• 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)

Reader mode
TheMindOfThomas said:
The tech is pretty much there, it's the software that isn't.
Click to expand...
Exactly this! Though instead of developing games that are only playable in VR, I think just adding good implementation to normal games would usually be a better option, but I definitely wouldn't be opposed to both coexisting.
Shoulder said:
Chiming in with this VR talk, I got to thinking about my childhood memories of using a Sony Walkman tape player.

[collapsed for cleanliness]
Click to expand...
I would love this, too. It'd fit perfectly with the Switch branding.
 
Seboss said:
I'm not really expecting them to do it though. How likely is it that Nvidia has managed to fit an X1 in the larger Switch 2?
Click to expand...
If they wanted to, they could probably do it. In the age of Proton and Rosetta, that is no longer a good solution to BC though.

Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
 
Seboss said:
I mentioned it in the past, that would be the most Nintendo thing ever to somehow downclock Drake to X1 level of performance in BC mode 😂

I'm not really expecting them to do it though. How likely is it that Nvidia has managed to fit an X1 in the larger Switch 2?
Click to expand...

The tricky thing for Nintendo is that getting exactly TX1 levels of performance isn't really going to be possible. Back on the Wii, when playing Gamecube games Nintendo would clock the CPU and GPU down to the same clocks as the Gamecube ran at, and likewise with Wii games on the Wii U. Because Nintendo reused the same hardware architectures across generations, they were able to guarantee that games would run precisely the same as on the older hardware*. There's a good reason to do this, as it's possible that some games would have unexpected bugs when running at higher clock speeds or on more powerful hardware, so running at the same clocks speeds on functionally identical hardware means not having to test hundreds of games to see if any of them break in strange ways.

This approach to BC really hobbled Nintendo with the Wii U, and thankfully they've dropped it for Switch 2, where instead of using the exact same A57 CPU cores and Maxwell GPU architecture as before, they're using newer iterations with (presumably) A78 cores and an Ampere GPU. On the CPU side, the A78 will be able to directly run code compiled for the A57 cores in the original Switch, but because it's a different architecture, it's not going to run it at the same speed. Even clocked down to the same 1GHz, the A78 is going to run almost all code faster, and may even run some code slower, if it was very tightly optimised for the A57.

On the GPU side, it's no longer binary compatible with shaders compiled for Maxwell, so there will have to be a compatibility layer which converts compiled shaders from Maxwell to Ampere. This will again mean potential performance differences compared to the original Switch, even at identical clock speed, with some code running faster or slower depending on architectural differences (and how good the compatibility layer is at generating performant Ampere shader code). Because Nintendo can't guarantee identical performance to the original Switch, they will likely have to err on the side of higher performance, to ensure that, even in the worst case scenarios, performance is at least as good as the original hardware.

One side effect of this change in approach is that they have to do a lot more testing. In fact, I suspect that this is the reason news is coming out about BC now. Nintendo needs to test as many games as possible under BC, and they're not going to do that all in-house. My guess is that Nintendo have been testing their BC solution with their first-partly titles for some time, and have recently become happy enough with the performance and compatibility to expand out to third parties, asking them to test their games and provide feedback. Hence Nate's comment that "testing is underway". I would assume the main focus will be on the GPU compatibility layer, catching edge cases not present in Nintendo's games, but they'll also be on the lookout for any performance issues.

My guess is that we'll get slightly improved performance in BC mode, with games hitting their framerate caps a bit more often and running at slightly higher resolutions in cases where dynamic resolution is present, but I don't expect major performance improvements. One important factor will be battery life, so I'd expect Nintendo to choose clocks in BC mode that are just high enough to guarantee as good or better performance than the original Switch and leave it at that. I actually wouldn't be surprised to see Nintendo use an increase in battery life on Switch 1 games as part of their pitch for people to upgrade, so there's an incentive to lean more towards modest clocks and better battery life than trying to squeeze every bit of performance they can out of previous-gen games.


* I suspect there are probably some very small performance differences due to smaller things like changes to the cache, but it's clear than Nintendo were trying to achieve as close to identical performance as possible.
 
