Personally, I feel if we focus on the performance of the card vs the power draw, it is not a bad trade off. The card in most cases is keeping up or faster than a 5700 XT which draws 220W or more. However, AMD may have fallen short of the 50% efficiency improvement from RDNA to RDNA2 that they advertised. In my opinion, the only card that I can compare from AMD's RDNA2 lineup is the RX 6800 which see a significant step up in performance for the power it draws over the 5700 XT at around the same TDP.
In any case, I feel the MSRP is very high for a card at this range. The cooler may have been over engineered for this card, resulting in low load temps.
Promises like that are always pretty idealized though - no doubt there's a very specific comparison that manages a 50% improvement, but it won't be a maxed-out SKU with clocks pushed high.
The efficiency gap between the 6700 XT and 6800 illustrates that quite nicely. In a similar vein, RDNA(1) covered a huge efficiency span, with the 5700 XT being pretty bad, the 5700 being good, and
the 5600 XT being amazing for its time before suddenly receiving a last-minute factory OC pushing it into "still pretty good, but no longer class leading" territory. Specific implementations and tuning matters
a lot. As does the test scenario, resolution, etc.
An example: The RX 6800 in TPU's reviews is 61% more efficient than the RX 5700 XT at 2160p, but only 33% more efficient at 1080p. At 1440p it hits the 50% improvement mark dead on. On the other hand, other RDNA(1) SKUs fare quite differently in the same comparison (1080p/1440p/2160p):
Improvement over the 5600 XT (new BIOS): 10% / 27% / 37%
Improvement over the 5700: 11% / 27% / 37%
Improvement over the 5700 XT: 33% / 50% / 61%
For a more like-for-like comparison (high end XT vs. XT): 6800 XT perf/W improvement over 5700 XT: 15% / 33% / 27%
So there's defnitely a 50% improvemnt to be found if you look hard enough, but something slightly above 30% seems far more realistic when comparing the range of implementations of each architecture more broadly across a range of test scenarios. What will be really interesting to see is what smaller RDNA2 SKUs bring to the table in efficiency at lower resolutions. Sadly TPU doesn't post per-resolution efficiency numbers any more, which likely makes it look less efficient overall due to its poor performance at higher resolutions compared to lower ones. The 6600 XT Strix review puts it at 18% more efficient than the 5600 XT overall, but it delivers 33% more performance at 1080p, so assuming both GPUs are at their listed gaming power draw (166W vs. 152W), that's something like a 24% efficiency increase at that resolution. Again, not ground-breaking, and certainly not 50%, but kind of promising for lower SKUs given how far the 6600 XT is pushed in terms of clocks.
But you're probably right - given the difference in my understanding of everything computer related now vs. then, I probably just accepted it as how things were at the time. Of course there's also the question of how much bugs actually manage to bother you if you're playing a game at 20fps in 640x480 
Still quite the increase from the 5-SKU 900 series though!
