Is it worth entertaining clocks lower than OG switch clocks? With a much more advanced node comes more headroom for clocks despite the increase in cores. We see this in the consumer GPU market with new GPUs occupying roughly the same space as their predecessors but boasting more cores and higher clocks at the same time and the cooling on GPUs in PCs haven't really changed much in recent years.
I think if Nintendo wants a battery life more like Mariko and less like Erista then sure, maybe they use the OG 460mhz switch portable GPU clock. But for docked mode they are going to make the absolute most out of that big GPU by clocking it right to the point where it doesn't throttle with its given cooling solution.
My guess, pulled right from my ass based on guestimation for minimum clocks at a given node is.
6nm - 460mhz portable, 900mhz docked - 1.41 TFLOPS portable, 2.76 TFLOPS docked.
4nm - 460mhz portable, 1.1 ghz docked, 3.38 TFLOPS.
Personally don't think we will see much departure from that OG switch 460mhz clock in portable for battery life, but in docked I think they will push it to 1-1.2GHZ, especially if they are targeting higher than HD resolutions and the screen on the switch is still 720p.
No, it isn't really worth entertaining, tbqh... A point comes where there are no gains in battery life, cooler temperature or savings in power consumption. 2017 portable clocks are well below that point. It's also worth remembering that on a 7nm lithograph, Steam Deck (with less efficient x86 architecture and without a docked mode) can hit up to 1.6GHz. With the better GPU, T239 can eclipse it at 550MHz before RT and DLSS. I feel anything in the 500-600MHz range is plausible at the lower end for portable mode. This would be a relatively modest increase from 460MHz, which is understood to be the higher end of the portable boost clocks in the 2017 Switch (About 8% increase at 500MHz, about 18% increase at 550MHz, and about 30% increase at 600MHz), and as your post states 6nm and 4nm, the capacity to have an even higher minimum portable clock is there. However, I feel that the lower end of frequencies are fine in portable mode, as the graphics have more scalability, and will likely start at a lower resolution... I also expect the CPU to be the more significant upgrade across the board (I think a 2-2.2GHz CPU is possible).
We have established that a 1.6GHz GPU clockspeed under worse present-day conditions is possible in Steam Deck for portable mode (OK, 7nm is hardly "bad" conditions, but it isn't the 6 and 4nm in your post...) - 977MHz would be relatively modest for a docked system where ARM architecture is more efficient than x86, putting it in 3TF territory before RT and DLSS. However, we know that the Mariko SoC had a docked GPU boost clock at 1.267GHz. 1.3GHz docked mode on a better lithography process is not only about 19% lower than the Steam Deck's portable clock, it would put it in 4TF territory before RT and DLSS, AND it would be barely a 2% increase on that Mariko boost clock. As for the battery life, there are natural gains in performance and efficiency of the various parts, as well as from the significant improvements in lithography processes.
"Super Switch" I could see (not because I think it's a good name, but because nothing else in the world uses that naming scheme, so there'd be no confusion), but "Switch Ultra" probably falls into the category of suffixes I mentioned in my post.
The thing is, you DO want to leverage the Switch logo and branding to some extent, while also clearly delineating it from it's predecessor. That why I think Switch 2 is the best of both worlds. It's not very Nintendo-like, but if Nintendo could buck tradition with the Switch itself, I think they should with the naming of its successor as well.
I know it isn't a popular view, but I wholeheartedly disagree with the idea that "Switch 2" is the easiest or "best of both worlds". It works for the PlayStation brand because they've sold boxes with the same tired-ass Dual Shock controller variant for 25 years. With regard to hardware specifically, the cardinal numbering convention spells "More of the same". It doesn;t spell "imagination" in the way that Wii, Switch, Game Boy or Dreamcast had done. It isn't the same as with software, where you can see the difference with the screenshots, and in motion. It means an era of predictability, and as I wrote elsewhere, that leads to complacency, stagnation, and ultimately, decline. It doesn't always help an established brand, either. Personally, I like the idea of Switch Advance, and even the relatively safe Switch Ultra. However, I am of the belief that "relative safety" is where we're heading, until new gameplay concepts and form factors emerge (I believe they will at some point) - I feel that Nintendo will adopt the ordinal numbering convention instead, because it would likely be perceived as "different enough" to what PlayStation uses, and older models will be discontinued quietly. So, It would be "Nintendo Switch (2nd Gen, 3rd Gen, 4th Gen, etc.)", but whichever model you own, people will still call it "Switch". Most people don't call their phones "iPhone 14" or "Note 20" but "iPhone" or "Note", and I suspect Nintendo is aiming for something like that. It would be in line with architectural roadmaps, and when there's a new one, they can see the infographics for specs, etc. - I feel Nintendo will do more of this.
My questions just regarding A100 in comparison to what T239 will utilize is mostly due to GA100 SM diagram not having RT cores at all and FP64 units allocated to the die.
A100 SM
GA102 SM
My main concern with TSMC 7nm is that we have the Steamdeck and we know what kind of efficiency that gets on the node.
An 8 CPU core with 12SM GPU part just screams absolute terrible battery life and we know the Switch family of systems will never include a battery anywhere close in size to the Steamdeck (not factoring in Ampere being less efficient than RDNA2) ...
Except x86 devices are much less effiecient than ARM-derivative ones. They also don't have neural units, which the next Switch will have for DLSS, so, their GPUs are pushed more intensively - In Steam Deck's case, its GPU clocks up to 1,6GHz. This is for a portable, when home console clock speeds are 1.565GHz for XSS, 1.8GHz for XSX and 2.2GHz for PS5, hence the poorer battery life. For this successor, we don't know what elements of Lovelace will be on it, but they were reported. Even without, the clock speed wouldn't need to be anywhere near as high as 1.6GHz in portable mode (as mentioned above, it eclipses Steam Deck's 1.6TF at 550MHz, and even a lower frequency hasn't factored in RT or DLSS, or the likelihood that the battery would power a lower resolution screen. So, it's not quite the same here.