Thraktor said:
My guess is that we'll get slightly improved performance in BC mode, with games hitting their framerate caps a bit more often and running at slightly higher resolutions in cases where dynamic resolution is present, but I don't expect major performance improvements. One important factor will be battery life, so I'd expect Nintendo to choose clocks in BC mode that are just high enough to guarantee as good or better performance than the original Switch and leave it at that. I actually wouldn't be surprised to see Nintendo use an increase in battery life on Switch 1 games as part of their pitch for people to upgrade, so there's an incentive to lean more towards modest clocks and better battery life than trying to squeeze every bit of performance they can out of previous-gen games.


* I suspect there are probably some very small performance differences due to smaller things like changes to the cache, but it's clear than Nintendo were trying to achieve as close to identical performance as possible.
Click to expand...

Since, as you have documented extensively there isn't really much battery life gain in clocking it below a certain threshold (420mhz?), do you believe they will disable hardware in compatibility mode?

Lets say BC mode runs 4SM, with 8 disabled.
And 4 A78s.
 
Ojoloco said:
Imagine to testing all of the shovelware
Click to expand...
There is so much shovelware on the eshop, I feel like it would be impossible to test all of those games tbh. I wonder if Nintendo will only allow BC on games that have passed some type of Switch 2 performance certification.
 
Anyone ever seen MVG video where he raised the clock speeds of a regular jail broken Switch and was able to achieve better/more stable frame rates in TotK?

That’s what I am imagining for BC upgrades for the Switch 2; games just hitting their intended frame rates smoothly where maybe they lacked and less low dynamic resolutions.
 
TheSentry said:
Anyone ever seen MVG video where he raised the clock speeds of a regular jail broken Switch and was able to achieve better/more stable frame rates in TotK?

That’s what I am imagining for BC upgrades for the Switch 2; games just hitting their intended frame rates smoothly where maybe they lacked and less low dynamic resolutions.
Click to expand...
For totk/ botw specifically, memory bandwidth was the main bottleneck if I recall. He slightly inchreased the memory clock, and removed almost all drops if I recall correctly.
 
Thraktor said:
My guess is that we'll get slightly improved performance in BC mode, with games hitting their framerate caps a bit more often and running at slightly higher resolutions in cases where dynamic resolution is present, but I don't expect major performance improvements. One important factor will be battery life, so I'd expect Nintendo to choose clocks in BC mode that are just high enough to guarantee as good or better performance than the original Switch and leave it at that. I actually wouldn't be surprised to see Nintendo use an increase in battery life on Switch 1 games as part of their pitch for people to upgrade, so there's an incentive to lean more towards modest clocks and better battery life than trying to squeeze every bit of performance they can out of previous-gen games.
Click to expand...
Interesting perspective. Personally, when considering what would be a good pitch for people to upgrade, I would peg performance in the form of higher resolutions (for DR games) and more stable frame rates as a bigger feature than additional battery.

But there might be a happy medium where most if not all games can hit (close to) their upper end resolutions and frame rate target while still not running with full power (and thereby saving on battery as well). Battery on the order of 6-7 hours would be quite enough I think (in the ballpark of Switch V2).

But yeah, just my two cents. Definitely up for discussion.
 
Thraktor said:
The tricky thing for Nintendo is that getting exactly TX1 levels of performance isn't really going to be possible. Back on the Wii, when playing Gamecube games Nintendo would clock the CPU and GPU down to the same clocks as the Gamecube ran at, and likewise with Wii games on the Wii U. Because Nintendo reused the same hardware architectures across generations, they were able to guarantee that games would run precisely the same as on the older hardware*. There's a good reason to do this, as it's possible that some games would have unexpected bugs when running at higher clock speeds or on more powerful hardware, so running at the same clocks speeds on functionally identical hardware means not having to test hundreds of games to see if any of them break in strange ways.

This approach to BC really hobbled Nintendo with the Wii U, and thankfully they've dropped it for Switch 2, where instead of using the exact same A57 CPU cores and Maxwell GPU architecture as before, they're using newer iterations with (presumably) A78 cores and an Ampere GPU. On the CPU side, the A78 will be able to directly run code compiled for the A57 cores in the original Switch, but because it's a different architecture, it's not going to run it at the same speed. Even clocked down to the same 1GHz, the A78 is going to run almost all code faster, and may even run some code slower, if it was very tightly optimised for the A57.

