• Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
  • Furukawa Speaks! We discuss the announcement of the Nintendo Switch Successor and our June Direct Predictions on the new episode of the Famiboards Discussion Club! Check it out here!

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)

Between a nice OLED screen that plays Xenoblade 2 at 360p with fuzzy TAA and a laminated IPS LCD that plays that game at native 1080p with anti-aliasing and all settings cranked up, I'll take the latter.

I'd like a high quality OLED for the latter too, but that feels inevitable, Nintendo loves their hardware revisions. They've done it every single generation. The first model of Switch 2 will not be the last. First models of a new console always have their 'growing pains'. The original DS is the ugliest handheld Nintendo ever made, the DS Lite is one of the best. The original DS was still popular and desirable for the new games and experiences it brought.

It's a safe prediction that next year folks will be foaming at the mouth to play something like a new eye-melting 3D Mario and are willing to make compromises just as with the original Switch and BotW. Not that I think these are particularly large or drastic compromises.
 
yeah this is predicated on it being on top of a table or console or whatever

if me or my sister bought a switch 2 we'd actually have to make room for it
Join Sakurai and I on team ‘laying it flat’.

It looks ugly af but the switch airflow is good enough that it can fit in some pretty small gaps.
 
Join Sakurai and I on team ‘laying it flat’.

It looks ugly af but the switch airflow is good enough that it can fit in some pretty small gaps.

Impossible to join the team with children in the house. You might as well drop it on the floor on purpose. ;D
 
I just get frustrated by people who think Switch 2 is a good name because they can't engage with the idea that Nintendo would ever deliver anything but a 4k60 Breath of the Wild machine, and that Nintendo might want to communicate a more complex idea.
I apologize for how grouchy this was. It got a little over the top at the end.
 
That contradicts what I know about 3nm. 4nm is essentially a optimization over 5nm, and the transistor density on different variations of both do reflect that.

TSMC's N4 (which Nvidia is using) has a transistor density of 143-ish million/mm^2. Compared to that TSMC's 3E node has a transistor density of 215 million/mm^2, which is a sharp 50% increase in density. I don't see how that is not a generational increase.

In fact 2nm will also similarly be an optimization of 3nm, just like 4nm is of 5nm.

Do you have a source for your info?

 
Chill man, I expressed my opinion on how a numbered naming scheme would make sense if the Switch 2 isn't the last of the Switch series, and it is based on the assumption that the Switch 2 is, in fact, just a more powerful Switch. Unlike the past 3 generations this one doesn't seem to have any major new gimmick to it, as far as I can tell so far, so there's not much to communicate.
Yeah, I'm sorry about that. it was frustration about the conversation, not directed at you at all.
 
I feel like naming has two options, but both deliver the same communication to the consumer:

  • Switch 2
  • Something completely different

The message is: This is not an addon, this is not a new version of XYZ, this is THE NEXT console from us.

Any name that's Switch + suffix (outside of a 2), or even prefix + Switch have a too high risk of giving people the wrong idea.

If it's Nintendo Swap, okay. Nintendo Attach, okay too (although too close to 'Attack'). Super Switch, Switch Next, Switch Attach and so on ... nonono badbadbad.
 
That contradicts what I know about 3nm. 4nm is essentially a optimization over 5nm, and the transistor density on different variations of both do reflect that.

TSMC's N4 (which Nvidia is using) has a transistor density of 143-ish million/mm^2. Compared to that TSMC's 3E node has a transistor density of 215 million/mm^2, which is a sharp 50% increase in density. I don't see how that is not a generational increase.

In fact 2nm will also be a similar optimization of 3nm, just like 4nm is of 5nm.

Do you have a source for your info?
2nm is a technology node. img source.
Untitled_design_2023_05_03T103919_711_638186873254688873.png
 
Did a quick and dirty Geekbench 6.3 run on my desktop and laptop, and seeing the single core scores I cannot help but be a little concerned about low CPU clocks on the T239.

https://browser.geekbench.com/v6/cpu/6094682
https://browser.geekbench.com/v6/cpu/6094871

Ofcourse I'm well aware even the 4700U is somewhat faster than the PS5's CPU without multi-threading on paper, but do keep in mind mine's on a Windows environment and heavily power and clockspeed constrained.

