So on TV is already atrocious? Got it. 

I am with you that 3rd party developers aren't willing to evolve on playability stuff only visuals, so that's what we are going to see for next years: PS4/X1 games with visuals effects and graphical performance that please gamers that are focused on this aspect. 

So while the next Nintendo console won't match PS5/X1, it can receive those AAA games if the developers find the incentive to make the game run in a playable state (or terrible state for those sensitive eyes that can't play nothing that isn't in ultra performance)

On TV it's going to be docked and can use more pixels then to create a better looking DLSS image. From 720p, DLSS Super Performance mode can totally cook and make a 1440p or 4K image with anti-aliasing too. 

Most important question is production node. Samsung 8nm (like the nvidia 3000 series) or TSMC 7 or 5, 4N make all the difference for performance per watt. Is there any confirmation who is printing those SOCs?

I would have absolutely preferred nVidia 30 series GPU in the Playstation 5 and Xbox Series X as they are better performing parts, especially in Ray Tracing.

But then price becomes a factor.

The difference between a Tegra X1 and Tegra X2 is actually not big, even in regards to the price.

Now underclock the Tegra X1 from 1Ghz to 0.3Ghz and try those benchmarks again.

I never claimed that the Tegra X1 wasn't a high-end part in it's class. Those are your words, not mine.

I claimed there was better options. And there was.

Debatable.

I always want consoles to have the best possible hardware at the lowest possible price, it benefits us, the consumer.
The fact there is stuff all difference between the Playstation 5 and Series X is not a bad thing, but both companies came to similar conclusions in regards to price/performance with their hardware choices. - I would have preferred better hardware as it was available however.

The fact that historically Microsoft, Nintendo and Sony have essentially "picked and chosen" what hardware they want, just like a subway store... I.E. Geforce/Radeon/PowerPC/x86/MIPS/PowerVR etc' flies in the face of your last statement.

Actually agree with you here.

Docked mode would be more beneficial for DLSS as it has more information to build a higher resolution output from.

Still don't wan't DLSS on the next gen Switch however as it's propriety nVidia technology... And we know what happens if a technology/licence (Be it Music or otherwise) expires... The titles are pulled from stores and tend not to be backwards compatible on the next iteration of hardware.

Samsung 8nm node confirmed by pro leaker Kopite7?

https://www.youtube.com/watch?v=qSUYzzEfTm4

That would be less than ideal as Samsung 8nm is a power hungry node (used in the Nvidia 3000 GPUs). We would be looking at less than 400MHz GPU clock in mobile mode and maybe 1 Tflop performance. For comparison: Steam Deck has ca. 1.6 TFlops and is produced with the better TSMC 7nm node.

One comment explains the main issue:

Hey Doc, it's almost certainly not going to be 8nm, simply because 1536 cuda cores at 422mhz already draws 6watts for just the GPU via Nvidia's own numbers and delivers only 1.29TFLOPs, pushing 8 A78C cores even at 1.5w would use 2w, that means for the SoC alone, it would draw 8w, the screen, ram, other components would all draw 3 or 4 watts minimum, but the issue with 8nm is deeper than this, it's the configuration of the chip itself, see a GPU with only 1024 cuda cores at 600MHz would do 1.23TFLOPs, draw about the same amount of power, while also costing much much less. The reality is that Drake is too big of a configuration for these low clocks to make sense economically. It's also worth noting that Kopite7 has specifically gotten nodes wrong, and T239 info wrong, even in your video he had the wrong code name, and thought both Orin and Drake were Ada Lovelace and not Ampere. They are GA10B and GA10F, GA standing for Geforce Ampere, Ada Lovelace is specifically AD10X chips.

He did definitely get Tegra T239 correct, but then seems to have gotten some of the finer details incorrect. Said the codename was "Dane Whitman", when it has "Tim Drake", initially said 2048 CUDA cores I think, and said it was Lovelace. It could be 8N sure, but at 8N even at what they're describing, the chip would be a lot bigger than a Tegra X1, so is Nintendo looking to make something significantly larger this time? Did they look at the Steam Deck and get jealous? lol. What he's describing is a very large device. 

I think the direct leaks from Nvidia are more correct because they are (well) direct actual data leaks from Nvidia themselves that was not supposed to be shared with the public. In those hacks we do see a Tegra T239 but it's listed at 1536 CUDA cores, 12SM configuration, Ampere. At 8nm that chip would be massive, like I believe larger than the chip in the Legion GO (huge ass handheld). 

My guess would be he got some info mixed up as he's made a few mistakes there, or maybe since his leak is very old (March 2021) as well. Who knows maybe Nintendo was planning to launch earlier initially (March 2023?) and moved back to fall 2024 opening the door to a better node process being available. The other thing is in one of his tweets they write "SEC8N ?", the question mark implying they're not sure of that info. 

Well Nintendo had a pro model planned we think and the Pandemic put an end to those plans.

LOL of course it would be the possible worst out come like for the past 20 years. not surprising i wouldn't be surprised if nintendo starting making switch pro and just decided to make the hardware for switch 2 they probably won't have any issues at launch with more then 20 million plus units ready. they probably got a sweet deal.

Where the dude you already lost the bet lol.

I think that  was plans for switch 2. maybe they started the making the chips very early for some kinda special deal and to have enough supply for switch 2. If they wanted to make a pro they could have considering its only gonna do around maybe 5 million units a year it wouldn't be that hard to keep up, its not gonna sell like a switch 2.

From the pastebin: https://pastebin.com/V5nTeh4h

Tegra X1 on TSMC 20nm is 116mm2, while the Tegra X1+ on TSMC 16nm FinFet is 100mm2. Nvidia Orin (full T234) is 455mm2. If we assume that T239 (= Switch2) is about 50% the area, we end up with a die size of 227mm2 on Samsung 8N.

That would be a big boi die. From what I can find on the interwebs, Steam Deck's APU is 163mm² and a desktop 4070 is 294,5mm².