@Picao84 Even the gtx 480 and 5870 (both DX11) will boot any current game assuming there isn't a Vram limitation. Wii U is an DX10 variant, possibly with some features from cypress? Anyways, it's not DX11 so yeah no Doom or anything else current without major changes. Dice and 4A would rather troll Ninty on twitter than develop for Wii U and now switch
There's vids of Doom running on 6970 albeit seemingly not well
Wolfenstein does not run on DX11, hence why Kepler is the minimum required GPU on nvidia side (despite promises Fermi did not get proper DX12 support).
But in any case @Recoup, more proof that 1) switch is closer to PS4 / Xbox One performance levels
2) raw power, as you define it, means jack
Doom does not run well at all on a HD 6970, a GPU that boasted 2.7 TFlops of raw power!! The nvidia competitor at the time was the Gtx580, a GPU with 1.6TFlops of computing power, more than one TFlop of different, yet the Gtx 580 was around 30% faster, despite the TFlop deficit!
The OG PS4 has 1.8 TFlops, almost a full TFlop of difference!!! According to your logic, the PS4 is inferior to HD 6970 and should not be able to run Doom better!! But it does it, way way better!
The Switch undocked has around 158 GFlops of raw power, a mere 5.8% of the raw power of the HD6970. It has no business whatsoever running Doom AT ALL, according to your logic. Yet it does.
Here you have: empyrical evidence of how much TFlops in the WiiU GPU architecture means squat!
The WiiU GPU makes use of the same architecture present on the HD6970. That architecture had lots of theoretical raw power, but was extremely inefficient at using it. Why? Because it had a VLIW architecture which is much harder to extract performance, than the modern GCN and Kepler/Maxwell/Pascal architectures.
In other words, the WiiU GPU raw power, in the real world, was fraction of the stated TFlops.
Since you like numbers so much, let's extrapolate the real performance difference between the HD 6970 (Wiiu GPU architecture) and Gtx580 (closer to Switch GPU architecture, although older).
HD6970: 2,7 Tflops
GTX580: 1.6 Tflops
Gtx580 is around 30% faster, so to get HD6970 up to speed let's increase HD 6970 TFlops by 30%, giving us 3.5 TFlops.
Now let's divide 1,6 by 3,5. It gives 45%. So, in the real world, the HD6970 behaves like a GPU with only 45% of its raw power! Shocking... if someone is new to these things, like you are.
Now let's extrapolate to the WiiU vs Switch undocked. 45% of 352 GFlops gives 158 real world GFlops. That is around 40% of the Switch raw power docked! In other words the Switch docked is 2,43 times faster than the WiiU!
And this is even before taking into account that Switch uses Maxwell architecture, which is also more efficient than the gtx 580 architecture was!
This means that Xbox One, being closer in architecture to the PS4 and Switch, has 8 times the raw power of the WiiU (1.3 TFlops vs 158 GFlops), while it is "only" 3.38 times faster than the Switch docked.
Ergo, the Switch is closer to Xbox One than WiiU, in docked mode.
Edit - OK, got a little confused with Switch numbers. I seemed to remember that in undocked mode the Switch had 384 Gflops, but that's actually in docked mode. Unless we consider FP16, which doubles processing power in some situations, the Switch in portable mode is not that faster (circa 20%) than WiiU, although docked it is way more powerful.
Still, the point remains that WiiU would find it much harder to run PS4 / Xbox One ports than the Switch, given the absense of hardware support for the new APUs.
Add to that the whole game isn't even on the cart and the amount of cut corners are evident. And it still dips to the twenties?!
