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

[Dakhil]

though 2.1 might not be terrible as it's not far off from the Series systems anyway

Click to shrink...

I think 2.1 GB/s is the highest data transfer speed that's possible with Samsung's UFS 3.1 chip. So I'm guessing sustained data transfer speeds for Samsung's UFS 3.1 chip is going to be roughly around 1.5 GB/s (assuming sustained data transfer speeds are around 25% lower than the highest data transfer speeds).

I know that RedGamingTech is not the most reliable source for news and/or rumours, but apparently what kopite7kimi has said about Nvidia's 5 nm consumer GPUs corroborates with what RedGamingTech heard about Nvidia's 5 nm consumer GPUs.

• fabricated at TSMC's 5 nm nodes
• planned for release in 2022
• codenamed "Lovelace" (after Ada Lovelace, a British mathematician who was one of the first programmers)
• is a monolithic GPU rather than a MCM GPU
• more of a evolution of Ampere than a brand new GPU architecture

(It starts at 0:09 if the time stamp doesn't work.)
This is what kopite7kimi said about Nvidia's 5 nm consumer GPUs.

 

[KamiCrit]

I feel the current Nintendo Switch generation is wrapping up. We've had all the big mainline games hit the Switch, now it's time for the next big thing with Zelda bridging the gap. Gaming wise the future still seems to be X86, so I can see the next gen hybrid being backwards compatible with Switch games.

All the "Switch Pro" talk I'm betting is now getting rolled into the next console. Winter 2020 would of been the time to do the pro move but covid bringing on supply, engineering, planning disruptions said otherwise.

We're nearing 4 years of say a 5 year life cycle. The Switch U is real.

 

[Hermii]

I feel the current Nintendo Switch generation is wrapping up. We've had all the big mainline games hit the Switch, now it's time for the next big thing with Zelda bridging the gap. Gaming wise the future still seems to be X86, so I can see the next gen hybrid being backwards compatible with Switch games.

All the "Switch Pro" talk I'm betting is now getting rolled into the next console. Winter 2020 would of been the time to do the pro move but covid bringing on supply, engineering, planning disruptions said otherwise.

We're nearing 4 years of say a 5 year life cycle. The Switch U is real.

Click to shrink...

The future for low powered computing devices is absolutely not x86, gaming or not.

Apple is moving their entire lineup over to arm, and the rest of the laptop industry will follow sooner rather than later. It's x86 days that are numbered, arm has a long future ahead of it.

 

[Dakhil]

Apple is moving their entire lineup over to arm, and the rest of the laptop industry will follow sooner rather than later.

Click to shrink...

I think that depends on how competent Microsoft is in doing emulation of 64-bit programs. And if Microsoft continues to make Windows 10 on Arm run exclusively on Qualcomm's SoCs.

 

[Hermii]

I think that depends on how competent Microsoft is in doing emulation of 64-bit programs. And if Microsoft continues to make Windows 10 on Arm run exclusively on Qualcomm's SoCs.

Click to shrink...

Apples move, and the amazing performance they demonstrated on the m1 chip will light a fire under their asses, don't you think?

 

[Dakhil]

Apples move, and the amazing performance they demonstrated on the m1 chip will light a fire under their asses, don't you think?

Click to shrink...

I think that depends on if Qualcomm makes the same mistake as Intel since most Android smartphones run on Qualcomm's SoCs due to Qualcomm's 4G, 4G LTE, and 5G patents.

 

[ILikeFeet]

Apples move, and the amazing performance they demonstrated on the m1 chip will light a fire under their asses, don't you think?

Click to shrink...

all the more reason I think Nvidia has something to gain by making a Tegra for Windows. particularly in gpu performance

 

[Hermii]

I think that depends on if Qualcomm makes the same mistake as Intel since most Android smartphones run on Qualcomm's SoCs due to Qualcomm's 4G, 4G LTE, and 5G patents.

Click to shrink...

Not sure what mistakes you are referring to here.

 

[NineTailSage]

I think 2.1 GB/s is the highest data transfer speed that's possible with Samsung's UFS 3.1 chip. So I'm guessing sustained data transfer speeds for Samsung's UFS 3.1 chip is going to be roughly around 1.5 GB/s (assuming sustained data transfer speeds are around 25% lower than the highest data transfer speeds).

I know that RedGamingTech is not the most reliable source for news and/or rumours, but apparently what kopite7kimi has said about Nvidia's 5 nm consumer GPUs corroborates with what RedGamingTech heard about Nvidia's 5 nm consumer GPUs.

