space-efficient") coolers should be able to cool more with the same/less volume (and I suppose weight, price lol)
It would be interesting to see some size-TDP-noise-temperature normalized testing:
I wish they used the 30% TDP cut of the 9070 non-XT Reaper to make the card 30% smaller, it's the only dual slot at all to begin with... (and it's not like the 2.5 slot cards are equivalently shorter to compensate for .5 slot increase)
R9 Nano happened almost a decade ago, 2 slot 1 fan short 175W card,
Nvidia (partners) did similarly with GTX 1080, 2070, 3060Ti, and a single rare late 4070 model,
all of these are 180-200W and fit into less than a "liter" of space, so > 200 "Watts per liter"
AMD (partners) did the same with Powercolor 5700 ITX and (higher TDP!) Vega 56 Nano e.g., but nothing for 6650XT or 7600XT
(R9 nano is 15.5x4x11 e.g. which comes to about 0.7 Liter, so 175 divided by 0.7 is 250W/L)
the 9070 / XT isn't 175-180W, but with this metric, at 220W and 300W, then they should be like 20% and 66% larger. The 9070XT Reaper is close enough...but the 9070 has nothing
Another example is 5070 FE design, we had Zotac 1080Ti Mini 8 years ago, same TDP, same volume. Admittedly ignoring thermals / noise, you would think over 8 years later we would get better designs not same/stagnation.
Apparently the 5070 cooler is worse than the 4070 (S?) FE since one of the fans don't even work properly lol
The only other reasonable length dual slot/fan are from Galax and Inno3D, 3 models not bad, idk if it's available in all regions (let alone at MSRP lol)
Another (crazy) example is Galax 1070 Katana, 4x smaller than 5090 FE (great cooler) for about 1/4 TDP, but...8 years ago (both are >300 W/L, maybe the only non-enterprise/server ones to do so)
I don't know why GPU companies can't just reuse 8-10 year old GPU designs, and I don't know why after 8-10 years we are mostly getting less "dense"/"efficient" cooler designs
If Nvidia (FE) and
Inno3D can make 2 slot (5080) GPUs, (only dual slot 5070Ti/5090 is from Inno3D/FE respectively) then others don't have much excuse, it's not like they are lower length to compensate for thickness
But even Inno3D (while still compact),
used X2 on 4070Ti Super (285W) but not on 5070Ti (300W), despite advertising one; do they really need an extra fan for 15W?
Even if noise or thermals are a concern, nowadays undervolt + PL will cut 25% TDP for <5% perf loss, since cards are pushed for diminishing returns / "overclocked OOTB"
I don't know if AMD/Nvidia partners have to strictly adhere to TDP, or they can go down a small bit (like they can go up on 9070XT, or kinda ignore entirely on laptop)
Ultimately we would need to see testing for if say, a 50% size difference (at same TDP) leads to some mix of 50% noise/temp difference (and need to scale both to linear?)
Sorry for large comment
I want an ITX / low length/single fan RTX 4070 with 16GB VRAM. (Or rather something like the 1080Ti mini for the 4070Ti (Super/16GB VRAM)) It's ~200W just like the numerous ITX 1080/2070(S)/3060Ti, all ~1 "Liter" in size, so ~200 "Watt per Liter" We should be getting better/more...
www.techpowerup.com