• 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)

A next-gen AMD APU in 2026 may not be that big of a step down from the Series X and PS5. I think whatever that chip is that's being worked on at AMD is what Valve is waiting on for the Steam Deck 2, and I would guess is probably being considered for an Xbox and even Playstation handheld as well.
You are far to optimistic imo. Im not expecting it to be much more power efficient than Drake if at all (assuming Drake is on 4nm). Sony and MS might be willing to make a bulkier handheld and go higher on power budget though.

There are inherent disadvantages to using X86 in a handheld and AMD is behind Nvidia.
 
I may well be wrong, but we'll see. No point dragging this conversation further as I don't have much else to add. Not much longer until we find out anyway.
 
On the matter of performance, outside of RAM, suggesting Nintendo would (or can) sacrifice a better SOC to cut costs is not in tune with reality. To be absolutely clear, on 4N T239 would be both performant and affordable, while changing to a smaller design now after years of working with T239 would be uneconomical.
 
I may well be wrong, but we'll see. No point dragging this conversation further as I don't have much else to add. Not much longer until we find out anyway.
We're all wrong in at least one way or another. We have a bunch of info on T239 at least.

I wish we could do multiple polls in a single thread. Do one on RAM, one on Storage, one on Expandable Storage Type and probably a few other things that will become known around announcement/GDC.
 
0
In 2017 the Tegra X1 was still a top performer for graphics processing. Its true that the X1 was originally released in 2015, but it was still going toe to toe with the best SOC's in 2017 when it came to graphics. The top of the line Snapdragon SOC in 2017 started to edge out the Tegra X1, but just barely, and those were only used in devices that were much more expensive than Switch. So when someone makes the claim that the X1 was outdated when Switch launched in 2017, ask them to provide a benchmark from March 2017 or older that shows the Tegra X1 getting bested in a graphics rendering benchmark. When people make these claims, they typically make them boldly, insinuating that not only was the Tegra X1 outdated, but was outperformed by competing SOC's by a significant margin, something that didn't really start materializing for 2-3 years after Switch released.

Fast forward to now with the T239, you can argue that its dated technology as well, but in practice its going to be a top performer in the mobile SOC market when it releases later this year. So while it is technically factual to say its dated tech because the GPU cores are Ampere based and the Arm A78 CPU cores released all the way back in 2020, they will still by competitive in 2024. Assuming T239 is on 4N, it will be about as impressive as the Tegra X1 was back in 2015 when it released.

Nintendo's president recently talked about the longer than normal life cycle with Switch being tied to the software now being unified onto one platform, no longer split between portable and home console. This tells me that Nintendo's plan going forward will be to have generations that do last 7-8 years before releasing a successor. So when we are looking as some potentially choices Nintendo has made as it pertains to SNG, we have to evaluated it with the mindset that Nintendo plans on this new hardware being their primary platform for the next 8 years. Its reasonable to believe this will play into making some choices based on the long term health of the platform rather than choices that save a few dollars in 2024 but result in performance complications down the road. With the tech industry, you also have to consider what will be more mainstream a few years down the road. Going with two 6GB modules is probably cheaper than two 8GB modules, but if things are trending towards phones and tablets using 8GB modules, then that we may see the cost of those come down significantly and 6GB modules could end up being phased out before SNG has reached the end of its life cycle. To summarize, there are choices that may seem a bit premium in 2024 that will pay dividends in a few years when those parts are now used in millions and millions of low cost phones and tablets.
 
and I don't think I would expect that from Nintendo this time around, despite what the reports seem to indicate.
Multiple members of this thread including myself have read the contents of the hack. It is one reason I am particularly stubborn about this, as I do not consider source code to be on the same level of credibility as secondhand reports. Without that hack we would have a much different perspective on what to expect, it's the strongest bit of informed speculation we have and why many are feeling positive about the device. Our 'worst case' estimates have risen since then.

Stuff like RAM, storage, node, etc. are all up in the air at the moment but we have a reason to expect certain values just based on what is suggested by the leaked info. Just to clarify, the leak did not give any definitive information about any of those - but there is enough to suggest that they may not go with, let's say, 8 GB of RAM because it would starve the GPU. It's why the debate has been around 12 vs. 16 GB.

