Based on how tech has progressed since the Switch, how powerful could a Switch Pro in early 2021 actually be?

Dakhil · Oct 21, 2020

bmfrosty said:
I hope it will use NVMe for external storage whether the physical format is m.2, cfexpress, xfmexpress, or microsdexpress.

I think pro could be an abca form factor instead of a faster unit - that being a metal body, glass screen, and maybe different joycons that use full size analog sticks like a pro controller. Otherwise it will use the same clocks as all of the other switch models of it's generation.

As much as I would like the next Nintendo Switch model releasing in 2021 to use CFExpress Type A cards, SD Express 8.0 cards, etc., for external storage, unfortunately, I think Nintendo's going to continue to use UHS-I cards asexternal storage for the next Nintendo Switch model releasing in 2021. And I think Nintendo using CFExpress Type A cards as external storage for the next Nintendo Switch model releasing in 2021 would be as much as a PlayStation Vita situation as Nintendo using SD Express 8.0 cards as external storage for a 2023 console since CFExpress currently supports up to a PCIe 3.0 x4 interface, not a PCIe 4.0 x4 interface, unfortunately. And as a result, Microsoft did have to collaborate with Seagate to design the Seagate Storage Expansion Card for Xbox Series X|S, which I think is basically a custom CFExpress Type B Card with SK Hynix's H25BFT8G5M3P 128-Layer 4D NAND chip and Phison's PS5019-E19T PCIe 4.0 x4 NVMe SSD controller chip. But saying that, I think there's a chance a 2023 console might use CFExpress Type A cards for external storage, especially if Microsoft and Seagate manage to sell a good amount of the Seagate Storage Expansion Cards for Xbox Series X|S, and more companies are encouraged to adopt CFExpress cards for external storage to the point that CFExpress cards become reasonably inexpensive to produce and purchase.

And as much as I'd love for the rumour of the next Nintendo Switch model releasing in 2021 to use a magnesium alloy housing, DigiTimes can be a hit or a miss when it comes to rumours, so unfortunately I think there's a good chance the magnesium alloy housing rumour from DigiTimes won't pan out.

SiG said:
Then we have the whole ARM thing: I'm wondering if with Nvidia acquiring ARM, they could custom make a "big little" system that has the beefiest ARM chip as its "big" paired with several weaker "little" ARM chips, and because its an SoC there would be a direct bridge connector between CPU and GPU to allow for increased bandwidth.

Nvidia needs regulatory approval from the UK, European Union, China, and the US before Nvidia officially acquires Arm, so Nvidia has yet to acquire Arm. (But whether or not Nvidia successfully acquires Arm is another story altogether.) And since Nvidia believes that 18 months (from when it was announced that Nvidia is planning on acquiring Arm for $40 billion) are needed to get regulatory approval, I don't think the next Nintendo Switch model releasing in 2021 and a 2023 consoles are going to see any benefits when the CPU is concerned.
As for designing a custom Arm based CPU, Nvidia attempted to do so with Denver, Denver 2, and Carmel. But since Nvidia's using Arm's Cortex-A78AE for Orin, I think it's safe to say Nvidia wasn't successful in designing a custom Arm based CPU.
And by "custom make a "big little" system", do you mean something like designing a CPU with heterogeneous configuration of 8 Cortex-A78 cores and 4 Cortex-A55 cores? Otherwise, what you're describing sounds like what Arm big.LITTLE and Arm DynamIQ are essentially.

SiG said:
Alsright, I'm constantly getting confused about the use of the X1 nomeclature:

Tegra X1 = Switch OG 2017 = 20nm
Tegra X1+ = 'Mariko' = Red Box Switch 2019 = 12nm
Is there some other "TX1" chip that I'm confusing it with?

No, that's correct.

z0m3le · Oct 21, 2020

ZachyCatGames said:
I've tested compiling software and Gamecube emulation. Power consumption and temps only starts getting crazy at about 2GHz.
Dolphin stresses both the GPU and CPU, but the system remains cool and I still get around 2.5-3 hours of battery (this is on Erista)

I can't find any other articles with that 16W number.
You can't compare mdarcy and abca, they're different devices.
Higher clockspeeds have been tested on IowaMariko hardware and the results are very much in my favor.
L4T sources point to Mariko using far less power than Erista.
I'm not responding any further as this is fucking stupid.

We don't have access to the same information as you, what we can do is look at public info and look at the charts and graphics, mdarcy uses Mariko, yes it has different hardware configurations, but the power consumption here is almost entirely from the SoC clocks, I've estimated Switch Mariko at 14 watts, which would leave some head room for the differences in hardware configuration and makes sense when you look at idle and video streaming consumption.

You don't have to respond, this is an open forum, I am not posting this information to disprove you, but to further everyone's understanding of Switch hardware so we can discuss the potential of Mariko in a Pro model, and it seems unlikely from what I can see, though you have different information, so for you it might look different, but what you said about CTCaer's comments makes sense with what we are seeing in that article I linked, the drop from 1.9GHz to 1.785GHz is about 1 watt, the drop from 1267MHz to 921MHz is going to be 3 or so watts, the difference in hardware would also likely result in 1 less watt or so, that puts it in line with low end consumption of Erista to hit the 1785MHz/921MHz profile of ~11watts. You aren't being disproven by my posts, they aren't conflicting, what is conflicting is the idea that Nintendo could use the higher GPU clock as well, which will definitely result in ~14watts or so, that isn't out of the TDP range of the Switch, but it is higher than Erista with it's much lower clocks.

Sorry if this is frustrating you, but I'm not trying to attack you here. I do apologize if it is coming off that way.

fwd-bwd · Oct 21, 2020

Did everyone try the chicken? I thought the chicken was lovely. Joking aside, from where I sit z0m3le's posts do not appear to be an attack. While most of us are speaking in hypotheticals here, ZachyCatGames has been making some rather definitive claims. It's within reason that some would probe the veracity of said claims—not the individual.

On to a different topic: I've been experimenting with piping the Switch video into a Shield to simulate how the latter's "AI" upscaling could potentially be employed by the 2021 Switch. Here are the cropped screenshots of Rayman Legends with the "basic" 4K upscale (top) and "AI" 4K upscale (bottom).

It's no DLSS but not too shabby. I'll post more when I have the time later this week.

MP! · Oct 21, 2020

fwd-bwd said:
Did everyone try the chicken? I thought the chicken was lovely. Joking aside, from where I sit z0m3le's posts do not appear to be an attack. While most of us are speaking in hypotheticals here, ZachyCatGames has been making some rather definitive claims. It's within reason that some would probe the veracity of said claims—not the individual.

