Switch 2 Speculation

[Lurkmass]

Indeed, I should have say moving to AMD "or any other vendor" "could" mean no backward compatibility.
As you say, BC is not a simple thing, but granted there is no legal stuff preventing Nintindo to emulate its previous NVidia based hardware, a technical solution may ultimately be found.

Now, I really think Nintendo is sticking to NVidia with Switch 2. It just makes sense, the rumored T239 chip sounds just about perfect for the new console.
I'm just really curious as to what was the reason to recall devkits as this action does not make any sense to me. I just think that letting the devs work on deprecated devkits is better than recalling them and ditch whatever work they were doing. And I guess the only reason this could make sense (to me) is the move to another vendor indeed. But moving to another vendor does not make sense either so...
I'm looking forward the next batch of rumors regarding this situation. Fun times ahead

Considering another vendor would make sense if Nvidia won't provide binary compatibility even if they intended to initially stick with them and it's a more likely scenario than you or anyone would believe. Again. AMD only recently introduced backwards compatible GPU designs and in the past they would often make breaking changes rather than extend the GPU ISA for each generation. The signs of BC for a successor aren't great given Nvidia's history since for every generation they've broken binary compatibility on a constant basis from Maxwell, Pascal, Volta/Turing, Ampere/Ada and now Hopper. They've basically developed 5 distinct architectures that are incompatible with each other in just the span of 8 years. Nvidia did make recent attempts to share a common ISA subset between Volta/Turing only to restart again with Ampere and Ada is their last architecture (compute capability 8.9) to have commonality with Ampere so it seems Nvidia cannot make any commitments to keep compatibility for more than 2 major GPU generations and that's a far cry from them being able to exhibit that their more recent architectures are actually backwards compatible with Maxwell ...

As for using emulation to provide BC, I don't think Nintendo are interested in that route. For more modern systems they've pretty much always relied on a hardware solution for BC and you can see this in their system designs with duplicate components (CPU/GPU). They only entertain the idea for much older platforms so emulation is very likely out of the question. It's either a hardware solution or no BC out of which are the most likely ...

If there were significant hardware changes in the final devkit as opposed to the deprecated devkit then any work developers had invested in the latter is already redundant. That's just the reality with hardware/platform development ...

 

[Chiimaero]

Considering another vendor would make sense if Nvidia won't provide binary compatibility even if they intended to initially stick with them and it's a more likely scenario than you or anyone would believe. Again. AMD only recently introduced backwards compatible GPU designs and in the past they would often make breaking changes rather than extend the GPU ISA for each generation. The signs of BC for a successor aren't great given Nvidia's history since for every generation they've broken binary compatibility on a constant basis from Maxwell, Pascal, Volta/Turing, Ampere/Ada and now Hopper. They've basically developed 5 distinct architectures that are incompatible with each other in just the span of 8 years. Nvidia did make recent attempts to share a common ISA subset between Volta/Turing only to restart again with Ampere and Ada is their last architecture (compute capability 8.9) to have commonality with Ampere so it seems Nvidia cannot make any commitments to keep compatibility for more than 2 major GPU generations and that's a far cry from them being able to exhibit that their more recent architectures are actually backwards compatible with Maxwell ...

As for using emulation to provide BC, I don't think Nintendo are interested in that route. For more modern systems they've pretty much always relied on a hardware solution for BC and you can see this in their system designs with duplicate components (CPU/GPU). They only entertain the idea for much older platforms so emulation is very likely out of the question. It's either a hardware solution or no BC out of which are the most likely ...

If there were significant hardware changes in the final devkit as opposed to the deprecated devkit then any work developers had invested in the latter is already redundant. That's just the reality with hardware/platform development ...

Well, I'm not sure how important is hardware BC vs "emulation". I put emulation with quotes because I think a port of the NVN API is all that's needed, but I may be wrong here.
But what I'm quite sure is that BC will be super important for Nintendo if they want to successfuly handle the transition to Switch 2. Maybe it's just a bias, or wishful thinking, but I don't see how a Switch 2 with no BC can be successful. Switch 2 with no BC fate could be in line with 3DS (best case) or Wii U (worst case)... ie only the fans and the hardcore gamers will be inclined to buy the system.
We'll see.

 

[Lurkmass]

Well, I'm not sure how important is hardware BC vs "emulation". I put emulation with quotes because I think a port of the NVN API is all that's needed, but I may be wrong here.
But what I'm quite sure is that BC will be super important for Nintendo if they want to successfuly handle the transition to Switch 2. Maybe it's just a bias, or wishful thinking, but I don't see how a Switch 2 with no BC can be successful. Switch 2 with no BC fate could be in line with 3DS (best case) or Wii U (worst case)... ie only the fans and the hardcore gamers will be inclined to buy the system.
We'll see.

If you're idea is to create "translation layers" like DXVK or VKD3D-proton then that won't work either and neither will "porting NVN" as well. NVN just like other console gfx APIs are "statically linked" with the game's binaries which allows for optimizations like precompiled shaders (useful for avoiding the coined #stutterstruggle term) and inlining function calls ...

An HLE approach to graphics emulation is doomed for failure since games almost always relies on specific graphics hardware behaviour to render graphics correctly. Nearly all successful emulators out there that you see including the Switch itself requires an LLE approach for their graphics system which gets less feasible the more modern the target platform is ...

Either Nvidia does have a new design (whether it'd be an extension or a 2nd GPU) that's compatible with Maxwell shader binaries or they don't. Backwards compatibility for their successor is very likely going to rest on that fact alone. Steam Deck can't even reach full speed for Switch emulation in a lot of cases and it still remains to be seen if the highest-end handheld solution like the AMD Ryzen Z1 Extreme is capable of such a feat but I don't think their new system will come close in graphics performance near the latter. Touting emulation just isn't feasible right now since there's not enough performance headroom to emulate their previous system with reasonable accuracy ...

 

[Chiimaero]

If you're idea is to create "translation layers" like DXVK or VKD3D-proton then that won't work either and neither will "porting NVN" as well. NVN just like other console gfx APIs are "statically linked" with the game's binaries which allows for optimizations like precompiled shaders (useful for avoiding the coined #stutterstruggle term) and inlining function calls ...