On the GPU side, it's no longer binary compatible with shaders compiled for Maxwell, so there will have to be a compatibility layer which converts compiled shaders from Maxwell to Ampere. This will again mean potential performance differences compared to the original Switch, even at identical clock speed, with some code running faster or slower depending on architectural differences (and how good the compatibility layer is at generating performant Ampere shader code). Because Nintendo can't guarantee identical performance to the original Switch, they will likely have to err on the side of higher performance, to ensure that, even in the worst case scenarios, performance is at least as good as the original hardware.

One side effect of this change in approach is that they have to do a lot more testing. In fact, I suspect that this is the reason news is coming out about BC now. Nintendo needs to test as many games as possible under BC, and they're not going to do that all in-house. My guess is that Nintendo have been testing their BC solution with their first-partly titles for some time, and have recently become happy enough with the performance and compatibility to expand out to third parties, asking them to test their games and provide feedback. Hence Nate's comment that "testing is underway". I would assume the main focus will be on the GPU compatibility layer, catching edge cases not present in Nintendo's games, but they'll also be on the lookout for any performance issues.

My guess is that we'll get slightly improved performance in BC mode, with games hitting their framerate caps a bit more often and running at slightly higher resolutions in cases where dynamic resolution is present, but I don't expect major performance improvements. One important factor will be battery life, so I'd expect Nintendo to choose clocks in BC mode that are just high enough to guarantee as good or better performance than the original Switch and leave it at that. I actually wouldn't be surprised to see Nintendo use an increase in battery life on Switch 1 games as part of their pitch for people to upgrade, so there's an incentive to lean more towards modest clocks and better battery life than trying to squeeze every bit of performance they can out of previous-gen games.


* I suspect there are probably some very small performance differences due to smaller things like changes to the cache, but it's clear than Nintendo were trying to achieve as close to identical performance as possible.
Click to expand...
why not have a low power mode for that if it gives modest performance and relation boost? I would rather have a large performance boost knowing that games can run at max resolution and frame rate and if I want better battery life I can choose low power mode
 
0
So far we have talked a lot about patches on a case-by-case basis to improve performance. But what I would like to understand is whether it was theoretically possible to implement a universal software solution not to improve games but simply to ensure that backward compatibility is systematically functional?

I mean, is it technically possible to do an automatic translation of instructions for any switch 1 game, or do the developers already have to intervene individually to adjust the compatibility?

My idea has always been that backwards compatibility would go through the game library linked to your account. If you have a game in your library, buy in digital or already launched in physical with your account, then a patch downloads automatically when you want to launch the game on switch 2.

But does translating the instructions automatically mean you don’t need to touch the games at all, much like an emulator? Is there potentially a stable environment that could work for everything?
 
If I remember correctly, in the internal data of the Arkham Knight port on Switch there is a profile of the game at 1080p that is not used on Switch and does not work on this one, with some graphical improvements or things like that. Perhaps it's the kind of patch you should expect from a Switch to Switch 2 game without making a native version
 
boodream said:
So far we have talked a lot about patches on a case-by-case basis to improve performance. But what I would like to understand is whether it was theoretically possible to implement a universal software solution not to improve games but simply to ensure that backward compatibility is systematically functional?

I mean, is it technically possible to do an automatic translation of instructions for any switch 1 game, or do the developers already have to intervene individually to adjust the compatibility?

My idea has always been that backwards compatibility would go through the game library linked to your account. If you have a game in your library, buy in digital or already launched in physical with your account, then a patch downloads automatically when you want to launch the game on switch 2.

