• 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.

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)

I had forgotten to comment on this before, but when the new Nintendo device is finally announced, I would be very interested in hearing more about the historical information you know about, even if it's filled with redacted details. The currently available timeline of events derived from rumors/leaks/hacks don't seem to connect all that well, so new information that could help make more sense of some things would be a good addition to the thread.
Excited for the eventual Schreier expose.
 
I’m curious, but how significant of an effect does a movie have in people buying a whole new console for a game
The same effect the Pokemon animé has on the selling of video games and merchandise. Media has always been seen as a marketing mechanic for video games, which is why the Megaman Battle Network series has an anime, and so did the Star Force series (but I think that was a special).

The movie is also because Nintendo now has a theme park.

From a Marketing point of view, the end effect will still be a consumption of Nintendo’s brand, which is why Apple’s innovations are centered in their ecosystem of devices:
• Need a phone? iPhone
• Need wireless earphones? AirPods
• Need a Smartwatch? Apple Watch

We can’t compare accessories to media and theme parks, but the idea is the same: hook and maintain the hook. Hook consumers via media/theme/games and keep them there with your products
 
Oldpuck's leaving #teamleapday? Guess I'll have to continue flying the flag solo.

I salute you, but where you go I cannot follow, my brother.
From leakers, it appears that Nvidia's typical tape out to product launch is under a year these days, with around 8 months for the first Ada GPUs. I'd expect a Nintendo chip to have a longer gap than that, between wanting a higher volume launch than is typical for GPUs, and a bit of a buffer for safety (if they need to do a second stepping).
I totally agree that the Switch example isn't ideal, but it's also the only internal example from Nintendo we have. I'm not sure Nvidia is comparable. Nvidia produces some reference designs, but depends on (ie "abuses") downstream manufacturing partners to make cards, which can vary wildly in quality and configuration.

Best data I've got says that Apple turned around the iPhone 14 Pro in 11 months from tape out, so perhaps you're right about the scheduling, but if I had to guess I'd say Nintendo 5th Switch (being generous) is not as refined a development process as Apple 38th iPhone.

Put another way, if Nintendo has enough games to keep Switch alive another year, they have enough games to keep [redacted] going through its first. Sacrificing the latter for the former doesn't make much sense to me.
Yeah, this is a pretty compelling argument, and I agree. But when it comes to timing it out, you're starting from a 2023 target (which you've justified!) and then working toward the delay date. And in that context, I see what you're saying, and I would agree.

When it comes to [redacted], I would be very surprised if Ubisoft had any exclusive games in development.
Also agree.

Sparks of Hope underperformed and I suspect the studio has been assimilated into the Assassin's Creed machine by this point.
At least some rump Ubisoft Milan continues to work on Sparks of Hope DLC, and has been a Nintendo oriented studio since inception. So short of some Assassin's Creed port, which could be done by Virtuous or similar, I imagine that's where a Nintendo game would be.

Having only a small number of people working on a port makes leaks much less likely, both by there just being fewer people to leak, and also because they will know that it's much more likely that any leaker will be found and fired. And while Ubisoft is probably not that concerned about leaks, Nintendo certainly is, and I would expect them to pressure Ubisoft to keep a much tighter leash on teams working on [redacted] than they usually do.
I agree that it is is plausible that the usual 3rd parties are doing their usual thing and that it's a tight ship. But it's not just that Ubisoft is a leaky studio usually, it's that Ubisoft's financial troubles of late shook a lot of leaves off the tree, and still nothing.

All in all, I don't find any date unbelievable, I just happen to weigh the scraps of data we have one way, and I totally understand someone else weighing them another.
Buddy. Are you doing OK? I don't mean that in a mean way, just. This reads. In a concerning way. You good? Hope this thread isn't doing your mental health in. I know I've had to take a few breaks from here for my own mental well-being.
Honestly, probably not great - I was in the middle of an all nighter trying to get some work done when I wrote this - but it isn't the thread's fault. Sorry for being a little loopy and letting it all get out there. Thank you for asking.
 
I'm splitting this off to a separate post as the previous one is long enough as it is, but several people have discussed the apparent inevitability of Ubisoft leaks, and I felt I should chime in.

First off, yeah, Ubisoft are very leaky, and have been for a while. The typical Ubisoft project is developed by a dozen studios and about a thousand people, and the bigger you go with a production the more likely you are to get one or two people leaking what they know about the project. I also suspect that Ubisoft management aren't particularly concerned about these leaks, but even if they were it would be difficult for them to do much about it, given the scale of their operations.

In Switch's first year, Ubisoft had one exclusive game in development, Mario+Rabbids, and several ports. Mario+Rabbids, unsurprisingly, got leaked, with several hundred people working on it. I may be mistaken, but I don't think the ports like Rayman Legends were leaked. Again, probably not surprising, given how few people would typically work on these ports.