On to a different topic: I've been experimenting with piping the Switch video into a Shield to simulate how the latter's "AI" upscaling could potentially be employed by the 2021 Switch. Here are the cropped screenshots of Rayman Legends with the "basic" 4K upscale (top) and "AI" 4K upscale (bottom).

It's no DLSS but not too shabby. I'll post more when I have the time later this week.

wow that's a definite improvement
this is being ran native not streaming, correct?

fwd-bwd · Oct 21, 2020

MP! said:
wow that's a definite improvement
this is being ran native not streaming, correct?

Yes, it's local. I used an HDMI capture card to pull the Switch video (1080p) into a computer, Chromecasted the video (1080p) to Shield, upscaled it (4k), and used a second capture card to pull the Shield video (4k) into another computer for the screenshots, all in real-time. The second set of capture card and computer was needed because Shield can only take 1080p screenshot natively.

Pokemaniac · Oct 21, 2020

SiG said:
Alsright, I'm constantly getting confused about the use of the X1 nomeclature:

Tegra X1 = Switch OG 2017 = 20nm
Tegra X1+ = 'Mariko' = Red Box Switch 2019 = 12nm
Is there some other "TX1" chip that I'm confusing it with?

Those are the only two that exist right now.

That said, a theoretical 8nm version of the TX1 has been proposed as a candidate for if Nintendo is going the traditional Pro route, since Mariko's headroom seems like it would be kind of limited for that sort of device.

z0m3le · Oct 21, 2020

So from the digital foundry power consumption test of Zelda BotW Switch Erista vs Switch Mariko, we have 145 Minutes and 258.5 Minutes of game time. The battery life of the Switch is 4350mah @ 5V for 21.75WH. At 1Ghz CPU + 460MHz GPU: Erista consumes 7 watts, while Mariko consumes 5 watts.

Docked the Mariko unit consumes 6 to 8 watts during Zelda botw, calling it 7 watts for the average here means that Mariko could run docked GPU clock with the same power consumption as launch Switch units, but importantly I feel, 12nm Mariko cannot see a CPU upclock or GPU upclock beyond this without a larger battery capacity, running the CPU at 1785MHz and GPU at 768MHz should result in a battery life around ~2 hours 20 Minutes in Zelda BotW, this is portable performance but it's a bit too low, now you could use the joycons 1070mah in a new model to get you to 3 hours, that is the solution for Mariko.

Docked, to hit 1267MHz on the GPU for ~650GFLOPs, it will draw around 14 watts, charging the battery consumes an additional 5 watts, so you are looking at 19 watts for the Switch when docked and those clocks with Mariko. These are good enough numbers to be honest, I don't know if the whole joycon thing makes sense however, I feel like Nintendo would limit the charge from the joycons, so that when you dock the Switch, the joycons can still have a charge to use them, leaving 30% would drop the battery life when portable to just over 2 and a half hours in BotW. I don't think Nintendo will go this route.

Shrinking Tegra X1 further is viable, because you don't need to increase the CPU more for Switch gen 1 games, meaning you can increase the GPU to say 1.6GHz for 800GFLOPs and have a similar battery life to launch Switch units without joycon batteries needing to be drained. That is about the limit of Tegra X1 though, 7nm would be marginally better, say 30 more minutes of playtime but a similar clock speed.

With Nvidia hiring a senior software engineer for AI and DLSS on future console hardware, I think this likely means that that hardware already exists, since they would need to work on the hardware, it should be far closer than 3 years off, and that leads me back to Bloomberg's report about 4K in next year's model, it lines up with that report and the job listing, but of course I've only explained the 3 options ahead of Nintendo, I want to narrow it down further, but really all 3 options have pros and cons, I think current Mariko is the weakest and only really being discussed because Nintendo is actively testing unreleased hardware with Mariko, but any Nvidia produced SoC could replace Mariko for any work being done. We will have to wait and see.

fwd-bwd said:
Did everyone try the chicken? I thought the chicken was lovely. Joking aside, from where I sit z0m3le's posts do not appear to be an attack. While most of us are speaking in hypotheticals here, ZachyCatGames has been making some rather definitive claims. It's within reason that some would probe the veracity of said claims—not the individual.

On to a different topic: I've been experimenting with piping the Switch video into a Shield to simulate how the latter's "AI" upscaling could potentially be employed by the 2021 Switch. Here are the cropped screenshots of Rayman Legends with the "basic" 4K upscale (top) and "AI" 4K upscale (bottom).

It's no DLSS but not too shabby. I'll post more when I have the time later this week.

This is amazing, very interesting to see the technology, is the lag noticeably worse with the upscaling vs just the video feed?

It still doesn't fit bloomberg's report that developers are readying software for 4K on the Switch, but it's obviously a viable option and I wonder if the 32bit version of Tegra X1 found in the cheaper shield tv could be used for this, because it is a much cheaper variant.

Blomkvist · Oct 21, 2020

I have enjoyed reading all pages of the thread. Here is my take.
Overclocked Mariko won't cut it for "Pro" revision because it is bottlenecked by CPU and RAM. Shrunk down TX1+ is not happening since there is no chip with A57 on 8 nm or 7 nm node. My bet is on Xavier derivative. The rumors started to spread when the Xavier NX started mass production. With its 384 cores, DLA, tensor cores etc. it is a great candidate for a dev kit of a compatible device. There is nothing else it could be. And last thing: upcoming Orin SOC will be backwards compatible with Xavier.

Jraphics Horse · Oct 21, 2020

z0m3le said:
This is amazing, very interesting to see the technology, is the lag noticeably worse with the upscaling vs just the video feed?

It still doesn't fit bloomberg's report that developers are readying software for 4K on the Switch, but it's obviously a viable option and I wonder if the 32bit version of Tegra X1 found in the cheaper shield tv could be used for this, because it is a much cheaper variant.

I think there could be a lot of ambiguity there if it does mean anything for next year. The result of a smart 4K scaler is going to look better the higher the input resolution is, and if such scaling would be in the tablet itself rather than the dock, you could also be layering over higher resolution HUD.

Mercury_Sagit · Oct 21, 2020

fwd-bwd said:
Did everyone try the chicken? I thought the chicken was lovely. Joking aside, from where I sit z0m3le's posts do not appear to be an attack. While most of us are speaking in hypotheticals here, ZachyCatGames has been making some rather definitive claims. It's within reason that some would probe the veracity of said claims—not the individual.

On to a different topic: I've been experimenting with piping the Switch video into a Shield to simulate how the latter's "AI" upscaling could potentially be employed by the 2021 Switch. Here are the cropped screenshots of Rayman Legends with the "basic" 4K upscale (top) and "AI" 4K upscale (bottom).