But does translating the instructions automatically mean you don’t need to touch the games at all, much like an emulator? Is there potentially a stable environment that could work for everything?
Click to expand...
the expectation is that the automatic translation is the default because of the change in gpu architecture. the environment is built into the system and devs don't have to do anything unless they want to break out of the switch constraints
 
Charamiwa said:
How do we think they would handle 720p Switch 1 games on a 1080p screen?
Click to expand...
They will probably do a combination of the following things:
1. Run them as is with some basic upscaling
2. Offer a 1:1 pixel mode like they did for DS games on the 3DS
3. Patch them to run at native 1080p (would require every individual game to be patched)
 
Ojoloco said:
If I remember correctly, in the internal data of the Arkham Knight port on Switch there is a profile of the game at 1080p that is not used on Switch and does not work on this one, with some graphical improvements or things like that. Perhaps it's the kind of patch you should expect from a Switch to Switch 2 game without making a native version
Click to expand...
I missed that, would appreciate a source.
 
boodream said:
So far we have talked a lot about patches on a case-by-case basis to improve performance. But what I would like to understand is whether it was theoretically possible to implement a universal software solution not to improve games but simply to ensure that backward compatibility is systematically functional?

I mean, is it technically possible to do an automatic translation of instructions for any switch 1 game, or do the developers already have to intervene individually to adjust the compatibility?

My idea has always been that backwards compatibility would go through the game library linked to your account. If you have a game in your library, buy in digital or already launched in physical with your account, then a patch downloads automatically when you want to launch the game on switch 2.

But does translating the instructions automatically mean you don’t need to touch the games at all, much like an emulator? Is there potentially a stable environment that could work for everything?
Click to expand...
Yes, in principle this is possible. The compatibility layer needs to translate on the fly from one binary language to another, but considering that most GPU commands are vectorised commands, the translation is probably not the most time-consuming if done in a smart way. A more pressing issue is whether the translation is consistently optimal or whether performance leaks out during the remapping of one binary language to another (Thraktor explained it much better up above). But I imagine NVIDIA are significantly involved in this, and they know both binary instruction sets through and through, so I expect it to work well. There can be individual games that have some quirky code that relies on the specifics of the Switch system and don't work (fully) out of the box, but most games should work in this manner in my view.

Having to re-download your old game for BC sounds to me like a sub-optimal scenario, definitely when you want to play them from the physical card.
 
How do some of you think this patch system will work and is it possible we could see Game Freak do it for this years Pokemon game? I had the Pokemon game penciled in as a cross gen game.
 
Zedark said:
Yes, in principle this is possible. The compatibility layer needs to translate on the fly from one binary language to another, but considering that most GPU commands are vectorised commands, the translation is probably not the most time-consuming if done in a smart way. A more pressing issue is whether the translation is consistently optimal or whether performance leaks out during the remapping of one binary language to another (Thraktor explained it much better up above). But I imagine NVIDIA are significantly involved in this, and they know both binary instruction sets through and through, so I expect it to work well. There can be individual games that have some quirky code that relies on the specifics of the Switch system and don't work (fully) out of the box, but most games should work in this manner in my view.

Having to re-download your old game for BC sounds to me like a sub-optimal scenario, definitely when you want to play them from the physical card.
Click to expand...
Does it even matter if the remapping isn't 100% optimal, when it's a much more powerful soc?
 
Hermii said:
Does it even matter if the remapping isn't 100% optimal, when it's a much more powerful soc?
Click to expand...
It will never be 100% optimal, but the worry would be that some chunks might be like 10% optimal, and then you could get degraded, rather than upgraded, performance. But I don't think we need to worry about it too much. However, for Thraktor's assertion, it is an important question how much raw horsepower you need and want to employ to balance between adding more battery life and more performance.
 
Dark Cloud said:
How do some of you think this patch system will work and is it possible we could see Game Freak do it for this years Pokemon game? I had the Pokemon game penciled in as a cross gen game.
Click to expand...
I think making a native version should be fairly straightforward/ small learning curve.

It will use an iteration of the same dev environment they're already used to. So as long as you're not trying anything ambitious it should be straight forward to get the game running imo. Nobody's forcing you to use anything beyond Maxwells feature set, and you should get a nice boost even if you just brute force it.
 
As long as the compatibility layer isn't introducing bugs, I do not think the increases performance from T239 is likely to create them. We have numerous videos showing lots of various Switch games running with massive overclocks and I do not remember seeing this ever break a game of introduce bugs. Switch games run close to the metal, bit not to the extent that older systems did where gamespeed was directly linked to clock speeds.