When it comes to [redacted], I would be very surprised if Ubisoft had any exclusive games in development. Sparks of Hope underperformed and I suspect the studio has been assimilated into the Assassin's Creed machine by this point. That leaves ports. Assassin's Creed games are by far the most likely of these, for obvious reasons, and I'd say there's a good chance we're getting either AC: Valhalla or AC: Mirage in [redacted]'s launch window. I would have said Mirage previously, but Valhalla is Ubisoft's most successful game of all time, and they could see it in a similar way to Skyrim on Switch, as a major "I can't believe that's running on a handheld" third party open world game for [redacted]'s launch window. Although Mirage is newer, it's reportedly a much smaller scale game than Valhalla, so they may feel it wouldn't be as "impressive" to see running on a portable device.

In any case, if Ubisoft is planning a Valhalla port for [redacted], that would likely mean a small number of people working on it, likely all working in the same studio, possibly not even an Ubisoft studio. Even if Mirage is being developed for [redacted], there's no reason for everyone working on the game to know about a [redacted] port. It's not like an exclusive game such as Mario+Rabbids where everybody involved would have to be aware it's a Switch game. In the scale of Ubisoft's 1000-person productions, the number who actually work directly on individual platforms is pretty small, and with development segmented across multiple studios it wouldn't exactly be hard for one studio to be responsible for the [redacted] build with most employees being unaware it's coming to the platform.

Having only a small number of people working on a port makes leaks much less likely, both by there just being fewer people to leak, and also because they will know that it's much more likely that any leaker will be found and fired. And while Ubisoft is probably not that concerned about leaks, Nintendo certainly is, and I would expect them to pressure Ubisoft to keep a much tighter leash on teams working on [redacted] than they usually do.


As usual, smart post. And if Ubisoft is outsourcing a port, it would be relatively easier to encourage that smaller studio to keep a lid on things since they have much more to lose (future work) than some random Ubisoft chair moistener from sector 7G (who would also likely be oblivious to the outsourcing decisions made by their higher-ups).

(it's just fear of Nintendo ninjas all the way down)
 
Last edited:
This doesn’t really make sense with what he said.
I don't know if the percentage of Switch users who would go out and buy a more advanced model when they can play the same games on the model they already have is that high. It's not like we're talking about 125 million Digital Foundry fans. It's going to be interesting to follow, and complicated to deal with, because Totk's numbers do show that Nintendo would have an immediate interest in doing the same thing that Sony was kind of forced to do with the shortages, but at some point there's going to have to be concrete reasons to buy the new hardware.
 
I don't know if the percentage of Switch users who would go out and buy a more advanced model when they can play the same games on the model they already have is that high. It's not like we're talking about 125 million Digital Foundry fans. It's going to be interesting to follow, and complicated to deal with, because Totk's numbers do show that Nintendo would have an immediate interest in doing the same thing that Sony was kind of forced to do with the shortages, but at some point there's going to have to be concrete reasons to buy the new hardware.
I mean, clearly there’s an audience of people under Nintendo‘s wing who care about a more premium experience. They literally are willing to pay more for a better experience. The switch lite has abysmal sales, while the switch OLED has stellar sales relatively speaking.

The method of delivery is how Nintendo telegraphs their product to the consumer to make them want it.

It just means that people can buy something regardless of what they already have.

The Wii U did not have many cross-gen games and still sold few units.

With so many users owning the Switch, the Switch U has to have reasons to buy it.
Cross gen doesn’t mean “no exclusive software.”

It just signifies a period of transition.
 
I mean, clearly there’s an audience of people under Nintendo‘s wing who care about a more premium experience. They literally are willing to pay more for a better experience. The switch lite has abysmal sales, while the switch OLED has stellar sales relatively speaking.

The method of delivery is how Nintendo telegraphs their product to the consumer to make them want it.


Cross gen doesn’t mean “no exclusive software.”

It just signifies a period of transition.
I don't think there will be many cross-gen games on the Switch U because of the file size that Switch U games will have. It's not like PS4-PS5.
 
I very much doubt nintendo is going to want to make two versions of the same game. They only did that with the 3DS/WiiU and WiiU/Switch because of how badly the WiiU sold and notably even then they still didn't make Wii/WiiU crossgen games.

Assuming they make sure REDACTED has backwards compatibility they would be better off just making sub-60FPS Switch games that on REDACTED will have a simple patch for 1440p/60fps and maybe one or two additional features. It's already at the point where most first party games nintendo has released on the switch the past few years are sub-60fps games anyway.
 
The same effect the Pokemon animé has on the selling of video games and merchandise. Media has always been seen as a marketing mechanic for video games, which is why the Megaman Battle Network series has an anime, and so did the Star Force series (but I think that was a special).

The movie is also because Nintendo now has a theme park.