I think I read earlier in this thread that T239 on 8nm would be a larger chip than in the Series S. If so, then I find that to be a diminishing possibility, as I don't expect Nintendo to ship a massive thick device. People also assume it to be 8nm because of Ampere GPUs, but Tegra X1 was fabbed on 20 nm while desktop Maxwell GPUs were all 28 nm. So Nintendo not only went with a top-of-the-line mobile SoC in 2016 but also fabricated it on a smaller node for their purposes.
 
In 2017 the Tegra X1 was still a top performer for graphics processing. Its true that the X1 was originally released in 2015, but it was still going toe to toe with the best SOC's in 2017 when it came to graphics. The top of the line Snapdragon SOC in 2017 started to edge out the Tegra X1, but just barely, and those were only used in devices that were much more expensive than Switch. So when someone makes the claim that the X1 was outdated when Switch launched in 2017, ask them to provide a benchmark from March 2017 or older that shows the Tegra X1 getting bested in a graphics rendering benchmark. When people make these claims, they typically make them boldly, insinuating that not only was the Tegra X1 outdated, but was outperformed by competing SOC's by a significant margin, something that didn't really start materializing for 2-3 years after Switch released.

Fast forward to now with the T239, you can argue that its dated technology as well, but in practice its going to be a top performer in the mobile SOC market when it releases later this year. So while it is technically factual to say its dated tech because the GPU cores are Ampere based and the Arm A78 CPU cores released all the way back in 2020, they will still by competitive in 2024. Assuming T239 is on 4N, it will be about as impressive as the Tegra X1 was back in 2015 when it released.

Nintendo's president recently talked about the longer than normal life cycle with Switch being tied to the software now being unified onto one platform, no longer split between portable and home console. This tells me that Nintendo's plan going forward will be to have generations that do last 7-8 years before releasing a successor. So when we are looking as some potentially choices Nintendo has made as it pertains to SNG, we have to evaluated it with the mindset that Nintendo plans on this new hardware being their primary platform for the next 8 years. Its reasonable to believe this will play into making some choices based on the long term health of the platform rather than choices that save a few dollars in 2024 but result in performance complications down the road. With the tech industry, you also have to consider what will be more mainstream a few years down the road. Going with two 6GB modules is probably cheaper than two 8GB modules, but if things are trending towards phones and tablets using 8GB modules, then that we may see the cost of those come down significantly and 6GB modules could end up being phased out before SNG has reached the end of its life cycle. To summarize, there are choices that may seem a bit premium in 2024 that will pay dividends in a few years when those parts are now used in millions and millions of low cost phones and tablets.
In full agreement here.

I think that as it gets more difficult to get to smaller nodes that the push to get a new generation will lessen as well. There's a wall somewhere for that, and at a certain point, Nintendo can't increase the processing power of the machine at the same cost or reduce the cost of the machine by going for a smaller process.

I think that maybe we get one or two more new generations out of smaller processes, and then in order to get more power, they'll need a console that runs at hotter temperatures and will be reliant on battery technology breakthroughs.
 
Multiple members of this thread including myself have read the contents of the hack. It is one reason I am particularly stubborn about this, as I do not consider source code to be on the same level of credibility as secondhand reports. Without that hack we would have a much different perspective on what to expect, it's the strongest bit of informed speculation we have and why many are feeling positive about the device. Our 'worst case' estimates have risen since then.

Stuff like RAM, storage, node, etc. are all up in the air at the moment but we have a reason to expect certain values just based on what is suggested by the leaked info. Just to clarify, the leak did not give any definitive information about any of those - but there is enough to suggest that they may not go with, let's say, 8 GB of RAM because it would starve the GPU. It's why the debate has been around 12 vs. 16 GB.