An HLE approach to graphics emulation is doomed for failure since games almost always relies on specific graphics hardware behaviour to render graphics correctly. Nearly all successful emulators out there that you see including the Switch itself requires an LLE approach for their graphics system which gets less feasible the more modern the target platform is ...

Either Nvidia does have a new design (whether it'd be an extension or a 2nd GPU) that's compatible with Maxwell shader binaries or they don't. Backwards compatibility for their successor is very likely going to rest on that fact alone. Steam Deck can't even reach full speed for Switch emulation in a lot of cases and it still remains to be seen if the highest-end handheld solution like the AMD Ryzen Z1 Extreme is capable of such a feat but I don't think their new system will come close in graphics performance near the latter. Touting emulation just isn't feasible right now since there's not enough performance headroom to emulate their previous system with reasonable accuracy ...

Well, you're right, the task at hand is hard.
Yet BC is already important, and will be more and more important moving forward. So whatever the actual technical solution is, be it 100% hardware, partial hardware support, software via HLE, LLE, per game patch or a mix of all of this, they must have smth ready, and they have access to all the resources they need. Hell, I'm convinced throwing in the full X1 maxwell GPU just for BC would be acceptable if that's the only feasible solution.

So to me, it just seems obvious BC will be there, whatever the cost is for Nintendo, because if they don't, I'm not sure how they will explain to their customers that their eShop library is not working on their new machine.

 

[Lurkmass]

Well, you're right, the task at hand is hard.
Yet BC is already important, and will be more and more important moving forward. So whatever the actual technical solution is, be it 100% hardware, partial hardware support, software via HLE, LLE, per game patch or a mix of all of this, they must have smth ready, and they have access to all the resources they need. Hell, I'm convinced throwing in the full X1 maxwell GPU just for BC would be acceptable if that's the only feasible solution.

So to me, it just seems obvious BC will be there, whatever the cost is for Nintendo, because if they don't, I'm not sure how they will explain to their customers that their eShop library is not working on their new machine.

I still wouldn't hazard that they'll look into a software solution given how relatively modern their current platform are ...

Duplicating complex systems like graphics hardware shouldn't be taken lightly. It's die space that could've been spent improving the new platform itself or saved to lower their budget. I'm sure that they'll find a way to justify the loss of BC by delaying their new system with a better lineup schedule of new software if that's a more reasonable outcome for them ...

Besides, Nintendo being able to shop around for other offers that better suits their own needs rather than just taking what Nvidia has by default isn't bad thing by any means ...

 

[Seanspeed]

I'm sure that they'll find a way to justify the loss of BC by delaying their new system with a better lineup schedule of new software if that's a more reasonable outcome for them ...

Things would have to be very bad for them to take this route. Without backwards compatibility, the Switch successor will be doomed to fall very short of the Switch's success. I mean, this is probably inevitable anyways, but no BC will make things much harder.

I think people forget how much third party support has propped up Switch. Getting a 'better lineup' with first party stuff alone will be extremely difficult in these days of ever-growing development cycles, and most importantly - needs to be sustained over time if third party support is lackluster.

If Nintendo are smart, they will do what it takes to achieve BC. But of course Nintendo are known for binning common sense and forging their own path, even when that path is full of thorns and failure, so who knows.

 

[eastmen]

Things would have to be very bad for them to take this route. Without backwards compatibility, the Switch successor will be doomed to fall very short of the Switch's success. I mean, this is probably inevitable anyways, but no BC will make things much harder.

I think people forget how much third party support has propped up Switch. Getting a 'better lineup' with first party stuff alone will be extremely difficult in these days of ever-growing development cycles, and most importantly - needs to be sustained over time if third party support is lackluster.

If Nintendo are smart, they will do what it takes to achieve BC. But of course Nintendo are known for binning common sense and forging their own path, even when that path is full of thorns and failure, so who knows.

maybe but at the same time how will they resell us Mario Kart 8 deluxe deluxe ?

But I agree with you , Nintendo really needs to get bc right this time. It's an easy way to lock people into moving forward with your ecosystem. The switch 2 is a non starter for me if my library doesn't move over esp now that there are other viable portable options for gaming. My steam purchases from launch day of steam work on my deck and will on a deck 2. They also work on the asus rog ally as does game pass and battle.net and so on. For me Nintendo launching a new zelda every what 6 years isn't enough for me to buy a system. Personally I have never taken to the 3d mario games myself and even with zelda , twilight princess really made me check out of the 3d zelda (I'm a much bigger fan of the 2d) games. BOTW brought me back in. They aren't consistent enough imo with their other games. I don't know if there is another smash in the future from them or not. Mario kart is basicly the same one as on my wii u. It even looks the same. Who knows when we will see a proper sequel to it.

I wonder what is really going to drive appeal for a new console from them. I surely thought they would use the new zelda to push a new console. But maybe mario ?

 

[Lurkmass]

Things would have to be very bad for them to take this route. Without backwards compatibility, the Switch successor will be doomed to fall very short of the Switch's success. I mean, this is probably inevitable anyways, but no BC will make things much harder.

Is it ? I reckon most of their big AAA projects will have already had anywhere between 2-4 years of mass production time when their new system launches which will give them a decent lineup schedule to launch successors in quick fashion for their biggest titles from their previous platform ...

The only way I can see them realistically exploring a software solution to BC is if they give up their handheld form factor designs. Do you believe that would be a better arrangement somehow ?

I think people forget how much third party support has propped up Switch. Getting a 'better lineup' with first party stuff alone will be extremely difficult in these days of ever-growing development cycles, and most importantly - needs to be sustained over time if third party support is lackluster.

The vast majority of their third party support comes in the form of late ports and that didn't stop them from being the most successful platform of that generation despite not even featuring BC. Publishers will come around eventually if the new system itself is good which will mostly be contingent on the platform concept and it's first party games regardless ...

