Lol what? You think his account got banned. Doesn't look like it. If you look at his post from 2013 as someone linked on this thread he clearly mentions that too. I also find it funny that you called that guy an idiot even though he pointed out that the HD 5970 was powerful than the HD7870 with graphs showing the HD 5970 being faster than the HD 7950, makes you look like a bigger idiot as the HD 5970 as he stated, "may lose in games if it doesn't have proper game support or driver support for certain games". LOL.

Regardless if you look at the description of the video:

"This is the award for the most powerful consumer based computer chip not Professional Level Computer Chips (ie. AMD's Fire GL line or nVidia's Quadro GPU lines)".

They aren't saying as it's the best GPU. But the most powerful consumer computer chip.

The company has PC roots. I think every game they made so far has come to the PC.

I didn't see the world "fastest GPU". I see the "Most Powerful Computer Chip".

Well actually they have improved. Shadow of Mordor now run much better at 4K maxed out:

But where is your proof that other 4GB cards doesn't have the same issues?

Of course not. But that wasn't what he was arguing. As post #4 mentioned that different supercomputers have different architectures but one can still have more teraflops than the other.

Also, other supercomputers may perform better than some other in certain applications even though they may not have the most raw teraflops.

Source? Also, the Fury and Fury X, AMD is doing special memory optimization via drivers. It's also dependent on driver updates. That's why you see some stuttering with Shadow of Mordor when the Fury X at 4K everything maxed out because it exceeds the 4GB frame buffer. But with the newer drivers you see that problem has been contained where you don't see the stuttering.

How was he wrong? He wasn't even talking about dual GPU, he was talking about single GPU.

Heck, I could go 4X crossfire and say see it has more teraflops but hey it still loses in some games because you know multi-gpu doesn't work in all scenarios.

It will only go to system RAM when it exceeds the 4GB. I don't see why this is unique to other cards with 4GB GDDR5. They both have 4GB frame buffer.

Of course the HD 5970 wouldn't get 2X the performance of a 5870. Crossfire or SLI doesn't scale linearly. Sometimes may get 70% - 80%+ depending on the game. It all depends on how well the game supports crossfire. And if it does it will beat it as the graph shows.

Frankly, TC was talking about single GPU. I don't even know why you dragged multi-gpu scenarios into the question. Anyone who knows about GPUs knows that multi-gpu configurations can be a hassle or not work properly.

For 1080P you can make that argument. But still people buy it because it's faster. If TC want's to buy it because it's faster than he can do so. Hell, I know one buddy who brought a GTX 970 and then upgraded to a GTX 980 series just because it's faster. A lot of people did so as I mentioned because of the 3.5GB issue with the GTX 980.

And where in the world did you get that 4GB HBM is an issue more so than 4GB GDDR5. They will both have the same 4GB frame buffer. Given the choice I would go with 4GB HBM because it's faster than 4GB GDDR5. Hence R9 Fury Nitro > GTX 980.