I think I read earlier in this thread that T239 on 8nm would be a larger chip than in the Series S. If so, then I find that to be a diminishing possibility, as I don't expect Nintendo to ship a massive thick device. People also assume it to be 8nm because of Ampere GPUs, but Tegra X1 was fabbed on 20 nm while desktop Maxwell GPUs were all 28 nm. So Nintendo not only went with a top-of-the-line mobile SoC in 2016 but also fabricated it on a smaller node for their purposes.
Fair enough. My point was more so addressing the collective set of expectations, including the 16GB of RAM and 512GB of storage some people have mentioned. I'd say 12GB and 256GB are actually the upper limit of what Nintendo will offer, if the chip is as powerful as we've been hearing. I didn't mean to say that the chip itself would necessarily be what's incorrect.
 
I think that maybe we get one or two more new generations out of smaller processes, and then in order to get more power, they'll need a console that runs at hotter temperatures and will be reliant on battery technology breakthroughs.
I do think software/ rendering technologies/ dedicated hardware is the best route forward after that, or even now. RTX is defined by the tensor and RT cores.

Since we can't rely on more powerful tech
at the same rate we are used to, we need more specialized hardware to do the heavy lifting.

The ps5 pro is reportedly doing AI upscaling. Nvidia is doing continuous upgrades to their DLSS package (most recently frame gen and Ray reconstruction).
 
Fair enough. My point was more so addressing the collective set of expectations, including the 16GB of RAM and 512GB of storage some people have mentioned. I'd say 12GB and 256GB are actually the upper limit of what Nintendo will offer, if the chip is as powerful as we've been hearing. I didn't mean to say that the chip itself would necessarily be what's incorrect.
Ask yourself if people need such expectations "addressed", or if they have arrived at such based on an their analysis of what we know and what we have heard.

The chip's power, as has been extensively explored, would not suggest that they underdeliver on specs elsewhere due to budgetary constraints, once again, T239 being powerful does not make it expensive, and indeed on the most likely node, 4N, it is very affordable.

We haven't "heared" the chip is a certain way, we know a lot about it as a matter of fact, and it being powerful would imply better specs elsewhere, not less, because for the final time, a powerful T239 is on 4N, which means it is very affordable. Of course there's also the fact that a more powerful SOC would indeed make use of more RAM, so once again suggesting there is a sacrifice to be made here is facile.

To be clear, 12GB of RAM and 256GB of storage are a firm possibility, but in as objective terms as we can be about this device, not the upper limit.
 
Fair enough. My point was more so addressing the collective set of expectations, including the 16GB of RAM and 512GB of storage some people have mentioned. I'd say 12GB and 256GB are actually the upper limit of what Nintendo will offer, if the chip is as powerful as we've been hearing. I didn't mean to say that the chip itself would necessarily be what's incorrect.
the upper limit would be what insiders have put forth and we've gotten 16GB and 512GB. those are still realistic to be used for a retail unit
 
What’s the GPU specs again in terms of the lowest they can clock it before it no longer saves any battery life. 1.5tflop handheld and 3tflop docked?

Also didn’t people shoot down 16gb ram a month or so when I suggested it because there was no need for any more than Series S?
 
the upper limit would be what insiders have put forth and we've gotten 16GB and 512GB. those are still realistic to be used for a retail unit
512 was always presented as an upper limit by insiders. Nate seems weirdly confident in 16gb, but he is also insistent that it's speculation.

In my opinion 255 and 12gb is a reasonable expectation.
 
Because I look at something like the Logitech G Cloud, which has a worse processor, a quarter of the reported RAM, one eighth of the 512GB of storage, a smaller screen, and no detachable controllers, and it still costs $300.

The Razer Edge (non-5G) has a more comparable processor, 6GB of RAM, 128GB of storage and it's still $400. Expecting Nintendo to excel in value for the specs seems unrealistic.
Nintendo expects this system to sell about 100 million consoles and hundreds of millions of software, so if you buy something for mass production the cost will be lower than if you buy if for a more niche product and less mass produced product.
 
Someone did actually run GFX bench scores on a Switch.

Fun to look at.

Screenshot-2024-02-12-154224.png
Just to clarify here - these benchmarks are on Android.

So no Horizon OS, no NVN, no other low-level console optimizations or whatnot. Real world graphical performance on Switch is better.
 
What’s the GPU specs again in terms of the lowest they can clock it before it no longer saves any battery life. 1.5tflop handheld and 3tflop docked?