If Nintendo are smart, they will do what it takes to achieve BC. But of course Nintendo are known for binning common sense and forging their own path, even when that path is full of thorns and failure, so who knows.

What if their only option their current partner is offering is a redundant 2nd GPU on die ? Are they supposed to somehow accept that deal when other vendors can potentially offer much better licensing costs, benefits and value if BC is not in play ?

Can AAA content developers even be a viable target for new upcoming software if the new system's budget is spent on BC when there's arguably a better chance of receiving new AAA content if the budget goes towards acquiring a more powerful platform. That can mean the difference between being solely stuck with last generation AAA third party content or potentially seeing current generation AAA third party content. Nearly all AAA 3rd party publishers including japanese publishers express discontent when developing for the Switch because they don't feel very productive on the platform ... (it's suboptimal for them to use deferred renderers on the system)

As I see it, they've got to consciously pick between having a portable form factor, a higher productivity system, or ultimately backwards compatibility. They can choose 2 out of the 3 but I imagine having the holy trinity itself is virtually off-limits in the near future ...

 

[eastmen]

As I see it, they've got to consciously pick between having a portable form factor, a higher productivity system, or ultimately backwards compatibility. They can choose 2 out of the 3 but I imagine having the holy trinity itself is virtually off-limits in the near future ...

But why? the xbox series s and x can play last gen xbox games and all the way back to the original. The steam deck can play games all the way back to steams first launch and even further if you boot into the os. Why wouldn't Nintendo be able to emulate the switch ? Hell the steam deck can emulate the switch and offers better performance and resolutions than the switch does natively.

The switch launched in march of 2017 so we are already 6 years into its life span. Realistically we wont see a switch 2 until 2024 at this point so that would be 7 years. The xbox one launched in 2013 and the series s/x launched in 2020 also 7 years. If MS can do it why couldn't Nintendo ? If valve can do it with 2021 hardware why couldn't nintendo with 2024 hardware?

I mean think about it. Nintendo would really only have to offer something on par with the steam deck but in 2024 or even 2025 to have switch bc. On a new process in that time frame they should be able to get good battery life. The original switch would get 3 1/2 hours with BOTW. So the bar is set pretty low regarding battery life

 

[cheapchips]

If MS can do it why couldn't Nintendo ? If valve can do it with 2021 hardware why couldn't nintendo with 2024 hardware?

MS and Valve invested heavily in BC. Valve arguably more than even MS did. SteamOS and associated stuff has been a decade long effort.

I'm sure Nintendo could offer a BC solution with the power they should have in Switch 2. For me, I'd say it's a question as to whether they have the motivation and technical resources to implement it. Have Nvidia onboard could help with the latter part.

 

[Chiimaero]

Is it ? I reckon most of their big AAA projects will have already had anywhere between 2-4 years of mass production time when their new system launches which will give them a decent lineup schedule to launch successors in quick fashion for their biggest titles from their previous platform ...

The only way I can see them realistically exploring a software solution to BC is if they give up their handheld form factor designs. Do you believe that would be a better arrangement somehow ?

The vast majority of their third party support comes in the form of late ports and that didn't stop them from being the most successful platform of that generation despite not even featuring BC. Publishers will come around eventually if the new system itself is good which will mostly be contingent on the platform concept and it's first party games regardless ...

What if their only option their current partner is offering is a redundant 2nd GPU on die ? Are they supposed to somehow accept that deal when other vendors can potentially offer much better licensing costs, benefits and value if BC is not in play ?

Can AAA content developers even be a viable target for new upcoming software if the new system's budget is spent on BC when there's arguably a better chance of receiving new AAA content if the budget goes towards acquiring a more powerful platform. That can mean the difference between being solely stuck with last generation AAA third party content or potentially seeing current generation AAA third party content. Nearly all AAA 3rd party publishers including japanese publishers express discontent when developing for the Switch because they don't feel very productive on the platform ... (it's suboptimal for them to use deferred renderers on the system)

As I see it, they've got to consciously pick between having a portable form factor, a higher productivity system, or ultimately backwards compatibility. They can choose 2 out of the 3 but I imagine having the holy trinity itself is virtually off-limits in the near future ...

No one can read into the future. But yes, I think investing in BC is critical. Now that the eShop is a thing, no BC basically means you're messing up with your customers. That's not acceptable anymore.
Sure, it would be impossible to sell MK8 again, and same is true for most of their games at this point. Their 'style' works fine with less powerful hardware, and for many people, the Switch games are just good enough. You won't sell a new machine to those guys with the only intent to sell the same games again wil slighltly better visuals. Not when they bought them on the eShop. Not when they already looks fine. Not when every other vendor has BC now. So if the choice is to give in 5% of perfs because they need the old X1 GPU to be part of the new design, of course it's a good trade-off. It's either that, or take the risk to alienate all your hardcore fans.

Now, if you can't sell MK8 again, what to do then? That's the tricky part. You obviously sell MK9, but you need the new game to be different enough. So you need time, a lot of time to let new ideas emerge.
My personal feeling is that no matter what Nintendo wants, the market does not need more than one Nintendo machine every 10 years or so. And this may even stretch to 12 or 13 (or more) once the move to HD/4K is done.
We all feel it. MK8 is an outstanding product, but it's basically the same game as MK Wii. Same for later 2D Mario. Obviously same for the shitload of remasters.
I think that as customers, we need a bit of a dry period before the need to (re)play some Nintendo game comes back.

So in the end, my expectations for Nintendo moving on:
- work hard to keep their existing gamers base happy by offering more of everything (BC obviously, better ecossytem, better online, more fan service)
- make sure their new machine is solid enough to stay relevant for 10 years or more so that third party support stays solid for as long as possible
- operate a smooth transition to the new machine with new Switch 1 games available for a few years more and only the bigger games exclusive to the Switch 2
- take time to invest on their major games (they already do that pretty well)
- continue to invest in porting their legacy games to the Switch platform (this includes Switch 2) to fill-in the blanks
- continue to print money with they non video game branches

Trying to force sell some new machine based on its advanced technical capabilities only is a path to failure. Nintendo investors may disagree though.

