Not sure what's different, I mean in everyday usage ... but not undertstanding what ya looking for perhaps.
1. Both range the cards by performance, THG is subjective ... techpowerup has numbers which I much prefer
2. Both offer the same "at a glance" look
3. Neither provides direct "at a glance" overclocking comparisons. I find this very disappointing ... especially when AMDs cards are OC'ing in single digits and nVidia's can break 30%. If a card is a tier below another card based upon stock performance and you will be using MSI AB, is it not important if one overlcocks 32% versus the other doing 4% or 8% ? ... should that change the rankings ? perhaps not, .... but will it affect my choice ? absolutely.
The differences are:
1. TPU gives ya the tools to get the relative OC comparison. I'd prefer it plainly laid out but it's betetr than nothing. THG does not.
2. THG gives a longer, "historical" listing and lets you say for example how fast is the card I have now verses the one from 5 years ago. But again that's subjective and you can't come away with "it's 3.4 times faster". TPU lists cards within just 2 or 3 generations so if ya wanna compare something today versus 5+ years ago, ya gotta work a little. Compare a 1060 w/ a 760 and then a 760 w/ a 560.
Both suffer from the limitations of their format ... to fit all the cards in, THG can't do the numbers ... to get the numbers in, TPU can't fit as many cards
One other thing I like is seeing multiple resolutions. I find this information invaluable and I will give ya a great example. How many times have ya seen lively discussions about VRAM ... some folks saying X GB is way too low and others saying its fine. My view is
1080p = 3 GB
1440p = 6 GB
2160p = 12 GB
If we are going to argue that 3 GB is inadequate, then it invariably follows that no card today is adequate for 4k as its simply 4 x 1080p. But is there any way to determine if more is needed ? Well some would argue that if GPuz says > 3 GB then it's "case closed". Problem is GPUz doesn't measure RAM usage, it measures RAM allocation. If the install procedure sees 3, it might allocate 2 GB, it it sees 6, it might allocate 4 and so on . Detail on this link
https://www.extremetech.com/gaming/213069-is-4gb-of-vram-enough-amds-fury-x-faces-off-with-nvidias-gtx-980-ti-titan-x
And theres dozens of test that have been performed where they load a game on X GB card and ten load it on 6GB card and there's no difference... even when CPUz says its **using** 3.5 GB on the 6GB card.
Well anyway, the reason I found the test results on the 3GB and 6GB cards at different resolutions useful is this:
a) The 3GB card gets about 6% less fps than the 6 GB card in the same game. But we can't account this to VRAM card because the 3 GB card had 10% less shaders than the 6GB card ... RAM is not the only difference.
b) But, logic dictates, that if 3GB was in any way contributing to that 6% difference at 1080p, it would be way more than 6% at 1440p ... and it isn't; it's the same 6%. Cleasrly, it's not enough for 4k as when we hjump from 1440p to 2160p there is a marked downturn in performance... as would be expected.
But i think we all would agree that the more info a site can provide it's users, the more they will come to that site to get their info. For notebooks we can do this by checking what GPUs we wanna compare and what games , resolutions, settings we want to see compared. Tho the presentation is a bit rough, but ya cab dig thru and see it tho it takes a bit of scrolling.