From a Marketing point of view, the end effect will still be a consumption of Nintendo’s brand, which is why Apple’s innovations are centered in their ecosystem of devices:
• Need a phone? iPhone
• Need wireless earphones? AirPods
• Need a Smartwatch? Apple Watch

We can’t compare accessories to media and theme parks, but the idea is the same: hook and maintain the hook. Hook consumers via media/theme/games and keep them there with your products
Guess it depends on the overlap between the moviegoers who own a Switch and those who don't, which I assume is still big. Mario is a household name and there are kids out there who will have money to spend, as well as adults who are nostalgic for the brand and haven't touched a Nintendo system in years.
I get that, but like the Nintendo switch has over 125 million users and it is over six years old. If it were 24 million, or 37 million, or 48 million, and it was like 2 to 4 years old in range, I could see an affect that’s worthwhile. But the system is not new and it’s already with a very large market. Exactly how much of an affect does the movie have for people buying a platform just for a one game?

We don’t even know if there’s a new game this year (yet), so this is hypothetical.

Currently on the switch, there’s a million Mario games but only 2 (well technically 3) that matter for the general audience: Super Mario Odyssey, Mario Bros Deluxe and Mario Kart.


The first two are the main series games and the rest are spin-offs.


If a new Mario game of the 2D and 3D variety were to release this holiday (doubt both at once, would be a MESS), it would sell absolute gangbusters within the first few weeks.

But, it does not necessarily mean it translates to a lot of hardware sales.

If there’s an OLED Mario for the holiday it can give a temporary boost, but like I said the switch is not a new system and it is not a small system (hehe). It has a massive install base right now, when a hypothetical Mario game with a Mario OLED release in the fall, the system will be over 130M consoles on the market.

I do think that what you two are alluding to does work though for something such as a PlayStation 5 in this scenario, where a Spider-Man movie releases, and the Spider-Man game hits this fall. Why? Because, the PlayStation 5 is still pretty new, and its install base is pretty small. Relatively speaking, of course.

I don't think there will be many cross-gen games on the Switch U because of the file size that Switch U games will have. It's not like PS4-PS5.
I don’t really think that’s a limiting factor. Plus, we don’t know what the file sizes are going to be. And Nintendo games are pretty small.

Wii and Wii U were a wider gap than Switch to Switch 2.

And they were fundamentally different concepts that made it incredibly awkward for a cross gen to work for that. A wiimote for one, and a tablet for the other as a controller.

You have to make two very bespoke versions of the same game for that to work out.

That’s really not comparable here to the switch and presumably the switch 2 that is supposedly similar (not the same) to the switch.
 
Would it be odd to say that I'm firmly sticking to 2023 mostly on the basis that nothing was stated on whether or not new hardware was coming soon or what? Because I still find that suspicious, when they should have said nothing was coming this fiscal year because they're seemingly so confident in moving 15 million units

No, I don't think it's odd. Even though I've swung over to thinking 2024 more likely, @fwd-bwd has had the best alternative perspective I've read to make me pause for thought ...

It seems to me that their confidence in the FY03/2024 forecast can be taken both ways: The hardware sales is still robust, hence no Switch 2 reveal this FY to detract from the goal. Or, the business fundamentals are solid enough that they believe a new model announcement wouldn't torpedo the sales. That brings me to the next point...
... so while I saw [redacted] being ruled out this year for one last holiday push for the Switch, @fwd-bwd could see both sides of the optical illusion. I still think it's 60/40 in favour of 2024, but I've been really enjoying reading some of the comments in the last couple of pages from both sides. Food for thought indeed ... 🤔
 
I agree that it is is plausible that the usual 3rd parties are doing their usual thing and that it's a tight ship. But it's not just that Ubisoft is a leaky studio usually, it's that Ubisoft's financial troubles of late shook a lot of leaves off the tree, and still nothing.

All in all, I don't find any date unbelievable, I just happen to weigh the scraps of data we have one way, and I totally understand someone else weighing them another.
As has been suggested by others, Ubisoft employees probably have more fear of consequences from Nintendo than their own employer. Both in general but especially now. Combined with Ubisoft probably being able to compartmentalize Drake ports of their existing/upcoming lineup fairly effectively, the leak attack surface seems fairly small, even accounting for their current issues.
 
I don’t really think that’s a limiting factor. Plus, we don’t know what the file sizes are going to be. And Nintendo games are pretty small.

Wii and Wii U were a wider gap than Switch to Switch 2.

And they were fundamentally different concepts that made it incredibly awkward for a cross gen to work for that. A wiimote for one, and a tablet for the other as a controller.

You have to make two very bespoke versions of the same game for that to work out.

That’s really not comparable here to the switch and presumably the switch 2 that is supposedly similar (not the same) to the switch.
Nintendo games are pretty small because of the cartridges/money.

The gap between Switch - Switch U will be as big as Wii - Wii U due to file size and performance.

Switch U games will undoubtedly be larger than 16 GB.
 
I mean, clearly there’s an audience of people under Nintendo‘s wing who care about a more premium experience. They literally are willing to pay more for a better experience. The switch lite has abysmal sales, while the switch OLED has stellar sales relatively speaking.