Also didn’t people shoot down 16gb ram a month or so when I suggested it because there was no need for any more than Series S?
470'mhz if I recall. There are gains below that, but they are diminishing.

Not sure what that is in tflops, but I don't think you are far off.
 
y'know, I will say this. I know by this point some of you might be tired of my negativity, but I think I've found the bottom line, and I think it's one the tech-minded people in this thread will agree with

at the end of the day, so long as it meets the most basic definition of a hybrid, I can just put it in the dock and never touch it. ostensibly they have prioritized specs over it being decently portable and having a good screen, and at the start of a new gen I think that is the correct decision. earlier today I saw someone, I think bonejack, call it future-proofing and that resonated with me. they are setting a baseline; quality of the rest of the experience can come later.

because here's the kicker: the v1 switch kind of sucked! pretty poor, though decent, battery; dogshit display; and most egregiously an abysmal kickstand. they "finished" the device later with the OLED model.

I bought the switch 1 v1. why should I skip the switch 2 v1 just because it sucks at half its job?

anyway this has been the latest installment of me staring out into the horizon and thinking about nintendo switch 2 I hope you've enjoyed
 
I do think software/ rendering technologies/ dedicated hardware is the best route forward after that, or even now. RTX is defined by the tensor and RT cores.

Since we can't rely on more powerful tech
at the same rate we are used to, we need more specialized hardware to do the heavy lifting.

The ps5 pro is reportedly doing AI upscaling. Nvidia is doing continuous upgrades to their DLSS package (most recently frame gen and Ray reconstruction).
I'd put those under efficiency features. I don't disagree that efficiency is a way forward. I just have no idea how far it takes us. I don't know that once we hit the process node wall if we get something worth a full new generation out of it. I think this is probably a concern for GPU designers as well. Let's say we get a 2031 Switch 3 and it has DLSS 7 or something like that on a 1nm chip. No more die changes available. Are there enough more improvements to DLSS style stuff for another new generation?

Timeline is a total hypothetical. I have no idea if we get to 2nm, 1nm, or sub-1nm or when that wall is.
 
0
Nintendo expects this system to sell about 100 million consoles and hundreds of millions of software, so if you buy something for mass production the cost will be lower than if you buy if for a more niche product and less mass produced product.
That only works to a certain point. Nintendo isn't going to get to a prices much lower than a wholesale distributor of electronics parts gets.
 
What’s the GPU specs again in terms of the lowest they can clock it before it no longer saves any battery life. 1.5tflop handheld and 3tflop docked?
Based on Thraktor's post using clock speed estimates - 1.7 Tflops in portable mode (12 SMs @ 550MHz), 3.4 Tflops docked (12 SMs @ 1.1GHz).
 
In 2017 the Tegra X1 was still a top performer for graphics processing. Its true that the X1 was originally released in 2015, but it was still going toe to toe with the best SOC's in 2017 when it came to graphics. The top of the line Snapdragon SOC in 2017 started to edge out the Tegra X1, but just barely, and those were only used in devices that were much more expensive than Switch. So when someone makes the claim that the X1 was outdated when Switch launched in 2017, ask them to provide a benchmark from March 2017 or older that shows the Tegra X1 getting bested in a graphics rendering benchmark. When people make these claims, they typically make them boldly, insinuating that not only was the Tegra X1 outdated, but was outperformed by competing SOC's by a significant margin, something that didn't really start materializing for 2-3 years after Switch released.

Fast forward to now with the T239, you can argue that its dated technology as well, but in practice its going to be a top performer in the mobile SOC market when it releases later this year. So while it is technically factual to say its dated tech because the GPU cores are Ampere based and the Arm A78 CPU cores released all the way back in 2020, they will still by competitive in 2024. Assuming T239 is on 4N, it will be about as impressive as the Tegra X1 was back in 2015 when it released.