It's no DLSS but not too shabby. I'll post more when I have the time later this week.

Thanks a lot for your analysis. The only missing piece in the puzzle would be latency, but at least IQ-wise I'm now way more confident in what DLSS can do for game with animated art styles like this (so far only games with (hyper)realistic art styles are supported)
Edit:

skittzo0413 said:
That's not DLSS, that's Nvidia's AI upscaling which is much easier to implement. Theoretically that could be done in the dock, whereas DLSS needs to happen during the rendering process.

Yeah what I meant is that the AI upscaling method in Shield models serves as a baseline and I expect DLSS to perform better (at least in term of IQ, whether the latency is improved is up to the next Switch model hardware)

Skittzo · Oct 21, 2020

Mercury_Sagit said:
Thanks a lot for your analysis. The only missing piece in the puzzle would be latency, but at least IQ-wise I'm now way more confident in what DLSS can do for game with animated art styles like this (so far only games with (hyper)realistic art styles are supported)

That's not DLSS, that's Nvidia's AI upscaling which is much easier to implement. Theoretically that could be done in the dock, whereas DLSS needs to happen during the rendering process.

Deleted member 79256 · Oct 21, 2020

z0m3le · Oct 21, 2020

ZachyCatGames said:
CTCaer said 921/1785mhz on Mariko would get the same battery life as Erista at its stock handheld clocks

imgur.com

Imgur: The magic of the Internet

imgur.com

It's fine, I was tired and sorta pissed off lol.

To throw another wrench into this ig, from what I can tell from limited testing 2295mhz CPU + 1267mhz GPU on Erista uses less than 18W.

It's alright, everyone has their own thing.

Mariko Switch units consume up to 8 watts during botw when docked, launch Switch units play botw at 7.1 watts to 9 watts depending on brightness settings. What CTCaer said doesn't seem to match up with what we are seeing.

We also know that clocking the CPU to 1.785GHz on the launch model consumes 17 watts when docked, so I don't think the other number could be correct either.

The erista shield tv throttles at 19 watts and tries to maintain the 1.9GHz CPU throttling the GPU, maybe these chips you are talking about are being undervolted, but 20nm Erista throttled even worse in Pixel C tablet, 1.9GHz on the CPU but only 200mhz on the GPU, I don't know what's going on with these numbers but they just don't make sense to me.

Pokemaniac · Oct 21, 2020

Blomkvist said:
I have enjoyed reading all pages of the thread. Here is my take.
Overclocked Mariko won't cut it for "Pro" revision because it is bottlenecked by CPU and RAM. Shrunk down TX1+ is not happening since there is no chip with A57 on 8 nm or 7 nm node. My bet is on Xavier derivative. The rumors started to spread when the Xavier NX started mass production. With its 384 cores, DLA, tensor cores etc. it is a great candidate for a dev kit of a compatible device. There is nothing else it could be. And last thing: upcoming Orin SOC will be backwards compatible with Xavier.

I don't think Xavier makes a whole lot of sense as a base. The CPU would have to be replaced entirely, and IIRC, based on some previous discussion here, the GPU lacks support for certain things Nintendo might be looking for, like DLSS. If the next Switch SoC is going to be based on an existing one, Orin makes more sense, but Nintendo's needs have diverged enough from most of the other Tegra customers that they might basically just get their own chip.

Tora · Oct 21, 2020

After experiencing VR with the Quest 2, I can totally see the next Switch offering VR capabilities

DLSS would really help to enable this, it seems like a no brainer for every single title

bmfrosty · Oct 21, 2020

Dakhil said:
As much as I would like the next Nintendo Switch model releasing in 2021 to use CFExpress Type A cards, SD Express 8.0 cards, etc., for external storage, unfortunately, I think Nintendo's going to continue to use UHS-I cards asexternal storage for the next Nintendo Switch model releasing in 2021. And I think Nintendo using CFExpress Type A cards as external storage for the next Nintendo Switch model releasing in 2021 would be as much as a PlayStation Vita situation as Nintendo using SD Express 8.0 cards as external storage for a 2023 console since CFExpress currently supports up to a PCIe 3.0 x4 interface, not a PCIe 4.0 x4 interface, unfortunately. And as a result, Microsoft did have to collaborate with Seagate to design the Seagate Storage Expansion Card for Xbox Series X|S, which I think is basically a custom CFExpress Type B Card with SK Hynix's H25BFT8G5M3P 128-Layer 4D NAND chip and Phison's PS5019-E19T PCIe 4.0 x4 NVMe SSD controller chip. But saying that, I think there's a chance a 2023 console might use CFExpress Type A cards for external storage, especially if Microsoft and Seagate manage to sell a good amount of the Seagate Storage Expansion Cards for Xbox Series X|S, and more companies are encouraged to adopt CFExpress cards for external storage to the point that CFExpress cards become reasonably inexpensive to produce and purchase.

And as much as I'd love for the rumour of the next Nintendo Switch model releasing in 2021 to use a magnesium alloy housing, DigiTimes can be a hit or a miss when it comes to rumours, so unfortunately I think there's a good chance the magnesium alloy housing rumour from DigiTimes won't pan out.

That's why I designated an NVMe solution as hope. I think it's equally or more likely that UHS-I will be deemed as sufficient.

As far as the magnesium allow body is concerned, I think that it's a far shot, but a premium model switch is a possibility with the same clocks as abca(x). I think of it as an enthusiast model that would probably use the same circuit boards (or minor changes to deal with a slightly larger body), but maybe a bigger battery. Probably a higher quality screen, or at least a guarantee of whatever type of screen that's deemed the best of the types that are currently in production.

Deleted member 79256 · Oct 21, 2020

Dakhil · Oct 21, 2020

z0m3le said:
The erista shield tv throttles at 19 watts and tries to maintain the 1.9GHz CPU throttling the GPU, maybe these chips you are talking about are being undervolted, but 20nm Erista throttled even worse in Pixel C tablet, 1.9GHz on the CPU but only 200mhz on the GPU, I don't know what's going on with these numbers but they just don't make sense to me.

I think that at least part of the reason is that the Pixel C is passively cooled whilst the Nvidia Shield TV is actively cooled.

fwd-bwd · Oct 21, 2020

z0m3le said:
This is amazing, very interesting to see the technology, is the lag noticeably worse with the upscaling vs just the video feed?

It still doesn't fit bloomberg's report that developers are readying software for 4K on the Switch, but it's obviously a viable option and I wonder if the 32bit version of Tegra X1 found in the cheaper shield tv could be used for this, because it is a much cheaper variant.