The method of delivery is how Nintendo telegraphs their product to the consumer to make them want it.
There is certainly an enthusiastic audience. I'm not saying there isn't. I'm just saying that we often tend to overestimate the size of the audience because we are a forum of enthusiasts.People bought the OLED because it was the best model available, that's true, but they bought it fot the Switch library, just like the other models in the line.
 
@Thraktor i have a question and I feel like you’re the best person to ask this, but is it possible to backport a feature of a later architecture, even if they are on different foundries?

If i remember right, T239 has that FLCG that Ada has, but ORIN and the other Ampere products don’t have.

Don’t you usually have to be at the same foundry to have something that’s built into the design for more efficiency that it is supposedly meant for?

Or are there instances, especially from Nvidia, in which you can backport a feature from a later architecture at a different foundry to an older architecture? Especially when the later architecture is similar to the old one for the most part?
I'm not Thraktor, but my guess is yes to both questions, with one example being that the Tensor cores on Orin's GPU inherited the doubling of the Half-Precision Matrix Multiply and Accumulate (HMMA), Integer Matrix Multiple and Accumulate (IMMA), and Binary Matrix Multiple and Accumulate (BMMA) instructions from the Tensor cores on the A100.
 
There is certainly an enthusiastic audience. I'm not saying there isn't. I'm just saying that we often tend to overestimate the size of the audience because we are a forum of enthusiasts.People bought the OLED because it was the best model available, that's true, but they bought it fot the Switch library, just like the other models in the line.
But presumedly it'd have the switch's library via BC as well as exclusives to it. Unless you believe that BC isn't there, this is basically how the PS5 played it out.
 
Except, they do. One insider does not all rumours and leaks make.
Ok, show me some os these rumours and I will edit the comment to "Natan and some others don't have any comment about a reease date in this fiscal year .
Forgive me if you were mostly just setting up for the punchlines at the end, but don't forget to add "LinkedIn and Linux info suggesting tapeout was last year" and "leaks suggesting production lines are being set up" to the reasoning here.
The punchlines is only because how ridiculous is to think that Nintendo will wait until 2026 to launch a new generation. And I will add these to the reasoning if you show me from where these leaks and info you put are.
- 2037: Nintendo will wait for the 20 years partnership with Nvidia to end so they can launch Switch’s successor 🤪
Good! hehehe
 
I very much doubt nintendo is going to want to make two versions of the same game. They only did that with the 3DS/WiiU and WiiU/Switch because of how badly the WiiU sold and notably even then they still didn't make Wii/WiiU crossgen games.

Assuming they make sure REDACTED has backwards compatibility they would be better off just making sub-60FPS Switch games that on REDACTED will have a simple patch for 1440p/60fps and maybe one or two additional features. It's already at the point where most first party games nintendo has released on the switch the past few years are sub-60fps games anyway.
Releasing a patch to make the game run natively on the new hardware would make it cross gen.
 
This discussion takes place often but I really think that starting from this premise does not take into consideration how much games are selling Nintendo consoles. It’s not the same for ps5, which sells almost on its own and then sells games. That’s why Sony and Nintendo don’t have the same market positioning.
and what makes you say that Drake can't do the same? we're already seeing a demo shift with the Switch

Owning a Switch doesn't stop people from buying another Switch lol

Would you say the same about the Wii - Wii U?
I have no idea what you're even trying to say here
 
Nintendo games are pretty small because of the cartridges/money.

The gap between Switch - Switch U will be as big as Wii - Wii U due to file size and performance.

Switch U games will undoubtedly be larger than 16 GB.
Er, no, it will not.

The state of technology during the time of Wii and Wii U is a world of difference than the state of technology from around 2020-2022.

The gap between the Wii and Wii U, most notably the GPU, is much larger of a jump than the jump from Switch to Switch 2.


As for filesize, again, we don’t know that. There are PS5 games which are smaller than the PS4 version.

It’s not as simple.

and the file sizes are not a restriction for cross gen. They aren’t going to be >100GB mandatory installs.

There is certainly an enthusiastic audience. I'm not saying there isn't. I'm just saying that we often tend to overestimate the size of the audience because we are a forum of enthusiasts.People bought the OLED because it was the best model available, that's true, but they bought it fot the Switch library, just like the other models in the line.
I meant that there is a sizable amount of people willing to go for the better experience over the standard experience. It’s not insignificant to the extent that people seem to believe.

Having exclusive titles pushes the system, but having patches for older games helps give more value to the system.


Even just res+FPS bumps where applicable.
 