Is 2.5GHz sustained simultaneously across all 8 A78C cores a realistic proposition? I know next to nothing about mobile hardware.

Hmmm, be aware that the makeup of the GB6 score can skew towards certain parameters such as; memory speed, clock speed, cache size etc.
if you look at the breakdown of the score and compare it with certain other processors, you'll see that one processor may score very high in a certain task (due to the availability of certain instructions), but in another it won't be as high.

I can stare down and look at the GB6 scores all day long, but it's just one estimate for the CPU to provide some relative comparison, because like you said the 4700U can score higher in ST compared to the PS5 CPU (through 4700S scores) or Xbox Series S|X (through 4800S scores). Moreover, compared to those consoles with GDDR6, LPDDR5X has lower latency, which is beneficial for the CPU [1]

Nonetheless, compared to the Switch, it's a huge jump and if you look at what Nintendo's first party were able to do on that processor it's quite amazing (TOTK alone is a technical marvel).

For 3rd party games, yes it might be a bottleneck in some ways, but unlike your PC, a console is configured to have a sustained performance with fixed performance profiles, for which you can optimise for, and you do that for a frame rate target (30 or 60fps). The OS should have minimal overhead and a fixed amount of resources to perform its tasks in.
Moreover, the Switch 2 is likely not going to be an outlier in the console development, but hopefully a main target platform that needs to be considered while the development is ongoing.

All I've got from these scores so far that it's feasible for the switch successor to clock the A78 to potentially have the same ST score as the current-gen consoles even on SEC8. Of course, it depends on how much power is consumed by the complete cluster, when you clock it that high and if that's realistic in a handheld format or not, but perhaps with the outlook of a smaller process node it all doesn't look that bad.

Moreover, people much smarter than me in the thread have already looked at how the IPC of the A78 compares, but I don't have all the bookmarks for the discussion at that time so I'll need some time to search for them again. Honestly, most of the things we discuss have pretty much been researched extensively within the first ~500 pages (?) 😂 . It's where I found a lot of answers to questions like yours.

[1] https://www.eurogamer.net/digitalfo...review-play-pc-games-on-the-xbox-series-x-cpu
 
That contradicts what I know about 3nm. 4nm is essentially a optimization over 5nm, and the transistor density on different variations of both do reflect that.

TSMC's N4 (which Nvidia is using) has a transistor density of 143-ish million/mm^2. Compared to that TSMC's 3E node has a transistor density of 215 million/mm^2, which is a sharp 50% increase in density. I don't see how that is not a generational increase.

In fact 2nm will also similarly be an optimization of 3nm, just like 4nm is of 5nm.

Do you have a source for your info?
small nitpick: nvidia actually uses TSMC 4N (not N4). "4N" is a node process customized for nvidia, so it's not exactly the same as N4. Weird/confusing, I know.
 
I'm going to be honest - I used to be high on the name "Super Switch" for this new hardware, but everyone has been using Switch 2 when discussing this console for so long that I'm become attached to the name. I feel like if the console has a different name I will be disappointed a bit ngl (doubly so if it's something like "Switch Attach", which I am pretty confident it won't be)
 
0
i'll fight for super nintento swiitch. i don't think it's a good name, it's simply the name i want.
 
I have argued that I think Switch 2 could be a good name, because it could also be used to describe sets of software features that I think are natural extensions of what Switch does, not just because of how it describes the hardware. That doesn't mean I think it's the only good name, or that it will be the name.
This is the specific reason why I prefer and believe in Switch '2'.