Since I used Chromecast to send the Switch feed to Shield, it introduced massive latency therefore I can't really answer your question. That said, when I tried the "AI" upscaling with GeForce Now, the latency is barely perceptible to me. Please note that I'm privileged enough to have a fast connection and also not a technically demanding player, so YMMY. FWIW, the GeForce Now PR claimed that the upscaling only adds an 1-2 frames lag (presumably he meant 60fps). The GeForce Now upscaling is only available on the 2019 Shield Pro, which runs on a 64bit Mariko. It probably indicates that the feature would be too laggy on the Erista or 32bit Mariko.

bmfrosty · Oct 21, 2020

fwd-bwd said:
FWIW, the GeForce Now PR claimed that the upscaling only adds an 1-2 frames lag (presumably he meant 60fps).

I care about 2 frames of lag on a game by game basis. For a Final Fantasy or Dragon Quest, it's pretty much not a problem. On a lot of FPS games, it's not a problem, but for some it is. Guitar Hero was able to build in frame lag compensation, but for some games, you just can't really. Street Fighter or Hades with 2 frames of lag is really problematic.

Dakhil · Oct 21, 2020

So VideoCardz has reported on a rumour about Nvidia allegedly cancelling the December launch of the RTX 3080 20 GB and the RTX 3070 16 GB. (I'm taking VideoCardz's rumour with a huge grain of salt since VideoCardz seems to be all over the place when Ampere GPU rumours are concerned.) I've heard people speculating that this is an indication that Nvidia plans on fabricating consumer Ampere GPUs releasing in 2021 at TSMC's 7 nm nodes for a consumer Ampere GPU refresh in 2021.

I assume it's safe to say this has no effect on Nintendo's and Nvidia's plans for the next Nintendo Switch model releasing in 2021, especially if Nintendo and Nvidia has decided to use a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021? (I don't know if Nintendo and Nvidia has planned on using a 8 nm Tegra X1+ or a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

BlackTangMaster · Oct 21, 2020

Pokemaniac said:
I don't think Xavier makes a whole lot of sense as a base. The CPU would have to be replaced entirely, and IIRC, based on some previous discussion here, the GPU lacks support for certain things Nintendo might be looking for, like DLSS. If the next Switch SoC is going to be based on an existing one, Orin makes more sense, but Nintendo's needs have diverged enough from most of the other Tegra customers that they might basically just get their own chip.

Once again, Xavier could be used as a base for Dev units as he said. You can easily consider Carmel cores being the equivalent of any ARMv8 cores from A57 to A78, especially if they are not running at full speed. And the GPU is fully compatible with CUDA and support DLSS via integrated tensor cores (I'm not talking about the DLA). But you are also right... A new device will probably be sold as a Orin-derivative chip with A78s (Orin-S?) and not as Xavier-derivative.

ZachyCatGames said:
While using 2295MHz CPU + 1267MHz GPU + 1862MHz EMC I didn't experience any battery drain, nor have I experienced battery drain from 1785MHz + 921MHz + 1600MHz in Horizon on any game due to charger limitations.
fwiw I have a better binning than most Switches and even certain Shield TV devices, but it shouldn't make a massive difference... idk...

What CTC has said also lines up with what people have experienced with overclocking on Mariko.

This is proper OC madness. You should ask for Linus S. to watercool this monstrosity. Is suprising me that your switch is consuming less than half of what the Xavier AGX board consume with twice the amount of CPU cores, 2*GPU cores and a lot of hardware accelerated parts. That speak volumes about what can be done on 8 nm with newer and more efficient CPU cores for a new model.

Dakhil said:
So VideoCardz has reported on a rumour about Nvidia allegedly cancelling the December launch of the RTX 3080 20 GB and the RTX 3070 16 GB. (I'm taking VideoCardz's rumour with a huge grain of salt since VideoCardz seems to be all over the place when Ampere GPU rumours are concerned.) I've heard people speculating that this is an indication that Nvidia plans on fabricating consumer Ampere GPUs releasing in 2021 at TSMC's 7 nm nodes for a consumer Ampere GPU refresh in 2021.

I assume it's safe to say this has no effect on Nintendo's and Nvidia's plans for the next Nintendo Switch model releasing in 2021, especially if Nintendo and Nvidia has decided to use a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021? (I don't know if Nintendo and Nvidia has planned on using a 8 nm Tegra X1+ or a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

There is not a lot of room for a price war with that amount of GDDR6X.

Pokemaniac · Oct 21, 2020

BlackTangMaster said:
Once again, Xavier could be used as a base for Dev units as he said. You can easily consider Carmel cores being the equivalent of any ARMv8 cores from A57 to A78, especially if they are not running at full speed. And the GPU is fully compatible with CUDA and support DLSS via integrated tensor cores (I'm not talking about the DLA). But you are also right... A new device will probably be sold as a Orin-derivative chip with A78s (Orin-S?) and not as Xavier-derivative.

The post I was responding to was specifically about using Xavier in the actual final device, not just early devkits.

BlackTangMaster · Oct 21, 2020

Pokemaniac said:
The post I was responding to was specifically about using Xavier in the actual final device, not just early devkits.

He said "it is a great candidate for a dev kit of a compatible device". I may be wrong here.

Pokemaniac · Oct 21, 2020

BlackTangMaster said:
He said "it is a great candidate for a dev kit of a compatible device". I may be wrong here.

I could be reading it wrong, but the "My bet is on Xavier derivative." bit seems to be referring to the final device.

z0m3le · Oct 21, 2020

ZachyCatGames said:
The Switch can only take 15v 1.2A from the power adapter, which is 18W, if the system uses over 18W on the official power adapter you'll experience battery drain.
While using 2295MHz CPU + 1267MHz GPU + 1862MHz EMC I didn't experience any battery drain, nor have I experienced battery drain from 1785MHz + 921MHz + 1600MHz in Horizon on any game due to charger limitations.
fwiw I have a better binning than most Switches and even certain Shield TV devices, but it shouldn't make a massive difference... idk...

What CTC has said also lines up with what people have experienced with overclocking on Mariko.

According to the Anandtech chart here, Erista uses similar power consumption playing botw as digital foundry records in their power consumption video of Mariko when docked.

EDIT: beyond the above information, I don't really know what's going on with other clocks, we do know Erista draws over 11 watts docked. Mariko does this with a power consumption of 6 to 8 watts. Maybe CTCaer was suggesting those high clocks (1.785GHz/921MHz) vs docked Erista, which seems to line up better, as those clocks should increase the power draw by about 4 watts to 10-12 watts, which is very similar to what digital foundry records for Erista when docked in that video.

plebc · Oct 21, 2020

Does this mean Pokemon Shied has a memory bandwidth bottleneck? GPU usage stays the same even in busy areas.

source: https://www.twitch.tv/zachycatgames livestream of the leaked dev build

Procheno · Oct 21, 2020

Could anyone summarize the general consensus from this thread?

