Getting better, but y'er still 'soft-clipping' some bits.
If noise is a major focus (we're talking fractions of the past, but still relevant to those not wearing headphones), why wasn't the 2x580 vs 2x7950 noise disimilarity noted, instead the 6990 was the focus? Seems odd to not mention that reversal. Also unlike other launches there are non-reference designs as of day 1 with Gigabyte and others offering SLIGHTLY quieter solutions, this also gets little mention, and when using the SAME non-reference design in Don's review for the GTX580 the omission of like for like seems strange one month later. Not surprisingly compare the Gigabyte HD-7970 with the Gigabyte GTX580 and *cough* REMARKABLY *cough* they have similar noise characteristics, but oh my, the temperatures are still dramatically different. Why might that be, and which might be more important then?
Also, why mention fan noise in relation to power consumption instead of fan noise vis-a-vis temperatures? The former seems to have little to do with each other directly whereas the later would have more to do with significant #s. While one might think the lower power consumption would essentially mean less generated heat to dissipate, it would only matter if the temperature #s were similar which they aren't. The 70s #s vs 80s would also imply that the fans are working much harder than needed to reach the competition's level of cooling or really just temperature control (more test would be required for the former, while the existing tests show the later).
As for whether it's worth revisiting the 7970 review, look at it again and see if things haven't changed since then and if the semi-permanence of it reflects the current situation, especially in light of the pre-launch reviews of the last generation change. Doesn't seem balanced in the least, and seems overly dramatic after the launch despite launch day availability and improved launch drivers (which still leave a bit to be desired, but better than the doom & gloom implied).
Definitely when compared to other reviews out on the web now it seems out of place.
PS, in your audio section of the 7950 reviews, you should correct it or alter the statement to reflect your experiences not the product limitations; BluRays support 192/24 audio, not limited to 96/24 the 96/24 is limted to 7.1 but 192/24 is available on 5.1 and lower (like DVD-Audio available on 192/24 2-channel audio) for both LPCM and bitstream (depends on format), and the realtek hardware supports transport of of that, and even theoretically 192/24 on all 8 channels (despite lack of codec & software support). To those of us who care, the difference between 96/24 and 44/16 is noticeable (especially on a good system or good headphones), but 192 v 96 would likely better fit your view of it unlikely being a noticeable difference. Likely the reason you see 96/24 is either the limitations of the BluRays you're chosing or the decoding capabilities of the receiver. Choose concert or music BluRays/DVD-A for content and check your receiver support.
Getting better than the last review, but still some work to do boys... and it'll be interesting to see what happens with next quarter's review (the next review will probably just be the mid-range of the same st-range) so expect little to change in the approach to that.
But enough with the trite pr-stuff (as if a core-i7 2630QM let alone 2620M or LOL i7-2610-UE were anywhere near a desktop core i7 2600K any more/less than the GPUs), especially since you didn't focus on it during true paper launches (how many GF-100 'pre-views did you write?).
Seriously, if I read another mention of numbering/nomenclature in your reviews I'll take it as an admission from you 3 conceeding you don't think TOM's readers know how to use the Google search on their smartphones before plonking down $$ when making a purchase. Really? C'mon focus on something that matters a little more (like actual tests) and is reflective of an authority on TECH not an autority on PR/Maketing/ABCs&123s