The Switch device and concept is inclusive of:
  • switching between tabletop, TV, handheld (and VR) modes
  • high-end console gaming and smaller, more dedicated handheld experiences
  • modularity of tablets, joy cons, and accessories
  • detachable motion controls that can simulate the Wii control scheme
  • additional joy con feature such as HD rumble and the IR camera
  • 'snappy' functionality, instant hybrid switching, quick booting of games, lean OS
  • fun extensions of the form factor, such as AR kart racing, wearable fitness games, and cardboard pianos
The 'Switch' brand, to me, already encompasses so many different ideas and experiences that calling the next device 'Switch 2' can be interpreted as a promise of more flexibility and more unique ideas that can be leveraged by its form factor, and it has the bonus of being immediately clear to the layperson.

What does it mean to have 'just a more powerful Switch'? I see that thrown around a lot. There is 0% chance it is just that. If Nintendo adds a photographic camera, quick resume, microphone, personal assistant to the OS, enhanced motion controls and haptics, speakers to the joy con, new peripherals for the console that attach via magnets, a LIDAR sensor, a pressure sensitive stylus - none of this would betray the concept of a Switch, and would describe a 'Switch 2' just fine. The number '2' can bring excitement and wonder, if what comes before it is already well received.
 
Yes, switch is a wonderful piece of hardware. But If only Nintendo didn't held it back by lowering clocks....
I know it's difficult because of battery life.

Clock is depending on form factor. You can't said PS5 is underclocked because it can goes higher given better cooling
 
This is the specific reason why I prefer and believe in Switch '2'.

The Switch device and concept is inclusive of:
  • switching between tabletop, TV, handheld (and VR) modes
  • high-end console gaming and smaller, more dedicated handheld experiences
  • modularity of tablets, joy cons, and accessories
  • detachable motion controls that can simulate the Wii control scheme
  • additional joy con feature such as HD rumble and the IR camera
  • 'snappy' functionality, instant hybrid switching, quick booting of games, lean OS
  • fun extensions of the form factor, such as AR kart racing, wearable fitness games, and cardboard pianos
The 'Switch' brand, to me, already encompasses so many different ideas and experiences that calling the next device 'Switch 2' can be interpreted as a promise of more flexibility and more unique ideas that can be leveraged by its form factor, and it has the bonus of being immediately clear to the layperson.

What does it mean to have 'just a more powerful Switch'? I see that thrown around a lot. There is 0% chance it is just that. If Nintendo adds a photographic camera, quick resume, microphone, personal assistant to the OS, enhanced motion controls and haptics, speakers to the joy con, new peripherals for the console that attach via magnets, a LIDAR sensor, a pressure sensitive stylus - none of this would betray the concept of a Switch, and would describe a 'Switch 2' just fine. The number '2' can bring excitement and wonder, if what comes before it is already well received.
The excitement comes from Switch, not 2.
 
Personally, i think Switch 2 is a good name, too. Yes, it's boring, but it delivers the message perfectly.

And, i mean going by all rumors, leaks and speculation, this is also simply was the system is. The next generation of Switch. Bigger, better.

So i would prefer Nintendo being boring for once.

If they ever leave the Switch form factor, or are able to "revolutionize" it to another "gimmick", then they can and should drop the Switch name for something new.

Wait, is this a RL Nomura character?

Nomura is belts, not bags.
 
I read it properly, but the number 2 isn’t exciting at all, you could use the same argument for Switch Super, Ultra, Advance or anything else.
The point is that '2' brings excitement if the name preceding it is well received because it's a promise of more of what that thing encapsulates. I'm not saying the number 2, by itself, is an exciting number, 'Switch 2' as a whole is an exciting name, not on an aesthetic level but as a promise. e.g. people being excited by Hades II because Hades was an amazing game.
 
Switch 2 is simply too good from an advertising perspective. “Switch 2 2 adventure, Switch 2 partying” etc.

I also just don’t see any realistic hardware innovations that could easily differentiate itself from the Switch 1 outside of fidelity and maybe a Kinect.
 