I know better than to expect a PS4 -> PS4 Pro jump, but I'd like something decently substantial

Onix555 · Oct 21, 2020

plebc said:
Does this mean Pokemon Shied has a memory bandwidth bottleneck? GPU usage stays the same even in busy areas.

source: https://www.twitch.tv/zachycatgames livestream of the leaked dev build

It's a development build of a game by a historically.. uh let's say 'inefficient' developer.
It's the norm for them to be all over the place in FPS.

bmfrosty · Oct 21, 2020

Procheno said:
Could anyone summarize the general consensus from this thread?

I know better than to expect a PS4 -> PS4 Pro jump, but I'd like something decently substantial

Two trains of thought:
Pessimistic - We get a die-shrunk Erista/Mariko based Switch with higher clocks.
Optimistic - We get a Switch with a new SoC based around the Cortex A78 and Ampere

EDIT for Additional:
Model abcd is a VR unit.
Extra Pessimistic is that we don't get a more capable Switch, just another die shrink.

Deleted member 79256 · Oct 21, 2020

fwd-bwd · Oct 21, 2020

bmfrosty said:
I care about 2 frames of lag on a game by game basis. For a Final Fantasy or Dragon Quest, it's pretty much not a problem. On a lot of FPS games, it's not a problem, but for some it is. Guitar Hero was able to build in frame lag compensation, but for some games, you just can't really. Street Fighter or Hades with 2 frames of lag is really problematic.

Fair point. The "AI" upscaler is a bolted-on feature to GeForce Now, and probably can be further optimized. With a faster SoC and full integration into the NVN, its performance could be much improved on the 2021 Switch. In case that it wasn't clear, I'm not advocating for the "AI" upscaler but simply providing an alternative theory to Bloomberg's 4K report. It might be an option if the DLSS doesn't make its way onto the 2021 Switch, or as a secondary option to developers who don't want to deal with motion vectors.

Edit: Real-time motion vector info is required for DLSS 2 to work. Thus even if the 2021 Switch is equipped with DLSS, there may be developers that don't support it. A simpler "AI" upscaler could be a fallback option for them.

Pokemaniac · Oct 21, 2020

plebc said:
Does this mean Pokemon Shied has a memory bandwidth bottleneck? GPU usage stays the same even in busy areas.

source: https://www.twitch.tv/zachycatgames livestream of the leaked dev build

I'm pretty sure that's supposed to be a development build, so it's probably not wise to draw any conclusions about how the final game performs from it.

bmfrosty · Oct 21, 2020

fwd-bwd said:
Fair point. The "AI" upscaler is a bolted-on feature to GeForce Now, and probably can be further optimized. With a faster SoC and full integration into the NVN, its performance could be much improved on the 2021 Switch. In case that it wasn't clear, I'm not advocating for the "AI" upscaler but simply providing an alternative theory to Bloomberg's 4K report. It might be an option if the DLSS doesn't make its way onto the 2021 Switch, or as a secondary option to developers who don't want to deal with motion vectors.

Edit: Real-time motion vector info is required for DLSS 2 to work. Thus even if the 2021 Switch is equipped with DLSS, there may be developers that don't support it. A simpler "AI" upscaler could be a fallback option for them.

Interesting. I'm thinking of a more complex pipeline personally.

One layer would be 3d rendering, and that would be done at whatever resolution and be scaled by some method or combination of methods to 4k. That could be DLSS to get 540p up to 1080p, and then AI upscaling to bring that up to 4k as an example. Maybe a bad example. I'm not that type of engineer. The UI itself could be a second layer that is natively 4k and could be overlaid on top of the 3d elements in the frame buffer. That would probably get you a reasonably high resolution image. If things are sufficiently advanced, the game could only render the 3D elements that would not be obscured by a UI overlay. All that said, most modern games don't have much in the way of permanent UI elements, so that may not even matter.

NineTailSage · Oct 21, 2020

Dakhil said:
So VideoCardz has reported on a rumour about Nvidia allegedly cancelling the December launch of the RTX 3080 20 GB and the RTX 3070 16 GB. (I'm taking VideoCardz's rumour with a huge grain of salt since VideoCardz seems to be all over the place when Ampere GPU rumours are concerned.) I've heard people speculating that this is an indication that Nvidia plans on fabricating consumer Ampere GPUs releasing in 2021 at TSMC's 7 nm nodes for a consumer Ampere GPU refresh in 2021.

I assume it's safe to say this has no effect on Nintendo's and Nvidia's plans for the next Nintendo Switch model releasing in 2021, especially if Nintendo and Nvidia has decided to use a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021? (I don't know if Nintendo and Nvidia has planned on using a 8 nm Tegra X1+ or a 8 nm Tegra SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

Maybe Nvidia are thinking about manufacturing their Super or Ti variations of Ampere on TSMC 7nm, but Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over.

Dakhil · Oct 22, 2020

I don't know if Nvidia's talking about sampling or if Nvidia's ahead of schedule when it comes to Orin, but Nvidia mentioned in a blog post about Xavier that "NVIDIA Orin, our next-generation SoC, is coming next year, delivering nearly 7x the performance of Xavier with incredible energy-efficiency." And it should be mentioned that Nvidia said it was "...targeting automakers' 2022 production timelines" when Nvidia first announced Orin on 17 December 2019.

NineTailSage said:
Maybe Nvidia are thinking about manufacturing their Super or Ti variations of Ampere on TSMC 7nm, but Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over.

I don't know. A new SoC, especially a SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade to me.

z0m3le · Oct 22, 2020

Dakhil said:
I don't know if Nvidia's talking about sampling or if Nvidia's ahead of schedule when it comes to Orin, but Nvidia mentioned in a blog post about Xavier that "NVIDIA Orin, our next-generation SoC, is coming next year, delivering nearly 7x the performance of Xavier with incredible energy-efficiency." And it should be mentioned that Nvidia said it was "...targeting automakers' 2022 production timelines" when Nvidia first announced Orin on 17 December 2019.

I don't know. A new SoC, especially a SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade to me.

PS4 Pro came out 3 years after the PS4, XB1X came out 4 years after the XB1, Switch's next model if it comes out in september would be 4 and a half years after Switch, the same amount of time as the wii u to Switch or just a few months less time than N64 to gamecube or gamecube to Wii.