If Nintendo and Nvidia get this right, it can be a big selling point and help encourage Switch owners to upgrade sooner rather than later.
 
boodream said:
So far we have talked a lot about patches on a case-by-case basis to improve performance. But what I would like to understand is whether it was theoretically possible to implement a universal software solution not to improve games but simply to ensure that backward compatibility is systematically functional?

I mean, is it technically possible to do an automatic translation of instructions for any switch 1 game, or do the developers already have to intervene individually to adjust the compatibility?

My idea has always been that backwards compatibility would go through the game library linked to your account. If you have a game in your library, buy in digital or already launched in physical with your account, then a patch downloads automatically when you want to launch the game on switch 2.

But does translating the instructions automatically mean you don’t need to touch the games at all, much like an emulator? Is there potentially a stable environment that could work for everything?
Click to expand...

That what it sounds like we’re getting. It will be something built into the Switch 2 that will try to translate the game to run on the new hardware. No way we are gonna have to patch individual games since developers aren’t going to go back and patch their games in huge numbers.

Much like emulators though it won’t be perfect for all games. There will be some per game adjustments but it will be done at the system level and not a patch to download per game. Some games will just not work and not worth patching. Kinda of like PS5 not supporting all PS4 games.
 
I'm glad to see we have some huge signs that Switch 2 is backwards compatible and not in some weird half BC way like a lot of people thought for some very strange reason.
 
Goodtwin said:
As long as the compatibility layer isn't introducing bugs, I do not think the increases performance from T239 is likely to create them. We have numerous videos showing lots of various Switch games running with massive overclocks and I do not remember seeing this ever break a game of introduce bugs. Switch games run close to the metal, bit not to the extent that older systems did where gamespeed was directly linked to clock speeds.
Click to expand...
That's where Nintendo has to held themselves to much higher standards than the homebrew community. If a game crashes because of an OC no one in the community would think of that as a big deal.
 
fantomena said:
Why use words like these? Why call it a piece of shit?
Click to expand...
I love the Steam Deck. It's a crass jokey dig at its size, not a scathing indictment of the whole unit.

Like, I use the word 'honking' there. That alone should tell you I'm being unserious.
 
Last edited:
Dimensio said:
They will probably do a combination of the following things:
1. Run them as is with some basic upscaling
2. Offer a 1:1 pixel mode like they did for DS games on the 3DS
3. Patch them to run at native 1080p (would require every individual game to be patched)
Click to expand...
Oh yeah 1:1 pixel mode sounds good to me, if the screen is that much bigger I wouldn't mind using less of it for Switch 1 games.
 
Thraktor said:
My guess is that we'll get slightly improved performance in BC mode, with games hitting their framerate caps a bit more often and running at slightly higher resolutions in cases where dynamic resolution is present, but I don't expect major performance improvements. One important factor will be battery life, so I'd expect Nintendo to choose clocks in BC mode that are just high enough to guarantee as good or better performance than the original Switch and leave it at that. I actually wouldn't be surprised to see Nintendo use an increase in battery life on Switch 1 games as part of their pitch for people to upgrade, so there's an incentive to lean more towards modest clocks and better battery life than trying to squeeze every bit of performance they can out of previous-gen games.
Click to expand...
As you said, they have to pick a clock speed that guarantees consistent performance in all games, and since they can't test every game, they likely have to overshoot, so I can see most (or all?) games maxing out their previous targets, not just occasionally. That's not a major improvement when games are falling just short of their targets at worst; even if it's a game with poor performance that falls short consistently, any given frame is not far off from making it.

But I'm not so sure that they're going to focus on battery life in this regard anyway. "As good or better" is a believable talking point for both battery life and performance in BC, but significantly better battery life in BC isn't the most important selling point, and moreover it would kind of detract from the perception of battery life in native titles if they say them side by side. It's the kind of thing you might find on a support page, not in marketing material.
 
Last edited:
Makes me wonder if we will need to eventually create a Fami thread to keep a running list of Switch 1 titles that was determined to be problematic with Switch 2 BC layer (crashing, stuttering, whatever). Hopefully no games end up on that list or is going to be very few.

Such a list would be helpful to me considering the size of my digital library. If none of the entries in my digital library have an issue, I can then start looking into selling away the orig Switch.
 
Reader mode
You must log in or register to reply here.
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