Sigh, the point is that '2' brings excitement if the name preceding it is well received because it's a promise of more of what that thing encapsulates. I'm not saying the number 2, by itself, is an exciting name, 'Switch 2' as a whole is an exciting name not on an aesthetic level but as a promise. e.g. people being excited by Hades II because Hades was an amazing game.
Of course, but people are also excited for Hollow Knight: Silksong. The excitement comes from it being a succesor, and as long as the name conveys that it’s fine.
 
I also just don’t see any realistic hardware innovations that could easily differentiate itself from the Switch 1 outside of fidelity and maybe a Kinect.
I wouldn't be surprised by some kind of Kinect feature like what the Pimax Pro does, maybe by setting the joy con with an IR camera on a table.

Though I don't believe this is enough to name the whole console after, Switch already has 'full body AR' type experiences, so the Switch 2 getting enhanced version of those would be expected.

Of course, but people are also excited for Hollow Knight: Silksong. The excitement comes from it being a succesor, and as long as the name conveys that it’s fine.
The point I'm making is that 'Switch 2' can be a good name. I'm not saying that it is the only name Nintendo can use or that it will be the name for sure.
 
Ah, I see. When it comes of semiconductors, the two main types are logic and memory. It is true that for the past few generations node shrink has had decreasing improvements gen over gen for memory applications, so that includes volatile memory such as Cache and different types of RAM. This trend will continue until it grinds to a halt, and then we're going to have to figure out a different way to make memory on silicon.

However, logic-based applications have not run into this problem, yet, which is why we're still seeing node shrinks in applications like CPUs, GPUs, NPUs, etc. Infact after 2nm we're going to be introduced to Angstrom, where next-gen node shrinks will be designated as 20A, 18A, and so on. Otherwise, we wouldn't still be getting any better/faster CPUs/GPUs, would we? Apple just introduced their M4, AMD is gearing up for Zen 5 and RDNA4, Intel is (hopefully) going to compete with Arrow Lake. Nvidia has already decided on 3nm for their next-next generation of GPU after Blackwell, whatever it is called, I can't remember right now.

A perfect example of this is RDNA3, which separated the L3 cache from the main die so they could be manufactured at different lithographies. The 96MB of Infinity Cache is broken into 6 separate 16MB chiplets, manufactured on 6nm while the GPU core is manufactured on 5nm. This way they can enjoy the benefits of cheap memory chiplets and a slightly smaller (thus cheaper) but improved GPU core. This saves a lot of money by not making the whole thing in on a single node (like RDNA2) on 5nm, which would've increased costs, but the L3 cache wasn't going to be reaping the benefits.

In this context, we're talking about logical silicon, an SoC which has a CPU and GPU in the same package. There is memory silicon mixed in there too, but that is the least of our concern. 3nm for a T239 node shrink is a very possible proposition.
 
Last edited:
small nitpick: nvidia actually uses TSMC 4N (not N4). "4N" is a node process customized for nvidia, so it's not exactly the same as N4. Weird/confusing, I know.
My bad, thanks for the correction. I'm kinda overwhelmed with all this googling over the past 24 hours, so I got a lot of stuff mixed up.
 
Yeah, I'm sorry about that. it was frustration about the conversation, not directed at you at all.
We good!

I enjoy this kind of discussion tbh. I'm not an expert by any stretch, but I like speculating and making informed and educated predictions on future computer hardware in general.
 
I'm inclined to think Switch 2 Lite release 2 years after Switch 2, and then Switch 2 Pro (with an OLED screen) whenever PS6 launches.

Yeah any OLED type upgrade will be to juice interest in the Switch 2, which shouldn't really be needed in the first 2-3 years of the console.
 
0
small nitpick: nvidia actually uses TSMC 4N (not N4). "4N" is a node process customized for nvidia, so it's not exactly the same as N4. Weird/confusing, I know.
Me Googling:
"The 4nm (N4) process technology is an enhanced version of 5nm (N5) technology with density improvement. It started volume production in 2022."
"NVIDIA "Blackwell" GeForce RTX to Feature Same 5nm-based TSMC 4N Foundry Node as GB100 AI GPU"
...

