The original Switch eventually had a peak mode that went up to 921mhz. This was relaxed as Nintendo became more familiar with a chip they inherited, and they wanted to stabilize performance in games. They aren't inheriting a chip this time around, Nvidia is designing one specifically for them. If we go with the original Switch's clock rates, we're looking at 1.38 Tflops in handheld mode, and 2.8 Tflops in docked mode for Switch 2. 

Misremembered, Nintendo only loosened handheld clock rates up to 450Mhz, 921mhz was done with mods and overclocking.

There is a limit to how low you can clock and still get efficiency gains. Thraktor on famiboards calculated it, based on the T234's power-voltage function, to be around 480Mhz. Anything below that would mean you should go with an 8SM instead of 12SM chip that is clocked higher. 

There is also evidence that Nintendo might allow the Switch 2 to run at higher TDP's, given that it seems like the cooling system is beefed up compared to the original Switch. 

Mind however that Ampere TFs are roughly comparable to 60% of RDNA2 TFs (due to doubled FP32 instructions in the former and different IPCs/compute efficiency).

The docked Switch 2 being 2 TF would be underwhelming even for 8 nm. I think they can afford to go a little higher with the dock fan and higher PSU.

I honestly just want the Switch 2 to be at least 1.3TF in Docked

There have been multiple leaks around this from different sources with different images. - Even lines up with the shipping manifest that leaked many months ago.
Currently in the hardware enthusiast circles there is a degree of "faith" that these leaks are legitimate because of that.

However... TLDR.

12GB Ram @ 120GB/s of bandwidth. - Extremely unfortunate it's not 16GB. Bandwidth is good for 720P-1080P gaming.

256GB Solid State Drive @2.1GB/s. - Acceptable, unfortunately likely low IOPS, but competitive with Xbox Series S/X which will be good for porting.

Audio is Realtek. - This seems bizarre as Tegra ORIN has it's own built-in integrated audio functionality complete with power management and drivers... But likely Nintendo has a reason for it, likely due to needing it to dynamically switch between integrated speakers and then pushing it over USB/HDMI.

The Ruiming network is Realtek once again. - Up to 1Gb/s. - Supports USB 3.0, likely required for the Dock. - Orin natively has support for two ethernet connections, one typically used for Cameras/Microphones and the other used for Internet connectivity, so relying on Realtek is another odd choice, likely Nintendo needed some extra functionality that nVidia couldn't offer, but Realtek could.

STMicroelectronics JSTM32G0OB0OCET6 is a small SoC/IC with a 64Mhz ARM M10 processor, 1MB Ram, 512KB of storage likely to assist with I/O relating to the dock.

The SoC is still the unknown.

Anywhere from 1 Terfalop to 3 Teraflops.

Not that it is relevant or important.

Teraflops isn't "meaningless" per-say.
It's just peoples understanding of what it fundamentally is, is the real issue.

Teraflops is a hypothetical (Not real world) metric for single precision floating point performance, it's the same regardless of architecture.
The issue is, there is more to rendering than just floating point... And it doesn't account for that. I.E. Integer, Double Precision, Half Precision, Quarter Precision, Fillrate etc'.

Ironically it's only since Sony has essentially come out and basically said it's a "useless" metric that people are only now jumping on the bandwagon and  starting to think it's a useless metric.

It will have some advantages and disadvantages over a Playstation 4 Pro.
It's a more efficient and modern chip than what the Playstation 4 Pro has, so it can do more work with less.

Hopefully the Switch 2.0 doesn't chase and waste resources on 4k.

Vega was able to dominate in tasks that could use those Teraflops. Think: Folding.

If you can keep the compiler output efficient to keep the dual-issue pipes fed, you can achieve high utilization and get to almost it's realized teraflop potential.

In purely compute tasks AMD hardware is extremely potent.

...It's just games want more than single precision compute and that is where AMD has always fallen short.

All I'm asking for is a stable 60fps in all games. I don't really care or understand anything posted here.

The hardware is irrelevant to achieving 60fps... Even the original NES from the 1980's had 60fps games.

It all comes down to developer priorities, some developers will prioritize visuals over framerate.

In some Nintendo Games, like Mario main games and spin offs, Metroid, Smash Bros, etc. I think Nintendo will prefer to focus on graphics instead 60fps in games like Zelda, Luigi Mansion, Fire Emblem, Xenoblade, Pokemon from GamesFreak, etc.

From Third Party's, honestly I expect many cheap ports than good ones (I'm being realistic). I expect ports at 60fps from Indies, Japanese Companies like Capcom with games like Resident Evil, Japanese small companies, Cod, Fifa and some free to play games like Fortnite, Overwatch, etc. 

Wanna sign my Letter telling Nintendo to prioritize 60fps?

Booo! Give me 60 or give me death.

The console has not been announced yet and u are already with the five stages of grief?

Jokes around, just wait and see, this is just a technical prediction based on hardware leak. Sadly, in my opinion this leak is 90% true.

I expect 540p on low DLSSing to 1080p on the dock, for multiplatform games from the rest of the decade.
And cutting resolution of assets, to get smaller instalation size, it has to fit enough games on thes 256gb.
Let them store the high resolution details on the weigths of the DLSS models, not on the game instalation files.