EDIT it just depends on Nintendo's plans, it's either a gen 2 Switch like an ios device or it's a slight upgrade to try and keep the Switch relevant for an extra year.

Executive Taxi · Oct 22, 2020

Orin really seems like a generational leap for all intents and purposes. I feel like Orin is so powerful that they can rely on it for several years and hold its ground against PS5 and XSX. I could definitely see Nintendo marketing the console that uses this as the Switch 2.

NineTailSage · Oct 22, 2020

Dakhil said:
I don't know if Nvidia's talking about sampling or if Nvidia's ahead of schedule when it comes to Orin, but Nvidia mentioned in a blog post about Xavier that "NVIDIA Orin, our next-generation SoC, is coming next year, delivering nearly 7x the performance of Xavier with incredible energy-efficiency." And it should be mentioned that Nvidia said it was "...targeting automakers' 2022 production timelines" when Nvidia first announced Orin on 17 December 2019.

I don't know. A new SoC, especially a SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade to me.

At this point it's probably semantics for whatever Nintendo chooses this device to be.
I agree that usually a new SoC with newer GPU features should indicate the next-gen device, but maybe they have a new hardware roadmap that consists of continuous upgrades evolving the platform over time.

Dakhil · Oct 22, 2020

z0m3le said:
PS4 Pro came out 3 years after the PS4, XB1X came out 4 years after the XB1, Switch's next model if it comes out in september would be 4 and a half years after Switch, the same amount of time as the wii u to Switch or just a few months less time than N64 to gamecube or gamecube to Wii.

EDIT it just depends on Nintendo's plans, it's either a gen 2 Switch like an ios device or it's a slight upgrade to try and keep the Switch relevant for an extra year.

I was responding to NineTailSage's comment that "...Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over" since in my opinion, a new SoC, especially if it's a new SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade. And the gap between the Nintendo 64 and the Nintendo GameCube was around 5 years, 1 month, and 20 days and the gap between the Nintendo GameCube and the Wii was around 5 years and a day in North America. However, the gap between the Wii U and the Nintendo Switch is around 4 years, 3 months, and 13 days in North America. So you'd be right about the gap being roughly a little more than 4 years and 6 months if Nintendo decides to release the next Nintendo Switch model on September 2021, although the gap between the Nintendo Switch andthe next Nintendo Switch model in this instance would be a little longer than the gap between the Wii U and the Nintendo Switch in North America. (I have no idea if Nintendo and Nvidia decide to go for a 8 mm Tegra X1+ or a 8 nm SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

NineTailSage said:
I agree that usually a new SoC with newer GPU features should indicate the next-gen device, but maybe they have a new hardware roadmap that consists of continuous upgrades evolving the platform over time.

One problem with iterative upgrades is that the performance improvements of Arm CPUs after the Cortex-A78 will only be marginal, which might be problematic, especially when it comes to ports of next-gen games, since the strongest consumer Arm CPUs will still be weaker than a Zen 2 based CPU used on the PlayStation 5, the Xbox Series S, and the Xbox Series X. Another problem is that the "big" performance cores of Arm CPUs released in 2022 and onwards will only support 64-bit, which might be problematic since there are 32-bit Nintendo Switch games; and I'm sure there will be 32-bit games developed exclusively for the next Nintendo Switch model releasing in 2021 if Nintendo doesn't mandate that all games must be 64-bit. And as a result, Nintendo might not go for newer Arm CPUs, especially if newer Arm CPUs don't support 32-bit, and if Nintendo wants hardware backwards compatibility, which seems rather likely. I should mention there's no guarantee Nvidia's going to successfully acquire Arm within 18 months (from when Nvidia announced its intentions to acquire Arm from Softbank).

NineTailSage · Oct 22, 2020

Dakhil said:
I was responding to NineTailSage's comment that "...Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over" since in my opinion, a new SoC, especially if it's a new SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade. And the gap between the Nintendo 64 and the Nintendo GameCube was around 5 years, 1 month, and 20 days and the gap between the Nintendo GameCube and the Wii was around 5 years and a day in North America. However, the gap between the Wii U and the Nintendo Switch is around 4 years, 3 months, and 13 days in North America. So you'd be right about the gap being roughly a little more than 4 years and 6 months if Nintendo decides to release the next Nintendo Switch model on September 2021, although the gap between the Nintendo Switch andthe next Nintendo Switch model in this instance would be a little longer than the gap between the Wii U and the Nintendo Switch in North America. (I have no idea if Nintendo and Nvidia decide to go for a 8 mm Tegra X1+ or a 8 nm SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

One problem with iterative upgrades is that the performance improvements of Arm CPUs after the Cortex-A78 will only be marginal, which might be problematic, especially when it comes to ports of next-gen games, since the strongest consumer Arm CPUs will still be weaker than a Zen 2 based CPU used on the PlayStation 5, the Xbox Series S, and the Xbox Series X. Another problem is that the "big" performance cores of Arm CPUs released in 2022 and onwards will only support 64-bit, which might be problematic since there are 32-bit Nintendo Switch games; and I'm sure there will be 32-bit games developed exclusively for the next Nintendo Switch model releasing in 2021 if Nintendo doesn't mandate that all games must be 64-bit. And as a result, Nintendo might not go for newer Arm CPUs, especially if newer Arm CPUs don't support 32-bit, and if Nintendo wants hardware backwards compatibility, which seems rather likely. I should mention there's no guarantee Nvidia's going to successfully acquire Arm within 18 months (from when Nvidia announced its intentions to acquire Arm from Softbank).

Realistically though for the next 3-4 systems, would Nintendo need anything more capable than the A78 for processing?
With the possibility of adding more cores, increasing clock speeds and CPU cache, there's quite a few things they could do to keep a Switch style device competitive with processing (in perspective to it's form factor) to receive 3rd party games.

We know the GPU potential for growth is there already and things like DLSS, VRS and HDR make the visual fidelity side of things a no brainer going forward, but CPU side I don't think Nintendo needs anything for the foreseeable future beyond A78/X1. Their biggest hurdles as the iterations become more powerful is memory bandwidth, storage data speeds/size and power draw and thermal dissipation...