Last edited:
Releasing a patch to make the game run natively on the new hardware would make it cross gen.
If BC through emulation plus a boost to frame rate and resolution counts as crossgen then sure they will probably do that with games that are announced or released before they announce their next hardware but the only new first party Switch stuff I expect to be announced after they announce new hardware is DLC,
 
But presumedly it'd have the switch's library via BC as well as exclusives to it. Unless you believe that BC isn't there, this is basically how the PS5 played it out.
In fact, I was thinking more about the games that people will play in the future than the ones they already had the opportunity to finish on their current switch. Because even though the OLED is a better version, it doesn’t change anything in terms of development.

Given the increasing development time of the games, this requires a lot of resources.

Even if the DLSS were to work miracles, at some point if you only release a Mario or a Zelda or another console seller every five or six years, it would be a waste to release a new hardware without planning these major games especially for.

However, I understand what you mean, I must admit I’m curious to see how they will manage to put for example added value in the switch 2 version of tears of the kingdom.
 
and what makes you say that Drake can't do the same? we're already seeing a demo shift with the Switch
The fact that the Wii u would certainly have sold a little more than 13 million units if it had had since its launch games likely to make people want to buy it.

It seems pretty clear to me that we buy more a Nintendo console for Mario and Zelda than for FIFA and Minecraft.

Having a Nintendo console is the only way to play Nintendo games. That's the main appeal.
 
@Thraktor i have a question and I feel like you’re the best person to ask this, but is it possible to backport a feature of a later architecture, even if they are on different foundries?

If i remember right, T239 has that FLCG that Ada has, but ORIN and the other Ampere products don’t have.

Don’t you usually have to be at the same foundry to have something that’s built into the design for more efficiency that it is supposedly meant for?

Or are there instances, especially from Nvidia, in which you can backport a feature from a later architecture at a different foundry to an older architecture? Especially when the later architecture is similar to the old one for the most part?

I can't really think of any examples of Nvidia backporting features, because usually once they move to a new architecture there aren't any more chips on the older architectures to be backported to. All their previous SoCs since K1, for example, have used at least as recent an architecture as their newest desktop GPU at the time. T239 is a bit of a special case due to Nintendo's requirements.

You could argue that Ada's FP8 tensor core support is backported from Hopper in a way, given that Ada's effectively an updated Ampere architecture. Still, Ada's tensor cores are very different from Hopper's in every other way, so perhaps it would be more appropriate to say it was separately implemented rather than ported.

In terms of moving things from one foundry to another, I feel like the answer is probably yes if you have the time and/or money. There's going to be a time and cost involved in reimplementing a feature or design on a different manufacturing process with different design rules, so it's probably a question of whether it's worth that investment. Qualcomm recently reimplemented an entire flagship SoC from a Samsung process to a TSMC process, which probably wasn't cheap, but may have been justified given the importance of maintaining their dominance of the high-end android phone market.

I'm not entirely sure how big a deal it would be to back port something like FLCG. Firstly because I'm not 100% sure what it is. I can guess it stands for First Level Clock Gating (I'm not sure if that's confirmed or not), but first level could either mean the highest level (ie the most coarse-grained) clock gating, or the lowest level (ie the most fine-grained) clock gating, and Nvidia doesn't seem to document their clock gating approaches anywhere. If it's coarse-grained, then my guess is that it's relatively simple to port, as there are fewer places where you're gating the clock signal. If it's fine-grained, then that would imply that there are a very large number of places you're gating the clock signal, likely deeply embedded within individual hardware blocks, and therefore would probably be much more difficult to port.
 
I just want my Ace Attorney 7
Considering TGA exists I doubt AA7 would struggle graphically (yes they're an HD-ing of a 3ds game but still)

AA7 feels like exactly the type of game to launch on Switch 1 when Switch 2 is already out, "cross-gen" I guess but maybe not even have a specific Switch 2 version, just works due to BC
 
If BC through emulation plus a boost to frame rate and resolution counts as crossgen then sure they will probably do that with games that are announced or released before they announce their next hardware but the only new first party Switch stuff I expect to be announced after they announce new hardware is DLC,
The games wouldn't necessarily still be running in BC mode after the patch, though even if they aren't, I think it's fair to call it cross gen if the game is aware of the hardware it's running on and reacts to it.

In practice, the division between native Switch 1 and Switch 2 games is far smaller than any previous case where Nintendo has continued to support their previous system past its successor launching. It will probably not be uncommon for games to straddle that line and be fully cross-gen, especially for major titles.
 
Nintendo games are pretty small because of the cartridges/money.

The gap between Switch - Switch U will be as big as Wii - Wii U due to file size and performance.

Switch U games will undoubtedly be larger than 16 GB.

No definitely not.

Wii U's CPU is roughly 5x faster than Wii (tri core 1.243GHz vs single 729Mhz) , and Wii U has about 50x more RAM than Wii (88 vs 2GB, though 1GB for the OS). For the GPU, it's 12 gflops vs 176 gflops (14.6x). Note that the CPU and GPU power only accounts for on paper spec frequency speed and not architectural differences.

I didn't even bother with storage and storage speeds 512MB vs 8/32GB.

