You do realize that your speculation is years outdated. 2 years ago on March 1st 2022, we got T239's specs INSIDE of NVN2 a custom API built for Nintendo by Nvidia, in it we got tests for DLSS, we have cuda core count with those specs confirmed, we have the CPU core count, the memory bus, we even have the GPU clocks... You can speculate all you want about your imaginary version of the console, or you can join the discussion on what the leak said.
1536 Cuda cores (confirmed inside the API to all be active) running at 660MHz portable in a DLSS test, consuming 4.2w on just the GPU, this is 2TFLOPs. A second GPU clock of 1.125GHz or 3.456TFLOPs also found in the DLSS test, this is likely the standard clock when docked, and 1.38GHz for 4.24TFLOPs, a boosted clock. This test was done in summer 2021 before silicon existed, so this last high clock isn't a stress test as originally thought, but a target clock for use, likely as I said, the boosted max clock.
48 Tensor cores, also capable of 2TFLOPs portable (fp16) at the same time as the fp32 cuda cores, and up to 4.24TFLOPs docked fp16 performance, though various other bottlenecks will limit the system's performance, so it's better to think of Switch 2 as half of the PS5's performance, with DLSS offering better than FF7 Rebirth's imagine quality (as this game uses nearest neighbor upscaling), I don't use this game as a means to bash it, but as a means to hype Switch 2's imagine quality.
12 RT cores, this should allow it to out do PS5 ray tracing.
12GB to 16GB ram, with a low chance of 8GB ram, though insiders have heard that it has at least 12GBs for games. LPDDR5 102GB/s or LPDDR5X 137GB/s.
8 A78C ARM cores that perform better per clock than Ryzen 2 CPU cores in PS5/Xseries consoles. clocked somewhere between 2GHz and 2.6GHz. These cores are over 3 times faster than XB1/Switch's CPU per clock, and there is 8 of them, 7 for games... This means at 2GHz, it offers 14 times the performance of Switch's CPU for games, or somewhere around 70% of the PS5's CPU for games.
"Before you lash out because you don't like the numbers you are facing, remember this isn't my opinion, these are the limits which the TDP and production process are imposing - as per Nvidia themselves."
You've overlooked that your opinion is all you are actually going over here. The API tests expose all 1536 cuda cores, and all tests were done on 1536 cuda cores, it's the full GPU, the chip would simply be smaller if it were to use less cores, that is how low clocked chips are, they don't need dead transistors, because lower frequencies improve chip yields. 8nm is pure speculation, and even the person bringing it up was told 3-4TFLOPs, so using it as some means to bring the chip to half that performance is just more nonsense. 8GB RAM doesn't align with the rumors, and we have reports that the Matrix awakens demo ran better than on other current gen consoles. That means the system isn't limited to 8GB ram and half the GPU performance than being reported. The linux kernel exposes the CPU as a single 8 core cluster, so it's 8 A78C cores, none of this is really up for debate anymore, it's been known for 2 years. That is why people like me have moved our discussions away from this thread, because we don't need to speculate any longer. All of these things are basically known, heck performance is more or less known at this point, Switch 2 is about half the performance of the PS5 when docked and something near a PS4 Pro when portable thanks to DLSS improving image quality and the small screen hiding imperfections in the IQ.