Pokemaniac · Oct 22, 2020

Dakhil said:
I was responding to NineTailSage's comment that "...Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over" since in my opinion, a new SoC, especially if it's a new SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade. And the gap between the Nintendo 64 and the Nintendo GameCube was around 5 years, 1 month, and 20 days and the gap between the Nintendo GameCube and the Wii was around 5 years and a day in North America. However, the gap between the Wii U and the Nintendo Switch is around 4 years, 3 months, and 13 days in North America. So you'd be right about the gap being roughly a little more than 4 years and 6 months if Nintendo decides to release the next Nintendo Switch model on September 2021, although the gap between the Nintendo Switch andthe next Nintendo Switch model in this instance would be a little longer than the gap between the Wii U and the Nintendo Switch in North America. (I have no idea if Nintendo and Nvidia decide to go for a 8 mm Tegra X1+ or a 8 nm SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

One problem with iterative upgrades is that the performance improvements of Arm CPUs after the Cortex-A78 will only be marginal, which might be problematic, especially when it comes to ports of next-gen games, since the strongest consumer Arm CPUs will still be weaker than a Zen 2 based CPU used on the PlayStation 5, the Xbox Series S, and the Xbox Series X. Another problem is that the "big" performance cores of Arm CPUs released in 2022 and onwards will only support 64-bit, which might be problematic since there are 32-bit Nintendo Switch games; and I'm sure there will be 32-bit games developed exclusively for the next Nintendo Switch model releasing in 2021 if Nintendo doesn't mandate that all games must be 64-bit. And as a result, Nintendo might not go for newer Arm CPUs, especially if newer Arm CPUs don't support 32-bit, and if Nintendo wants hardware backwards compatibility, which seems rather likely. I should mention there's no guarantee Nvidia's going to successfully acquire Arm within 18 months (from when Nvidia announced its intentions to acquire Arm from Softbank).

Based on how it's been used, the 32-bit game support seems to mainly for enabling quick Wii U and 3DS ports, and, especially with that well quickly drying up, I think the chances of that becoming a BC only thing on Switch 2 are looking pretty high.

z0m3le · Oct 22, 2020

Dakhil said:
I was responding to NineTailSage's comment that "...Samsung's 8nm would be more than enough for a decent mid-gen refresh Switch if they do choose to go with a new SoC or if it's to move TX1+ over" since in my opinion, a new SoC, especially if it's a new SoC with an Ampere based GPU, seems more like a successor upgrade than a mid-gen refresh upgrade. And the gap between the Nintendo 64 and the Nintendo GameCube was around 5 years, 1 month, and 20 days and the gap between the Nintendo GameCube and the Wii was around 5 years and a day in North America. However, the gap between the Wii U and the Nintendo Switch is around 4 years, 3 months, and 13 days in North America. So you'd be right about the gap being roughly a little more than 4 years and 6 months if Nintendo decides to release the next Nintendo Switch model on September 2021, although the gap between the Nintendo Switch andthe next Nintendo Switch model in this instance would be a little longer than the gap between the Wii U and the Nintendo Switch in North America. (I have no idea if Nintendo and Nvidia decide to go for a 8 mm Tegra X1+ or a 8 nm SoC with an Ampere based GPU for the next Nintendo Switch model releasing in 2021.)

One problem with iterative upgrades is that the performance improvements of Arm CPUs after the Cortex-A78 will only be marginal, which might be problematic, especially when it comes to ports of next-gen games, since the strongest consumer Arm CPUs will still be weaker than a Zen 2 based CPU used on the PlayStation 5, the Xbox Series S, and the Xbox Series X. Another problem is that the "big" performance cores of Arm CPUs released in 2022 and onwards will only support 64-bit, which might be problematic since there are 32-bit Nintendo Switch games; and I'm sure there will be 32-bit games developed exclusively for the next Nintendo Switch model releasing in 2021 if Nintendo doesn't mandate that all games must be 64-bit. And as a result, Nintendo might not go for newer Arm CPUs, especially if newer Arm CPUs don't support 32-bit, and if Nintendo wants hardware backwards compatibility, which seems rather likely. I should mention there's no guarantee Nvidia's going to successfully acquire Arm within 18 months (from when Nvidia announced its intentions to acquire Arm from Softbank).

Yeah, I was just pointing out that the timeline might make perfect sense for overlapping generations that Switch might be scheduled to follow. If you believe that the 4th year that the Switch is currently in, happens to be the middle of its life cycle, that would mean that the 8th year is the end of its life cycle, so March 2025 could be the last scheduled release for Gen 1 Switch for instance.

Nintendo has the choice to build an overlapping generation with next year's model, supporting it too for 8 years, putting the end of that device in fall 2029, while also launching Switch gen 3 in fall 2025. Nintendo could always have 4+ years of device sales to transition between Switch generations, and this can help them manage newer technologies like DLSS, while also leaving them room to maneuver and try new gimmicks and form factors. All of this while avoiding wii to wii u or even 360 to xb1 type of transitions, xb1x could have been a whole new generation for Microsoft that would only be transitioning from the xb1 at the end of next year.

I actually don't see a reason to abandon the Switch as a platform for the foreseeable future. We will have to wait and see what they do.

T002 Tyrant · Oct 22, 2020

So the cycle could be:

2021 - Switch 2 has all first-party software cross-generation with the Switch. Allow the third parties to bring out exclusive software but encourage cross-gen compatibility whenever possible
2022 - Start to make at least one exclusive first-party exclusive. Introduce new hardware (Lite/Home). Decrease the manufacture of Switch, Increase Switch 2.
2023- Discontinue Switch fully, increase the frequency of first-party exclusives to Switch 2.
2024 - Introduce Switch 3. OG Switch no longer supported by first-party software coming out 2024, so Switch 2 and 3 will have games only playable on those systems. Repeat Steps.

z0m3le · Oct 22, 2020

T002 Tyrant said:
So the cycle could be:

2021 - Switch 2 has all first-party software cross-generation with the Switch. Allow the third parties to bring out exclusive software but encourage cross-gen compatibility whenever possible
2022 - Start to make at least one exclusive first-party exclusive. Introduce new hardware (Lite/Home). Decrease the manufacture of Switch, Increase Switch 2.
2023- Discontinue Switch fully, increase the frequency of first-party exclusives to Switch 2.
2024 - Introduce Switch 3. OG Switch no longer supported by first-party software coming out 2024, so Switch 2 and 3 will have games only playable on those systems. Repeat Steps.

I think life cycle should naturally follow sales peak cycles...!!! (EDIT: Yes right here a light went off in my brain)

I think I know why they delayed the Switch upgrade from 2020 to 2021... The peak sales cycle...