Why they do this
 
Hmmm, be aware that the makeup of the GB6 score can skew towards certain parameters such as; memory speed, clock speed, cache size etc.
if you look at the breakdown of the score and compare it with certain other processors, you'll see that one processor may score very high in a certain task (due to the availability of certain instructions), but in another it won't be as high.

I can stare down and look at the GB6 scores all day long, but it's just one estimate for the CPU to provide some relative comparison, because like you said the 4700U can score higher in ST compared to the PS5 CPU (through 4700S scores) or Xbox Series S|X (through 4800S scores). Moreover, compared to those consoles with GDDR6, LPDDR5X has lower latency, which is beneficial for the CPU [1]

Nonetheless, compared to the Switch, it's a huge jump and if you look at what Nintendo's first party were able to do on that processor it's quite amazing (TOTK alone is a technical marvel).

For 3rd party games, yes it might be a bottleneck in some ways, but unlike your PC, a console is configured to have a sustained performance with fixed performance profiles, for which you can optimise for, and you do that for a frame rate target (30 or 60fps). The OS should have minimal overhead and a fixed amount of resources to perform its tasks in.
Moreover, the Switch 2 is likely not going to be an outlier in the console development, but hopefully a main target platform that needs to be considered while the development is ongoing.

All I've got from these scores so far that it's feasible for the switch successor to clock the A78 to potentially have the same ST score as the current-gen consoles even on SEC8. Of course, it depends on how much power is consumed by the complete cluster, when you clock it that high and if that's realistic in a handheld format or not, but perhaps with the outlook of a smaller process node it all doesn't look that bad.

Moreover, people much smarter than me in the thread have already looked at how the IPC of the A78 compares, but I don't have all the bookmarks for the discussion at that time so I'll need some time to search for them again. Honestly, most of the things we discuss have pretty much been researched extensively within the first ~500 pages (?) 😂 . It's where I found a lot of answers to questions like yours.

[1] https://www.eurogamer.net/digitalfo...review-play-pc-games-on-the-xbox-series-x-cpu
I've seen the IPC tables, and it's genuinely impressive. But IPC is only part of the equation; relatively low clocks can still bog you down, which is where my concern stems from. It is a low power tablet, after all. Current Switch's CPU is so massively underclocked that even a (admittedly poorly programmed) mobile game like Genshin could not run on it. Thus games like Dragons Dogma 2 feel uncertain.
 
TX1+ on Samsung 8NM, 6GB of RAM, 128GB eMMC, TN LCD screen at 50hz refresh rate.

If you want to Doom we can doom.

As I said Nintendo Started with Gamecube (not Wii, in my opinion) a power efficiency route, moving away from raw power. Wii and Wiiu in desktop and Nintendo DS and 3DS in portables also followed this path. And only because nvidia had TONS of shields in stock nintendo took X1 as the switch brain. It may be very likely that Nintendo chooses power efficiency and grat profit to make the new switch.

Switch was on a very poor node, the TSMC 20nm, which had leaking issues. We documented it on the old old site, The Nvidia Shield TV would try to run the CPU at 1.9GHz and the GPU at 998MHz, but would throttle whenever both were in medium usage range, the CPU clock would drop to 1.4GHz and the GPU would drop to 921MHz, if both usage was at 100% the clocks would drop even more. Nintendo should have went with a 16nm node, but the chip was designed and manufactured at 20nm, we see with Mariko, just 2 years into the Switch's life, how drastic the power usage dropped.

Switch 2 isn't in this situation though, it's a custom chip, made specifically for Switch 2, and thus, it's size and power consumption will align with the design of the chip, there is no reason to make the chip as big as it is, if you clock it TOO low, because a smaller chip with higher clocks would perform the same and be cheaper. Switch 2's specs are known, and the clocks are obviously in a range where a cheaper chip couldn't do the same job. That is what we know, and with that knowledge, I can confidently say that even on minimum specs for the size of this chip to make sense, it will be far closer to the PS5 than the Switch was to the PS4.