 

[Lurkmass]

But why? the xbox series s and x can play last gen xbox games and all the way back to the original. The steam deck can play games all the way back to steams first launch and even further if you boot into the os. Why wouldn't Nintendo be able to emulate the switch ? Hell the steam deck can emulate the switch and offers better

What happened with the Series X/S along with the PS5 from AMD were exceptional circumstances that has very little chances of being replicated elsewhere. RDNA wasn't quite as clean of an architecture design as we would normally think of for GPUs. There were design compromises like having multiple SIMD modes (wave32 vs wave64), having two distinct geometry pipelines (legacy vs NGG), and it brought back all of the dropped instructions from GFX7 IP GPUs (last gen consoles) to make BC possible ...

The switch launched in march of 2017 so we are already 6 years into its life span. Realistically we wont see a switch 2 until 2024 at this point so that would be 7 years. The xbox one launched in 2013 and the series s/x launched in 2020 also 7 years. If MS can do it why couldn't Nintendo ? If valve can do it with 2021 hardware why couldn't nintendo with 2024 hardware?

It's explained in the above paragraph ...

I mean think about it. Nintendo would really only have to offer something on par with the steam deck but in 2024 or even 2025 to have switch bc. On a new process in that time frame they should be able to get good battery life. The original switch would get 3 1/2 hours with BOTW. So the bar is set pretty low regarding battery life

Steam Deck is somewhat capable of emulating the system but as we see with the latest AAA release, there are still visual anomalies compared to the original system. A rule of thumb for software emulation is that as the solution gets more accurate (cycle accurate > LLE > HLE) the solution also gets slower too ...

I guess backwards compatibility is possible if owners are willing to accept an inferior experience compared to the last platform ...

No one can read into the future. But yes, I think investing in BC is critical. Now that the eShop is a thing, no BC basically means you're messing up with your customers. That's not acceptable anymore.
Sure, it would be impossible to sell MK8 again, and same is true for most of their games at this point. Their 'style' works fine with less powerful hardware, and for many people, the Switch games are just good enough. You won't sell a new machine to those guys with the only intent to sell the same games again wil slighltly better visuals. Not when they bought them on the eShop. Not when they already looks fine. Not when every other vendor has BC now. So if the choice is to give in 5% of perfs because they need the old X1 GPU to be part of the new design, of course it's a good trade-off. It's either that, or take the risk to alienate all your hardcore fans.

Now, if you can't sell MK8 again, what to do then? That's the tricky part. You obviously sell MK9, but you need the new game to be different enough. So you need time, a lot of time to let new ideas emerge.
My personal feeling is that no matter what Nintendo wants, the market does not need more than one Nintendo machine every 10 years or so. And this may even stretch to 12 or 13 (or more) once the move to HD/4K is done.
We all feel it. MK8 is an outstanding product, but it's basically the same game as MK Wii. Same for later 2D Mario. Obviously same for the shitload of remasters.
I think that as customers, we need a bit of a dry period before the need to (re)play some Nintendo game comes back.

But they aren't going to sell the same games again. The Switch was just an exception since most owners weren't able to play many of the games on Nintendo's previous system which was a failure ...

Where did you get the ~5% number from ? In my estimates it would be closer to ~15-20% as I would expect a minimalistic Orin SKU would need upto 6B transistors for a viable die. A Tegra X1 GPU probably consists of no less than 1B+ transistors. That's easily 15% of the die budget just spent on doing nothing but BC ...

So in the end, my expectations for Nintendo moving on:
- work hard to keep their existing gamers base happy by offering more of everything (BC obviously, better ecossytem, better online, more fan service)
- make sure their new machine is solid enough to stay relevant for 10 years or more so that third party support stays solid for as long as possible
- operate a smooth transition to the new machine with new Switch 1 games available for a few years more and only the bigger games exclusive to the Switch 2
- take time to invest on their major games (they already do that pretty well)
- continue to invest in porting their legacy games to the Switch platform (this includes Switch 2) to fill-in the blanks
- continue to print money with they non video game branches

Trying to force sell some new machine based on its advanced technical capabilities only is a path to failure. Nintendo investors may disagree though.

I think that your expectations of their new platform are going to come up really short ...

Staying relevant to third parties for a decade all the while achieving BC is basically trying to upsell a system based on advanced technical capabilities which is a concept you singled out. If developers are dissatisfied by working on the Series S then what makes you think that a weaker system will see more longevity ?

 

[Seanspeed]

Is it ? I reckon most of their big AAA projects will have already had anywhere between 2-4 years of mass production time when their new system launches which will give them a decent lineup schedule to launch successors in quick fashion for their biggest titles from their previous platform ...

The only way I can see them realistically exploring a software solution to BC is if they give up their handheld form factor designs. Do you believe that would be a better arrangement somehow ?

The vast majority of their third party support comes in the form of late ports and that didn't stop them from being the most successful platform of that generation despite not even featuring BC. Publishers will come around eventually if the new system itself is good which will mostly be contingent on the platform concept and it's first party games regardless ...

What if their only option their current partner is offering is a redundant 2nd GPU on die ? Are they supposed to somehow accept that deal when other vendors can potentially offer much better licensing costs, benefits and value if BC is not in play ?

Can AAA content developers even be a viable target for new upcoming software if the new system's budget is spent on BC when there's arguably a better chance of receiving new AAA content if the budget goes towards acquiring a more powerful platform. That can mean the difference between being solely stuck with last generation AAA third party content or potentially seeing current generation AAA third party content. Nearly all AAA 3rd party publishers including japanese publishers express discontent when developing for the Switch because they don't feel very productive on the platform ... (it's suboptimal for them to use deferred renderers on the system)

As I see it, they've got to consciously pick between having a portable form factor, a higher productivity system, or ultimately backwards compatibility. They can choose 2 out of the 3 but I imagine having the holy trinity itself is virtually off-limits in the near future ...

3rd party publishers aren't concerned with whether the system is 'good' or not, they care about if there's enough people who own the system to be worth making a port/fresh game. Ease of development is probably their next highest priority.