Switch 2's CPU vs Switch will be a much wider gap than Wii vs Wii U (3 A57 1Ghz cores for gaming vs 7 A78 ~1.5Ghz cores, with an IPC of ~3x for the latter over the former), but it won't have the same power gap in RAM and GPU (not counting DLSS).
 
Last edited:
The fact that the Wii u would certainly have sold a little more than 13 million units if it had had since its launch games likely to make people want to buy it.

It seems pretty clear to me that we buy more a Nintendo console for Mario and Zelda than for FIFA and Minecraft.

Having a Nintendo console is the only way to play Nintendo games. That's the main appeal.
There weren't any desirable games for the wii around wii u launch either. So it's not like cross gen would have helped there either

I don't know what relevance that third party vs first party part has when I'm talking about cross gen games
 
In 2021 this configuration is what I was expecting for Switch 2:

  • GPU doked: 1.5 Tflops docked (thanks to mix precision and more modern architecture is better than ps4)
  • GPU portable: 800 Gflops portable (4x the performance of current model)
  • RAM - 10 GB LPDDR5 memory with a 64-bit bus and a bandwidth of 52GB/s
  • CPU - 6 Arm Cortex-A78 @ 1,6 GHz + 2 Arm Cortex-A55 @ 1,3 GHz
  • Screen - 1080p - will run game with half that resolution and utilize the DLSS to achieve the native resolution
  • DLSS via Tensor cores and RT cores for ray tracing
  • New and improved joycons
  • Q1 2023
  • U$ 399 (64 GB internal memory)
  • U$ 449 (128 GB internal memory)

It's funny to see how weak this machine would be for now and it amazes me that we still believe that the CPU will be the same as we expected in 2021. Worse is the screen, with many people wanting for 720p again.

There is no other better CPU they can use than A78? Or the current A78 is better than the 2021 model?
 
I'm not entirely sure how big a deal it would be to back port something like FLCG. Firstly because I'm not 100% sure what it is. I can guess it stands for First Level Clock Gating (I'm not sure if that's confirmed or not), but first level could either mean the highest level (ie the most coarse-grained) clock gating, or the lowest level (ie the most fine-grained) clock gating, and Nvidia doesn't seem to document their clock gating approaches anywhere. If it's coarse-grained, then my guess is that it's relatively simple to port, as there are fewer places where you're gating the clock signal. If it's fine-grained, then that would imply that there are a very large number of places you're gating the clock signal, likely deeply embedded within individual hardware blocks, and therefore would probably be much more difficult to port.
There appears to be no documentation on FLCG, the only references to a change in Ada's clock gating over Ampere are in marketing materials, and they are only in reference to the memory controller. In fact, all of RTX 40 power savings (that aren't from the improved node) are memory related

  • GDDR6, stock, runs at lower voltages.
  • "Tri-Speed Memory Control" - DVFS for memory controllers.
  • "Improved clock gating" - no other explanation, but only mentioned in context of memory.

The general assumption is that FLCG must be some fine grained clock gating, because coarse grained clock gating is a given, and T239 is documented as the only Ampere GPU to use FLCG. But this may be an internal name for some alternative clock gating strategy.
 
I don't know what relevance that third party vs first party part has when I'm talking about cross gen games

You compare the Sony strategic choices for the Playstation 5 (in a shortage) with what Nintendo will do for the next switch.

I’m trying to explain why I think comparing PlayStation’s strategy with that of the next switch is not really justified. However, it is difficult to talk about their respective positions without evoking the core of Nintendo’s strategy, which is its job as a publisher. It’s their games that sell consoles, not GTA or call of duty.

If you find that this difference is not relevant, then indeed, switch 2 needs no exclusivity to sell.
 
Don't expect most games for the Switch U to be cross-gen. Thet need a reason for people to buy the Switch U.
I don't know if the percentage of Switch users who would go out and buy a more advanced model when they can play the same games on the model they already have is that high.
If no-new-games equaling no-new-console-buy is true, Nintendo wouldn't have built one of its product strategies on selling multiple Switches per household. In FY21/22 Furukawa disclosed that 40% of OLED and 30% of Lite purchases were by consumers who already owned at least another Switch. So I don't see a hypothetical lack of exclusives at launch, or a prolonged cross-gen period, would stop people from buying. Regardless whether consumerism is ecologically sound or even morally right, many consumers do desire newer, better versions of what they already possess—latest phone, faster car, supersize burger, etc.
 
If no-new-games equaling no-new-console-buy is true, Nintendo wouldn't have built one of its product strategies on selling multiple Switches per household. In FY21/22 Furukawa disclosed that 40% of OLED and 30% of Lite purchases were by consumers who already owned at least another Switch. So I don't see a hypothetical lack of exclusives at launch, or a prolonged cross-gen period, would stop people from buying. Regardless whether consumerism is ecologically sound or even morally right, many consumers do desire newer, better versions of what they already possess—latest phone, faster car, supersize burger, etc.
Even though the OLED model is of better quality, it shares the same architecture as all other switches. This will not be the case for the next equipment that will have distinct performances to put to use.