Nintendo's president recently talked about the longer than normal life cycle with Switch being tied to the software now being unified onto one platform, no longer split between portable and home console. This tells me that Nintendo's plan going forward will be to have generations that do last 7-8 years before releasing a successor. So when we are looking as some potentially choices Nintendo has made as it pertains to SNG, we have to evaluated it with the mindset that Nintendo plans on this new hardware being their primary platform for the next 8 years. Its reasonable to believe this will play into making some choices based on the long term health of the platform rather than choices that save a few dollars in 2024 but result in performance complications down the road. With the tech industry, you also have to consider what will be more mainstream a few years down the road. Going with two 6GB modules is probably cheaper than two 8GB modules, but if things are trending towards phones and tablets using 8GB modules, then that we may see the cost of those come down significantly and 6GB modules could end up being phased out before SNG has reached the end of its life cycle. To summarize, there are choices that may seem a bit premium in 2024 that will pay dividends in a few years when those parts are now used in millions and millions of low cost phones and tablets.
Outdated when it comes to consoles is a weird argument. Every console is outdated when launched because its based on tech from several years before the console is launched.
 
ostensibly they have prioritized specs over it being decently portable and having a good screen
No offence, but how would you expect this to be agreeable when this isn't true, as far as we know?

Seriously, T239's not much bigger, if at all, than Tegra X1, and on 4N, which is what they'd want if they "prioritise specs", it wouldn't need a bigger device or more cooling than the original Switch. The larger device allows for a bigger screen and bigger battery, which benefits it as a portable.

The screen, well, we still just plain don't know yet, but since when is a 1080p 8" HDR display not "good"? That seems to be about the expectation, which is, specs wise, better. Also, the screen being "good" doesn't really have an impact on performance at all. It's not like going with an "even better" screen would decrease performance, that's not how these things work.

Also, the "sucks at half its job", which again, not something we know or can know right now and which doesn't seem to be the case based on what we DO know.

I see what you mean by the rest of your comment but this particular statement seems to be a little ludicrous.
 
Based on Thraktor's post using clock speed estimates - 1.7 Tflops in portable mode (12 SMs @ 550MHz), 3.4 Tflops docked (12 SMs @ 1.1GHz).
He estimated that to be the efficiency sweetspot, doesn't mean there aren't gains whe going below that.

Makes no sense to design a custom chip and run it below its efficiency sweetspot though.
 
i don't think May is possible. manufacturing would have to be in full swing and we would have an announcement already.

June/July could be on the cards but things will have to happen pretty fast come March. i'd expect some sort of Funcle action before then too. still feels like Q3/ September is most likely.
 
Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
 
Seriously, T239's not much bigger, if at all, than Tegra X1, and on 4N, which is what they'd want if they "prioritise specs", it wouldn't need a bigger device or more cooling than the original Switch. The larger device allows for a bigger screen and bigger battery, which benefits it as a portable.
switch 1 is a bit too big. switch 2 is bigger than too big. I recognize and respect your position that bigger than too big == "eh the old one demanded special accommodations anyway" but as someone who has joyously stuffed the old switch in places it doesn't belong (weird backpack pocket, my literal actual pants) it's a clear downgrade.

regardless of the subjectivity, I'd contend that making it bigger is a direct response to a priority of power

The screen, well, we still just plain don't know yet, but since when is a 1080p 8" HDR display not "good"? That seems to be about the expectation, which is, specs wise, better. Also, the screen being "good" doesn't really have an impact on performance at all. It's not like going with a "better" screen would decrease performance, that's not how these things work.
I didn't spell this out but the screen specs connection was all about price. early in development they would've planned the chip with the knowledge of a cheapo lcd going into the device in favor of performance. whether that turns out to be the case obviously remains to be seen but it makes more sense to me than going with a premium lcd
 
Nothing. He is just interpreting everything in the most pessimistic way possible.
I'm not sure I entirely agree with this exact framing, but I'd go a step beyond that given things we DO know seem to being disregarded to get to the conclusions drawn, which is a bit more than pessimism.
 
i don't think May is possible. manufacturing would have to be in full swing and we would have an announcement already.

June/July could be on the cards but things will have to happen pretty fast come March. i'd expect some sort of Funcle action before then too. still feels like Q3/ September is most likely.
Don't we already have proof of larger joycons being shipped through customs? Does that not count for the start of production on the new console?
 
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