Sony and Xbox dont really have to deal with this chicken and egg problem because they basically get all the major 3rd party games by default, so they are basically guaranteed to have a decent lineup of games regardless of 1st party. Nintendo, through their own decision making, always makes things more difficult for themselves, where the 1st party has to do a lot of heavy lifting to get people to buy before 3rd parties see the platform as something really worth investing in. And getting a strong enough 1st party lineup these days, in one window, will be extremely difficult(especially after having blown their Zelda load already). Look at how much the Wii U struggled when relying mostly on 1st party software alone. It's not enough.

I just think BC is going to be incredibly necessary if Nintendo wants to see rapid adoption anywhere near like what the Switch got. I'm not gonna get into the technical arguments too heavy here cuz we're mostly just throwing around a lot of guesses on things we cant know much about, but I think even if Nintendo have to eat into hardware margins to a much more significant degree, it will be worth doing to provide the foundation for a more successful 7-8 years of the platform's life. It will pay off in the end. A floundering start to the generation that could potentially deter 3rd party support and whatnot will be far more harmful in the end. I know Nintendo like to make money on the hardware, but the software ecosystem is still where the biggest chunk of money comes from.

 

[Chiimaero]

But they aren't going to sell the same games again. The Switch was just an exception since most owners weren't able to play many of the games on Nintendo's previous system which was a failure ...

I hope so! But reality is there is little innovation in most of their games, and that's not a Switch exception at all. We play the same 2D mario, smash bros, animal crossing, donkey kong, pikmin, mario party, mario golf, mario tennis games since the N64/GC era. This is true for Zelda games as well up to SS. Same for metroid prime games.
The exceptions? The games with no successors (F-Zero...) and the "let's try anything because why not" Star Fox games...
This is a bit of a dark painting, because, of course the games come with a few new touches here or there. And despite all the similarities, we got a graphics update at each generations (well, except for the Wii generation...). And I love those games. But let's not pretend Nintendo does not resell us the same games. And I certainly expect it will be more difficult for them the next time as the Switch graphics are already nice enough for most people.
And just in case: this issue of rehashing the same games is worse for MS and Sony, it's not a Nintendo thing at all.

Where did you get the ~5% number from ? In my estimates it would be closer to ~15-20% as I would expect a minimalistic Orin SKU would need upto 6B transistors for a viable die. A Tegra X1 GPU probably consists of no less than 1B+ transistors. That's easily 15% of the die budget just spent on doing nothing but BC ...

I think that your expectations of their new platform are going to come up really short ...

Staying relevant to third parties for a decade all the while achieving BC is basically trying to upsell a system based on advanced technical capabilities which is a concept you singled out. If developers are dissatisfied by working on the Series S then what makes you think that a weaker system will see more longevity ?

I'd say we have zero basis to guesstimate the impact on the perfs. It can go from 0 to 20 indeed. You seem to imply the transistor count of T239 is a known value, and, AFAIK, it's not. My 5% figure is just a number acknowledging that it could be a measurable impact, nothing more. If Nintendo decides to have its SoC larger by 1B transistors, what's to prevent them to do so?

As for the other points, as I said, we can't see into the future. My view on the successes and failures over the past 30 years or so of Nintendo consoles gives me an opinion as to what I expect next, nothing more. And this opinion is the market does not need more than one Nintendo console per 10 years (or more), thus the need for the Switch 2 to be a bit more "future proofed" than the Switch 1.
Also on the longevity thing, we'll see. But as I said, I think it will be harder and harder for console vendors to sells new machines. We'll see how far the PS5 will go. But there are many many factors that may force the industry to slow down on the hardware side. Just because it's harder to make reasonable updates that are not too big, don't cost too much, on top of consuming too much power. This means developpers, happy or not, will have to support "old" hardware for longer periods of time. This is just the reality of increasingly more challenging new manufacturing process adding to the law of diminishing returns on the graphics side. Advances in GPU/graphics are slowing.

Nintendo knows better than us of course. If they decide to ditch BC, it's fine. If they partner with AMD/ARM/DMP/IBM, it's cool. If they come up with a new console 3 years after the Switch 2, well, so be it.
I just don't see a successful path for them in doing so.

 

[eastmen]

MS and Valve invested heavily in BC. Valve arguably more than even MS did. SteamOS and associated stuff has been a decade long effort.

I'm sure Nintendo could offer a BC solution with the power they should have in Switch 2. For me, I'd say it's a question as to whether they have the motivation and technical resources to implement it. Have Nvidia onboard could help with the latter part.

Sure but at the same time Nintendo should be investing in it also. They should have been investing in it for awhile now. Lets be real the switch isn't a powerhouse It's 4 cortex a57 cores at 1ghz , 4 gigs of memory and a tegra x1. It wasn't cutting edge then. Almost any decent mobile soc at this point in time should be able to emulate it just fine. Nintendo can go ahead an purchase an emulator team and give them jobs to do it. There are multiple switch emulators now.

What happened with the Series X/S along with the PS5 from AMD were exceptional circumstances that has very little chances of being replicated elsewhere. RDNA wasn't quite as clean of an architecture design as we would normally think of for GPUs. There were design compromises like having multiple SIMD modes (wave32 vs wave64), having two distinct geometry pipelines (legacy vs NGG), and it brought back all of the dropped instructions from GFX7 IP GPUs (last gen consoles) to make BC possible ...

I mean MS has games running from the xbox original that was intel and nvidia hardware , the xbox 360 which was IBM power pc and ati and then the xbox one which was full amd all running on the xbox series. They also had it all running on the xbox one.

It's explained in the above paragraph ...

Steam Deck is somewhat capable of emulating the system but as we see with the latest AAA release, there are still visual anomalies compared to the original system. A rule of thumb for software emulation is that as the solution gets more accurate (cycle accurate > LLE > HLE) the solution also gets slower too ...

How much is it that there are visual anomalies due to hardware differences vs the emulators are just a bunch of enthusiasts doing it on the side ? Aside from that there are still visual anomalies on the nintendo online emulated games. Seems like a few anomalies don't matter much to Nintendo.