Unlike an OLED screen, its performance will necessarily go through a dedicated development. That changes the situation considerably, especially when you have to wait half a decade between the releases of each big license.
 
0
There is no other better CPU they can use than A78? Or the current A78 is better than the 2021 model?
T239 probably uses A78C, which is a version of A78 designed for "always on" performance. A78C's design makes no sense in phones, was used almost nowhere, and ARM markets it for "game consoles" - and ARM has never returned to the "C" variant in later gens. If you wanted to wildly speculate, you might assume that A78C essentially was a custom Nintendo design.

There have been two ARM CPU gens since the A78, the A710 and the A715. A710 is an unfortunate chip, ARM backtracked on 32 bit support at the last minute and the resulting chip eats too much power without performing better in most cases. A715 might be a good chip for Nintendo, but it's too recent to likely have been used - and if Nintendo thinks they need 32 bit support, then it's not usable.

Short of having a sub 1 hour battery life, the A78C is basically as good as it can get for Nintendo.
 
I don't know if the percentage of Switch users who would go out and buy a more advanced model when they can play the same games on the model they already have is that high. It's not like we're talking about 125 million Digital Foundry fans. It's going to be interesting to follow, and complicated to deal with, because Totk's numbers do show that Nintendo would have an immediate interest in doing the same thing that Sony was kind of forced to do with the shortages, but at some point there's going to have to be concrete reasons to buy the new hardware.
It could be as low as 1% and that would be more than Metroid Prime Remastered has sold.
I can imagine there would be enough interest for them too make money with it..
 
T239 probably uses A78C, which is a version of A78 designed for "always on" performance. A78C's design makes no sense in phones, was used almost nowhere, and ARM markets it for "game consoles" - and ARM has never returned to the "C" variant in later gens. If you wanted to wildly speculate, you might assume that A78C essentially was a custom Nintendo design.

There have been two ARM CPU gens since the A78, the A710 and the A715. A710 is an unfortunate chip, ARM backtracked on 32 bit support at the last minute and the resulting chip eats too much power without performing better in most cases. A715 might be a good chip for Nintendo, but it's too recent to likely have been used - and if Nintendo thinks they need 32 bit support, then it's not usable.

Short of having a sub 1 hour battery life, the A78C is basically as good as it can get for Nintendo.
IIRC on the same process node the 78 and 715 are not dramatically different, so probably not a big deal. I'd imagine shrinking down to 4nm, regardless of which CPU it is, would get you most of the benefits.
 
0
In 2021 this configuration is what I was expecting for Switch 2:

  • GPU doked: 1.5 Tflops docked (thanks to mix precision and more modern architecture is better than ps4)
  • GPU portable: 800 Gflops portable (4x the performance of current model)
  • RAM - 10 GB LPDDR5 memory with a 64-bit bus and a bandwidth of 52GB/s
  • CPU - 6 Arm Cortex-A78 @ 1,6 GHz + 2 Arm Cortex-A55 @ 1,3 GHz
  • Screen - 1080p - will run game with half that resolution and utilize the DLSS to achieve the native resolution
  • DLSS via Tensor cores and RT cores for ray tracing
  • New and improved joycons
  • Q1 2023
  • U$ 399 (64 GB internal memory)
  • U$ 449 (128 GB internal memory)

It's funny to see how weak this machine would be for now and it amazes me that we still believe that the CPU will be the same as we expected in 2021. Worse is the screen, with many people wanting for 720p again.

There is no other better CPU they can use than A78? Or the current A78 is better than the 2021 model?
Before the Nvidia hack happened, My expectations were:
Docked: Xbox one performance
Handheld: Switch Docked performance
6GB of lpddr5 Memory
DLSS
8 CPU cores of a57 clocked at 1.3 Ghz
768 CUDA Cores
128GB of storage
2021 Holiday Release
$400
 
Before the Nvidia hack happened, My expectations were:
Docked: Xbox one performance
Handheld: Switch Docked performance
6GB of lpddr5 Memory
DLSS
8 CPU cores of a57 clocked at 1.3 Ghz
768 CUDA Cores
128GB of storage
2021 Holiday Release
$400
Dude, that is worse than mine. But at least you was expecting it for 2021. My configs was to release on Q1 2023. Maybe what you posted could have been used in the cancelled Nintendo Switch PRO.
 
If they stay in the realms of T239, It'll be great.
The only way they could fuck it up would be to do the equivalent of the Wii duct-taping of Gamecubes.
 
Before the Nvidia hack happened, My expectations were:
Docked: Xbox one performance
Handheld: Switch Docked performance
Same here, actually. As far back as the pre-Fami days I had assumed the new system would run the current docked profile in handheld, then run a "super" profile when docked, for games that were patched to take advantage. That way pretty much all games would work across both systems (with the exception of a few heavy 3rd party AAAs that needed the more power) and devs would only have to design/patch for one additional profile.