That's the key for me. To close the gab between Switch2 and PS5 making it less than Switch >> Ps4 and so making ports easier.

Clock is depending on form factor. You can't said PS5 is underclocked because it can goes higher given better cooling

Nvidia shield has similar form factor (similar, not same) and has higher clocks. As other user said, Nintendo could made some customizations to X1 and didn't do it.
 
Thanks a lot for the replies all. I'm so excited to see the next 3D Mario, Mario Kart and Zelda on a 3+tflop docked experience considering what they pulled off on 0.5tflops!
I know tflops aren’t everything but that number makes me dread how bad the Wii U was in comparison because I forgot the specs of it
 
Any name that's Switch + suffix (outside of a 2), or even prefix + Switch have a too high risk of giving people the wrong idea.
While you're being more flexible at least, this is the most annoying thing about name discussion. People are so spooked over the Wii U happening again or it having a disruptive gimmick that they keep pushing that "2 or DOA" hyperboles.

Before the Wii U and XB1, there were dozen of non-numbered consoles. Even before the internet era, there wasn't some mass confusion because of names.

Meanwhile, PS is the only console using numbers. PS2 was a huge success, sure. But the numbering didn't save the PS3 from doing much worse than the 3DS (people look at total sales, but 3DS was quite profitable while PS3 burned all the PS1/PS2 profit to get to that 80M). PS4 did much better, but still a signficant decline in userbase compared to the PS2 despite how bad MS and Nintendo messed up the generation. PS5 is slightly trailing behind the PS4 and Sony forecast the gap growing a bit further.
 
Oh, now i can see it. I could not figure it out but with "If 1536 cores is legit, then 4N it is." i see the light. Here my guess:
  • Less than1536 cores - there are exactly zero indication this has happened. we have 1536 from the nvidia leak
  • Samsung's 8nm - not likely with 12SMs
  • Lower than lowest cpu clocks expectations. - i'm not sure what this is supposed to mean. Higher clock does not necessarily mean better. we want the most efficient clock speed
  • 8GB Ram - categorically false based on what we know from the shipment data
  • LCD (not bad, I prefer it over oled) - we don't know if LCD panels we saw in shipment data can be tied to Nintendo/Switch 2. Nobody can.
 
I can't say I understand this perspective. Nintendo Switch basically requires a bag to carry around. The successor hasn't changed that.

Plus, you gotta care about comfort a little. What about playing handheld mode around the house? The comfort matters more than the size there!
Yeah I think they will keep the "bag friendly" aspect in mind. The Switch has always been pretty good when I'm just travelling with a personal item on a flight whereas something like the Deck is too annoying to take with me. Based on the mockups made so far I don't think the increased size will make it much less difficult to fit in a bag.
 
Nvidia shield has similar form factor (similar, not same) and has higher clocks. As other user said, Nintendo could made some customizations to X1 and didn't do it.
The Shield is always plugged in and twice the volume and it still clocks down to basically what the Switch is using at prolonged loads. Another hypothetical chip can clock higher but with the X1 being used, Nintendo used a good clock for what they have.
 
You were talking about battery size, I didn't see your updated post. I was commenting on Switch 2's ability to run the specs within a power budget similar to the original Switch. I expect the Switch 2 to have a battery around 5000mah, the original Switch's was 4310mah, but 5000mah is very common today.

As for the AI comment, I've been discussing Nintendo hardware for almost 25 years online. I predate watson.
A 5000mah battery with 7.7v give us 38 Wh. So Switch OZ can consume 12.8 watts and still run for 3 hours. The original switch consume 9.8 on portable. That is 30% more than the original, what can be the power necessary for a better APU and a bigger LCD screen.
 
Please read this new, consolidated staff post before posting.

Furthermore, according to this follow-up post, all off-topic chat will be moderated.
Last edited by a moderator:


Back
Top Bottom