I guess backwards compatibility is possible if owners are willing to accept an inferior experience compared to the last platform ...

Steam deck a system originally supposed to hit in the fall of 2021 but released winter of 2022

And this is just the begining , the emulators will get tweaked and continue to get better but 720p locked at 30fps

Remember the switch port goes also does 720p in handheld with a dynamic res and still has drops down to 20fps in some instances.


I mean are we expecting the switch 2 to be less performant vs the steam deck ? I think that would be a huge mistake on nintendo's part. 3rd party games were big deals on the original switch , if they come in lower than the deck its giong to have a lot of trouble playing those games. I'd be hoping from nintendo something like 2-3tflops. If they come in more powerful than the switch then Nintendo should surely be able to emulate the switch using a more powerful machine and optimizing for a single soc vs a bunch of hobbyists making a general emulator.

But they aren't going to sell the same games again. The Switch was just an exception since most owners weren't able to play many of the games on Nintendo's previous system which was a failure ...

I mean I am sure there are people who didn't buy the switch either right ? So we can keep using that logic to put out the same game.

 

[Lurkmass]

3rd party publishers aren't concerned with whether the system is 'good' or not, they care about if there's enough people who own the system to be worth making a port/fresh game. Ease of development is probably their next highest priority.

@Bold That's exactly what I meant by having a 'good' system ...

Sony and Xbox dont really have to deal with this chicken and egg problem because they basically get all the major 3rd party games by default, so they are basically guaranteed to have a decent lineup of games regardless of 1st party. Nintendo, through their own decision making, always makes things more difficult for themselves, where the 1st party has to do a lot of heavy lifting to get people to buy before 3rd parties see the platform as something really worth investing in. And getting a strong enough 1st party lineup these days, in one window, will be extremely difficult(especially after having blown their Zelda load already). Look at how much the Wii U struggled when relying mostly on 1st party software alone. It's not enough.

Sony and Microsoft console platforms compete on the basis of developer productivity hence why nearly all of the third parties congregate over there because they feel very productive ...

Nintendo's last console platform is an example where valuable games (we see this clearly from re-releases on Switch) were locked down to a bad system so we shouldn't even be questioning the strength of their 1st party software at all. As long as their new system is appealing (whether it has BC or not), a solid lineup schedule will do the rest of the work. This is why 1st party AAA game releases are drying up on the Switch so they can prepare for their next wave of AAA games for their new system ... (they have plenty of tentpole options besides just Zelda)

I'd say we have zero basis to guesstimate the impact on the perfs. It can go from 0 to 20 indeed. You seem to imply the transistor count of T239 is a known value, and, AFAIK, it's not. My 5% figure is just a number acknowledging that it could be a measurable impact, nothing more. If Nintendo decides to have its SoC larger by 1B transistors, what's to prevent them to do so?

Considering where they've spent nearly half of the die space on functionality other than graphics for their current SoC, I imagine that it was a fair estimate for what they have left for their own budget or you could even say that it's slightly more generous given that they now have ~6x as much logic to play around with ...

As for the other points, as I said, we can't see into the future. My view on the successes and failures over the past 30 years or so of Nintendo consoles gives me an opinion as to what I expect next, nothing more. And this opinion is the market does not need more than one Nintendo console per 10 years (or more), thus the need for the Switch 2 to be a bit more "future proofed" than the Switch 1.
Also on the longevity thing, we'll see. But as I said, I think it will be harder and harder for console vendors to sells new machines. We'll see how far the PS5 will go. But there are many many factors that may force the industry to slow down on the hardware side. Just because it's harder to make reasonable updates that are not too big, don't cost too much, on top of consuming too much power. This means developpers, happy or not, will have to support "old" hardware for longer periods of time. This is just the reality of increasingly more challenging new manufacturing process adding to the law of diminishing returns on the graphics side. Advances in GPU/graphics are slowing.

Nintendo knows better than us of course. If they decide to ditch BC, it's fine. If they partner with AMD/ARM/DMP/IBM, it's cool. If they come up with a new console 3 years after the Switch 2, well, so be it.
I just don't see a successful path for them in doing so.

Just because hardware improvement is slowing down doesn't mean the demand for greater productivity is. Other console manufacturers will just introduce upgraded mid generation refreshes so that PS5 becomes the absolute minimum shipping target while Microsoft may have no choice but to end production of their Series S earlier than planned ...

The chances of Nintendo seeing longevity in terms of credible third party support regardless get smaller each passing year. Even the likes of Steam Deck can't keep up anymore in recent games. If Nintendo does go down the BC angle they're way more likely to dial down in third party support as we see on their previous systems ...

Their current platform was more successful in third party support specifically because they were willing to forgo backwards compatibility to gain a more solid foundation in terms of productivity. They could have a more meaningful improvement in hardware capability if they don't have to focus on backwards compatibility ...

I mean MS has games running from the xbox original that was intel and nvidia hardware , the xbox 360 which was IBM power pc and ati and then the xbox one which was full amd all running on the xbox series. They also had it all running on the xbox one.

The more recent Xbox consoles don't have full compatibility with either the original Xbox or the 360 ...

How much is it that there are visual anomalies due to hardware differences vs the emulators are just a bunch of enthusiasts doing it on the side ? Aside from that there are still visual anomalies on the nintendo online emulated games. Seems like a few anomalies don't matter much to Nintendo.

Some of those reverse engineering hobbyists actually got hired by the likes of IHVs such as AMD/Nvidia or even console manufacturers like Sony themselves! So I don't know why you're even questioning their expertise in this area ...

Steam deck a system originally supposed to hit in the fall of 2021 but released winter of 2022
*snip*

And this is just the begining , the emulators will get tweaked and continue to get better but 720p locked at 30fps
*snip*
Remember the switch port goes also does 720p in handheld with a dynamic res and still has drops down to 20fps in some instances.