Found the article I remembered reading here (btw this analyst predicted 130M sales in 2022, Switch will easily pass 100M sales next year, if it sells like this year it could be at 115M by March 2022, it could indeed pass 130M in 2022):

"As Nintendo's Switch is a unique game platform for use as either a console or handheld, there was initially some uncertainty over whether its sales cycle would more closely resemble that of a console or handheld, Although the Switch is currently categorized as more of a console, we expect it to sell like a handheld. Many users are in fact using the Switch as a portable according to a Famitsu survey of users. We previously projected a sales peak at about three years after launch (i.e., in FY3/19) on the assumption of a console-like sales cycle. Now that the Switch looks more likely to sell like a handheld, we update our sales forecast to reflect a longer sales cycle. We now expect the Switch to achieve peak sales about five years after launch (i.e., around FY3/20–21) like a typical handheld game device. We also raise our peak-year sales forecast from 25mn to 31mn units."

Nintendo is releasing the model next year because it wants to introduce a new product during the peak of sales, this is used to extend the maturity stage of the product, which is usually very short for electronic devices, but just like the DS, Switch is in a niche, and that niche is actually even more rare than the DS, so the lack of competition will help keep the maturity stage longer.

The goal with releasing a model next year and not this year is because it will replace the current Switch in a more meaningful way, it won't be just a new tier, but will transition to a new "generation" of the device... Basically Nintendo's goal with Switch is indeed overlapping generations as Iwata said and Nintendo bulls are hoping for in the market place... They will try to release generations to maintain the maturity stage of Switch's product life.

A 'Pro' model is generally placed earlier, during the growth stage, this is because it is trying to expand the market sales, basically it's introducing a new tier to sell to new customers, while a gen 2 type product like iphone 2, is reinvigorating sales of current users, it replaces the device in the market place, at least overtime.

We see this with galaxy phones for an example, Samsung introduces 2 Galaxy Note 20 products, Note 20 and Note 20 Ultra... That is the Pro of Galaxy Note 20, but what this timing suggests is actually Galaxy Note 30.

Jraphics Horse · Oct 22, 2020

Nintendo still see themselves in a growth stage, and they certainly didn't plan for Covid to give them a premature peak.

I think the comment about 'families gathered around a console' will fit into their next design, whether that means a bigger screen model, or two screens or whatever will better support tabletop mode.

z0m3le · Oct 22, 2020

Jraphics Horse said:
Nintendo still see themselves in a growth stage, and they certainly didn't plan for Covid to give them a premature peak.

I think the comment about 'families gathered around a console' will fit into their next design, whether that means a bigger screen model, or two screens or whatever will better support tabletop mode.

Nintendo moved next year's model from this year in like January of 2019, before a pandemic, and I think it was because of how Switch had been selling for the past 2 years at that point and all the research they had done and explained to board members about the Switch selling as a portable largely.

I think the ultimate goal with families is probably getting to a $300-$400 price point for a group AR experience of 4 players, that sort of thing is something I bet Nintendo has tons of ideas around. With streaming and video upscaling technology demonstrated a few posts up of Rayman, I think you could get a very low latency video feed to 4 players via wifi 6 and have small safety glasses style AR Goggles upscale to 1080p-1440p in the goggles.

Being able to get Mariko into a tube like this, the next step is a board 1/4th that size, you could realistically fit the board on one arm and the battery on the other arm (of the glasses). This could enable AR, and I think Pokemon Go showed Nintendo the potential of such a device, this is all just a guess, but I believe this is what they would be looking at, the social aspect of AR and the ability to completely overtake everything in the entertainment space has plenty of businesses interested, but I think Nintendo is right at the front of the pack, especially working with a hardware provider in Nvidia, who basically is the only company that could support a device like that in a realistic time frame of the end of the decade.

We will see how things go, but I've been thinking about an AR device from Nintendo since 2007's anime Denno Coil went live and it seem to be going in that direction in 2010 when I saw Nintendo's 3DS AR stuff. In my opinion, its a matter of time until Nintendo does something big with portable AR devices, I really do have it pinned for the end of the decade currently. Ramble over.

Jraphics Horse · Oct 22, 2020

z0m3le what model do you think was moved beyond 2020? And remember that the quoted Nikkei article claim about a delay was a mistranslation.

z0m3le · Oct 22, 2020

Jraphics Horse said:
z0m3le what model do you think was moved beyond 2020? And remember that the quoted Nikkei article claim about a delay was a mistranslation.

The one that zakatana and myself had heard was coming in 2020, way back in 2018. Also, that Nikkei article being mistranslated was very confusing at the time, what was mistranslated has continued to change from translation to translation, for instance one of the translations said that it had power consumption problems, another said that they couldn't get the device to work properly with the os, and another said the project didn't even have a lead. Nikkei did write an article, and I'm not a native speaker, so I don't have a definitive translation, but for there to be a report about a delay to the model following the Switch lite, and to hear previously about a model that was scheduled for 2020, and all the rumors about a Switch Pro from 2017+, I think it's not hard to trace a delay in early 2019, since talks of a pro model stopped circulating and things became a lot harder to get information on, when this year started it was definitive in January that there would be no new Switch model this year, and that was directly to board members, it also happened before the pandemic really locked down the world, also talks about the device mysteriously dried up right around the time the Nikkei article was published.

Here is a digitimes report for new model in 2020 FYQ1.

EDIT: This article is actually a gold mine:

"This Switch news is the latest in a long line of reports that date back to 2018. In October 2018, a Wall Street Journal report said Nintendo wanted to maintain its momentum from the Switch's initial release and stay competitive in the current market by bringing in new models.

Japanese newspaper Nikkei reported in early 2019 that a cheaper, more portable Switch was on the way and that a next-gen model was also in the works. The report said Nintendo is experimenting with a supposed high-end model and would change the Switch's usability, image rendering and operating system.

Nikkei's story corroborated a Wall Street Journal report in March that said both versions of the Switch could be released as early as the summer of 2019. To make the "mini" Switch more cost-effective, Nintendo would take out features like the Joy-Con vibrations, the Journal said.

In a CNET hands-on event with the Switch Lite in July, Nintendo of America President Doug Bowser told the outlet that the handheld console is not a "sequel" but an alternative to the original Switch. He also said another Switch would be coming in the future."

Deleted member 37917 · Oct 22, 2020

Jraphics Horse said:
z0m3le what model do you think was moved beyond 2020? And remember that the quoted Nikkei article claim about a delay was a mistranslation.

yeah like z0mbi3 said, they had a hardware targeting 2020 back in 2018, and I know that it wasn't the refresh from 2019, which I had no idea existed (same for the switch lite which I discovered at the same time as everyone here). I have nothing else than my good faith to support this claim, which is why I preferred to pass it as a prediction at the time I learned about it.

Based on how tech has progressed since the Switch, how powerful could a Switch Pro in early 2021 actually be?

User requested account closure

The Enlightened Wise Ones

User requested account closure

Alt Account

User requested account closure