Boy was I shocked when I saw what the Nvidia leak laid out. 😅
 
I wanna give credit to @Dakhil here, who set the bar very early on. Before there were really rumors of a Pro, he gamed out what a Switch 4k would look like - Orin, stripped of automotive components, using DLSS 2 to get to 4k. With little info on the ground about what Orin was, the details were wrong - we often speculated about "little Orin" which seemed to be a separate chip (instead of the binned/floorswept chips we got), and the degree of automotive specific technology (the CPU) that would have to be stripped wasn't obvious. But he basically nailed it out the gate, and predicted a lot of the little details way in advance, like using A78C.

@Thraktor's posts guided me toward my final power estimations. Some folks were insisting that, architecturally speaking, 6SMs was the only possibility. Thraktor argued for 4, based on power consumption. When the leak said 12SMs, that was the first indication that something was up. When it became clear a device wasn't coming in early 2023, after months of being slagged on for being insufficiently optimistic, the Orin power numbers bore out his analysis. Great stuff.

Ultimately, that's what makes this fun for me, is seeing smart people get it right, or even when they get it wrong in interesting ways.
 
0
Same here, actually. As far back as the pre-Fami days I had assumed the new system would run the current docked profile in handheld, then run a "super" profile when docked, for games that were patched to take advantage. That way pretty much all games would work across both systems (with the exception of a few heavy 3rd party AAAs that needed the more power) and devs would only have to design/patch for one additional profile.

Boy was I shocked when I saw what the Nvidia leak laid out. 😅
I was shocked when the hack mentioned Ray-Tracing and 1536 CUDA Cores
 
In 2021 this configuration is what I was expecting for Switch 2:

  • GPU doked: 1.5 Tflops docked (thanks to mix precision and more modern architecture is better than ps4)
  • GPU portable: 800 Gflops portable (4x the performance of current model)
  • RAM - 10 GB LPDDR5 memory with a 64-bit bus and a bandwidth of 52GB/s
  • CPU - 6 Arm Cortex-A78 @ 1,6 GHz + 2 Arm Cortex-A55 @ 1,3 GHz
  • Screen - 1080p - will run game with half that resolution and utilize the DLSS to achieve the native resolution
  • DLSS via Tensor cores and RT cores for ray tracing
  • New and improved joycons
  • Q1 2023
  • U$ 399 (64 GB internal memory)
  • U$ 449 (128 GB internal memory)

It's funny to see how weak this machine would be for now and it amazes me that we still believe that the CPU will be the same as we expected in 2021. Worse is the screen, with many people wanting for 720p again.

There is no other better CPU they can use than A78? Or the current A78 is better than the 2021 model?
Both.

So far, the current Armv9 CPUs have worse performance per watt at lower watts than the Cortex-A78 as shown below.
So far, based on Geekerwan's reviews (here and here), the Armv9 CPUs have worse performance per watt in comparison to the Cortex-A78 at lower watts.
ocR1sOg.png

KFEWW3v.png


And Ada Lovelace is practically not that different from Ampere from a GPU architecture standpoint. And Thor, the latest Arm based SoC from Nvidia, has a GPU based on Ada Lovelace, and is planned for release in 2025. So Drake's still the best Arm based SoC from Nvidia Nintendo could use in late 2024/early 2025.

And 6 Cortex-A78 cores and 2 Cortex-A55 cores probably require 2 separate CPU clusters since the Cortex-A78 supports a max of 4 CPU cores per CPU cluster. And more CPU clusters means more latency since there's more hardware components that needs to be communicated with.

The Cortex-A78C on the other hand supports a max of 8 CPU cores per cluster. (There exists a 6 core configuration of the Cortex-A78, but without the Cortex-A55 cores.)
 
If no-new-games equaling no-new-console-buy is true, Nintendo wouldn't have built one of its product strategies on selling multiple Switches per household. In FY21/22 Furukawa disclosed that 40% of OLED and 30% of Lite purchases were by consumers who already owned at least another Switch. So I don't see a hypothetical lack of exclusives at launch, or a prolonged cross-gen period, would stop people from buying. Regardless whether consumerism is ecologically sound or even morally right, many consumers do desire newer, better versions of what they already possess—latest phone, faster car, supersize burger, etc.
I don't think Nintendo wants to start the "next-gen" off on the wrong foot. It is important to maintain the momentum.

I also think they want new consumers; PC-PS consumers.
 
0
I used to have very low expectations until the Nvidia Hack happened.
Back in late 2021 I’d been basing my expectations on the specs given for the Orin NX and was speculating about 1.8 tflops at 10w. I’d assumed the rest of the specs would be similar to the Orin NX (8GB ram, 8 SMs etc)
 
Please read this staff post before posting.

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


Back
Top Bottom