I mean are we expecting the switch 2 to be less performant vs the steam deck ? I think that would be a huge mistake on nintendo's part. 3rd party games were big deals on the original switch , if they come in lower than the deck its giong to have a lot of trouble playing those games. I'd be hoping from nintendo something like 2-3tflops. If they come in more powerful than the switch then Nintendo should surely be able to emulate the switch using a more powerful machine and optimizing for a single soc vs a bunch of hobbyists making a general emulator.

Once again the rule of thumb is that as a solution gets more accurate, it becomes slower. Steam Deck runs with tons of hacks to increase perf to make software emulation of modern consoles a possibility but it makes for a very high maintenance project which isn't ideal in terms of long-term support ...

I don't think Nintendo will fancy the idea of software BC for a modern system for three reasons below ...

The first being tons of engineering resources that could be better spent on optimizing their own upcoming projects or providing better third party developer relations technical support ...
The second reason is the potential amount of maintenance because of the introduction of said hacks they'd have to do to support both high compatibility (no regressions/bugs) and high perf (high framerate) of the said solution ...
The final reason being potential obsolescence of the said solution when a generation comes to an end. It's a lot of engineering work literally *wasted* if Nintendo decides that they want to change architectures again for their next system after the Switch's successor ...

If you look at Nintendo's virtual console (software emulation) efforts they like to prefer much older platforms since they don't have to spend a lot of time to get a workable solution when they have much more powerful hardware to brute force any slow downs that they'd see on weaker platforms ...

Also if Nintendo is going to attempt emulation for modern systems like the Switch what would be their justification for not doing the same or increasing the scope to include other weaker systems but still would require a highly complex solutions for them (GC/Wii, 3DS, WiiU and potentially even older rival systems like DC, PS2, PSP, Xbox) ? That would be a tons of resources spent on a huge multi-faceted project that might not even last for more than a generation!

I mean I am sure there are people who didn't buy the switch either right ? So we can keep using that logic to put out the same game.

I don't understand what the topic behind this statement is supposed to be about ...

 

[eastmen]

@Bold

The more recent Xbox consoles don't have full compatibility with either the original Xbox or the 360 ...

That is a license issue not a hardware compatibility issue

Some of those reverse engineering hobbyists actually got hired by the likes of IHVs such as AMD/Nvidia or even console manufacturers like Sony themselves! So I don't know why you're even questioning their expertise in this area ...

They did and guess what Nintendo can go out and hire their own.

Once again the rule of thumb is that as a solution gets more accurate, it becomes slower. Steam Deck runs with tons of hacks to increase perf to make software emulation of modern consoles a possibility but it makes for a very high maintenance project which isn't ideal in terms of long-term support ...

And yet it runs well enough that everyone is enjoying it. On my steam deck I am getting a better experiance performance wise than on the switch itself running native. I get a higher resolution and more fps

I don't think Nintendo will fancy the idea of software BC for a modern system for three reasons below ...

The first being tons of engineering resources that could be better spent on optimizing their own upcoming projects or providing better third party developer relations technical support ...
The second reason is the potential amount of maintenance because of the introduction of said hacks they'd have to do to support both high compatibility (no regressions/bugs) and high perf (high framerate) of the said solution ...
The final reason being potential obsolescence of the said solution when a generation comes to an end. It's a lot of engineering work literally *wasted* if Nintendo decides that they want to change architectures again for their next system after the Switch's successor ...

If you look at Nintendo's virtual console (software emulation) efforts they like to prefer much older platforms since they don't have to spend a lot of time to get a workable solution when they have much more powerful hardware to brute force any slow downs that they'd see on weaker platforms ...

Also if Nintendo is going to attempt emulation for modern systems like the Switch what would be their justification for not doing the same or increasing the scope to include other weaker systems but still would require a highly complex solutions for them (GC/Wii, 3DS, WiiU and potentially even older rival systems like DC, PS2, PSP, Xbox) ? That would be a tons of resources spent on a huge multi-faceted project that might not even last for more than a generation!

I don't understand what the topic behind this statement is supposed to be about ...

Then that is a huge failing on nintendos part. There are people with a lot of purchased games on the switch that will contemplate moving over to other consoles or hardware. IT is very compelling to be able to buy a game on xbox and play it forever on all future xboxs. Its compelling to buy a game on steam and play it on all future hardware capable of playing steam.

I've already had a few die hard nintendo fans (people who only have a switch) reach out to me about steam deck and the ally. One of them preordered the ally and has purchased 3 years of game pass for it

 

[Lurkmass]

That is a license issue not a hardware compatibility issue

Even ignoring licensing issues for content, there are still massive coverage gaps (633 out of 2154 total) for 360 titles and it gets even worse with the original Xbox (63 out of 996!) ...

They did and guess what Nintendo can go out and hire their own.

Emulation developers/contributors can absolutely seek out to make a gainful career out of reverse engineering but quite a few of them do it purely out of passion rather than financial reasons. Also if Nintendo does decide to hire these people, I imagine one of their conditions would be is that they cannot work on or contribute to competing reverse engineering projects (emulation) much like Nvidia or Sony outlined for any prospective employees ...

Not being able to actively to participate with the open source emulation community is a sticking point for some of these experts ...

And yet it runs well enough that everyone is enjoying it. On my steam deck I am getting a better experiance performance wise than on the switch itself running native. I get a higher resolution and more fps

If you look at their hardware requirements, they recommend a GTX 1660 or RX 5500. Steam Deck for all intents and purposes is the bare minimum as they warn low perf of integrated graphics solutions in general ... (which probably leads tof needing lower compatibility settings)

You see game specific workaround in the form of game mods all the time if you're doing modern console emulation which defeats the purpose of the concept (emulation) ...

Then that is a huge failing on nintendos part. There are people with a lot of purchased games on the switch that will contemplate moving over to other consoles or hardware. IT is very compelling to be able to buy a game on xbox and play it forever on all future xboxs. Its compelling to buy a game on steam and play it on all future hardware capable of playing steam.

I've already had a few die hard nintendo fans (people who only have a switch) reach out to me about steam deck and the ally. One of them preordered the ally and has purchased 3 years of game pass for it

Well Nintendo are just going to have to hope for a hail mary that Nvidia includes a 2nd GPU, ditch portable form factors or no BC ...

There's just too much complexity involved with modern emulation for low power systems for that to be an acceptable proposal for them. Also how do you justify Switch specifically when there are much easier lower hanging fruits that can be more easily picked besides their current platform ? Going from N64 to their current console as their emulation target without significantly increasing the complexity of that project would require a bigger leap in capability than the Steam Deck ...

Switch emulation would be a forsaken project with high implementation costs, high maintenance, and may not last beyond the duration of their next system ...

 

[eastmen]

Even ignoring licensing issues for content, there are still massive coverage gaps (633 out of 2154 total) for 360 titles and it gets even worse with the original Xbox (63 out of 996!) ...

You are speculating that it isn't licensing issues.

Emulation developers/contributors can absolutely seek out to make a gainful career out of reverse engineering but quite a few of them do it purely out of passion rather than financial reasons. Also if Nintendo does decide to hire these people, I imagine one of their conditions would be is that they cannot work on or contribute to competing reverse engineering projects (emulation) much like Nvidia or Sony outlined for any prospective employees ...

maybe but Nintendo can hire its own engineers or did the virtual console games just appear one day and nintendo said hey lets not pass up this gift from god and use them ?

Not being able to actively to participate with the open source emulation community is a sticking point for some of these experts ...

If you look at their hardware requirements, they recommend a GTX 1660 or RX 5500. Steam Deck for all intents and purposes is the bare minimum as they warn low perf of integrated graphics solutions in general ... (which probably leads tof needing lower compatibility settings)

Yes and I would be hopeful a 2024 or 2025 console from nintendo would beat those minimums and the hardware inside the deck. If not then Nintendo and nvidia are doing something terribly wrong. Also Nintendo isn't affraid of using resampling tech like the Fidelity FX in zelda. They shouldn't have an issue using any of nvidia's new technology

You see game specific workaround in the form of game mods all the time if you're doing modern console emulation which defeats the purpose of the concept (emulation) ...

Well Nintendo are just going to have to hope for a hail mary that Nvidia includes a 2nd GPU, ditch portable form factors or no BC ...

There's just too much complexity involved with modern emulation for low power systems for that to be an acceptable proposal for them. Also how do you justify Switch specifically when there are much easier lower hanging fruits that can be more easily picked besides their current platform ? Going from N64 to their current console as their emulation target without significantly increasing the complexity of that project would require a bigger leap in capability than the Steam Deck ...

Switch emulation would be a forsaken project with high implementation costs, high maintenance, and may not last beyond the duration of their next system ...

How would it be possible that nvidia can't make hardware compatible with their older hardware. If its true then Nintendo went with the wrong partner for their hardware.

 

[Lurkmass]

You are speculating that it isn't licensing issues.

Licensing issues only stops them from being able to sell obsolete games. It doesn't stop them from being able to improve their emulators to play the original copies of these said games ...

There's no valid excuse for these abysmal compatibility rates unless they were strictly technical in nature ...

maybe but Nintendo can hire its own engineers or did the virtual console games just appear one day and nintendo said hey lets not pass up this gift from god and use them ?

I never insinuated Nintendo can't hire their own engineers for this purpose ...

If they think they can actually make a far better solution like you imply then they can by all means try but they're going to come up shorter than you think. Such is a life of compromises ...

would be hopeful a 2024 or 2025 console from nintendo would beat those minimums and the hardware inside the deck. If not then Nintendo and nvidia are doing something terribly wrong. Also Nintendo isn't affraid of using resampling tech like the Fidelity FX in zelda. They shouldn't have an issue using any of nvidia's new technology

You can't just barely beat these minimums. To get a simple, high compatibility/accuracy and maintainable implementation they need to have a reasonable amount of headroom (at least 2x the perf of Steam Deck) to make it a worthwhile endeavor ...

Temporal resampling technology isn't going to help either because rendering original games at a lower resolution during emulation just to use temporal upsampling isn't going to work when the original games aren't guaranteed to have motion vector data so that just leaves them with spatial resampling techniques ...

How would it be possible that nvidia can't make hardware compatible with their older hardware. If its true then Nintendo went with the wrong partner for their hardware.

Long-term planning and vision isn't exactly either of these businesses strongest points. Nvidia are compelled by market force to do clean sheet architecture designs hence why they invest hard in their software ecosystem to keep it portable across their architectures. Binary compatibility isn't as much of a priority for Nvidia as much as maintaining a high percentage of dGPU market share even if it mean killing binary compatibility for future designs to achieve that objective ...

Nintendo sees hardware as nothing more than a proxy for their 1st party games so they were likely just shopping around for something that met their requirements and it just happened to be from Nvidia. Are they truly the 'wrong' partner if everything turned out well for them in the end ?

Not every action has to have a grand plan for the future. Sometimes entities act spontaneously according to their current interests. Microsoft and Sony had to make a bet that AMD would survive until the next generation came along so their plans for BC would've been ruined if AMD went bankrupt. Nintendo wasn't concerned at the time whether or not if they were going to work with Nvidia again for their next system. Nintendo wanted to make their new system as successful as possible so if the Switch ended up being a failure, they were probably just going to call it quits for console platforms at that point hence any planning for BC would've been irrelevant in that case ...

If Nintendo could have a crystal ball to gaze into alternate futures where the Switch would've been successful regardless of supplier then they very well could have optimized their initial move to include plans for BC or if they discovered no such future and their success was absolutely contingent on going with Nvidia then they arguably made the right move in the end by not concerning themselves with BC because the choice between immediately ending up with another failed platform or having future BC isn't hard. Everyone just has to live by their own choices they've made even if they were mistakes in hindsight and that includes Nintendo and Nvidia ...

Nintendo knew well what they were getting into when heading into a licensing deal with Nvidia while Microsoft and Sony were gambling with their very own future on AMD (share prices and market valuation was at an all time low for them). Well whatever regrets they have now about their past decision, they all at least live to see another day. If Nintendo made it out this far for so long, I don't think the possibility of losing BC will deter from releasing a new platform ...