• fabricated at TSMC's 5 nm nodes
• planned for release in 2022
• codenamed "Lovelace" (after Ada Lovelace, a British mathematician who was one of the first programmers)
• is a monolithic GPU rather than a MCM GPU
• more of a evolution of Ampere than a brand new GPU architecture

(It starts at 0:09 if the time stamp doesn't work.)
This is what kopite7kimi said about Nvidia's 5 nm consumer GPUs.

Click to shrink...

Well according to people here in this thread, we know absolutely nothing about Nvidia's roadmap (even though Kopite7kimi has been extremely accurate in the past when it comes to Nvidia information). As if Nvidia themselves don't talk about their future tech directions internally and in public, which is how information gets leaked in the first place...

 

[Dekuman]

Yeah I don't think they'll launch a Pro followed by a Switch 2 a year apart. They would be tapping the same audience. If Pro is different enough that it warrants a Pro Lite, I can see a Switch Pro Lite coming in 2022.

But late 2023 is probably the best case for a successor.

 

[Dakhil]

Not sure what mistakes you are referring to here.

Click to shrink...

Continue to release underwhelming SoCs, considering that Qualcomm seem to suggest that Apple's following Qualcomm's lead in pushing Arm over x86, and letting the fact that most Android smartphones use Qualcomm's SoCs get to Qualcomm's head, despite the fact that it's only because of Qualcomm's 4G, 4G LTE, and 5G patents, which Qualcomm has complete monopoly over in the US.

 

[ILikeFeet]

or there could be no Pro and Nintendo just rides out 2021

 

[ShadowFox08]

I like Jon as well. And I don't think anything he says here is controversial actually.

If Nintendo sticks to a die shrunk TX1, they can OC it as much as they can and get maybe 2x performance overall on the GPU side, maybe also 2x on CPU with 2x bandwidth and 2x the Memory, and call it a day. It will immediately smooth out a lot of performance issues while allowing devs to release patched up games running at a solid 30fps at 720/900 otr even 1080p or a dynamic 60fps. I'm thinking service games like fornite, warframe will all get 60fps /high resolution patches.

Sticking to TX1 also means they can bring the successor with the full suite of more modern GPU features, DLSS 2.0+ (or 3.0), and improvcements such that even if it's only a 4-5x improvement in flop performance over OG Switch (or 2.5 to 3x more powerfil than the Pro), its more modern architecture and GPU features would still be significantly more powerful than Pro could ever be.

Click to shrink...

I don't feel very confident they can double clockspeeds for everything and have it at 15 watts in docked mode on 8nm. They could probably hit 750-800 GFLOPs GPU (1.5GHz) with 8GB RAM at 50GB/, but 2GHz CPU also? I just don't know Bout reaching both those GPU and CPU numbers.

It's funny because this is what the TX2 was meant for at 16nm ages ago (not sure about 2Ghz CPU). Has TX2 clocks ever been tested 🤔

 

[NineTailSage]

or there could be no Pro and Nintendo just rides out 2021

Click to shrink...

This is a very real possibility as well...
I think the naysayers don't understand that just because most of us are discussing the potential for future Switch tech, it's not ruling out the chance they may also do nothing next year and wait to just release the successor in 2022 or so.

 

[Pokemaniac]

Jon from Spawn Wave uploaded a YouTube video today talking about what he thinks the next Nintendo Switch model releasing in 2021 would be. I'm sure many people would disagree with Jon, but I thought it was worth discussing. (The bolded is my opinion.)

• Jon thinks Nintendo would name the next Nintendo Switch model releasing in 2021 the "New Nintendo Switch" since he thinks it worked for the New Nintendo 3DS, but he wouldn't mind it being called the "Super Nintendo Switch" (I disagree with Jon about the "New" moniker working for the New Nintendo 3DS)
• Jon thinks there are two things Nintendo doesn't want to do with the next Nintendo Switch model
  • Jon thinks Nintendo doesn't want to spend a lot of money on the next Nintendo Switch model since Nintendo would prefer to spend a lot of money on the successor as technology continues to improves throughout the years leading up to the successor
  • Jon thinks Nintendo doesn't want the performance jump between the Nintendo Switch and the next Nintendo Switch model to be too big since doing so would make the performance jump from the next Nintendo Switch model to the successor seem like less of a jump, which he argues is the case when going form the PlayStation 4 Pro and the Xbox One X to the PlayStation 5 and the Xbox Series X (I completely disagree with Jon about the performance jump between the PlayStation 4 Pro and the Xbox One X to the PlayStation 5 and the Xbox Series X seeming like less of a jump since the CPU on the PlayStation 5 and Xbox Series X is significantly better than the CPU on the PlayStation 4 Pro and the Xbox One X)
• Jon thinks Nintendo isn't too interested about third party developers developing games exclusive to the next Nintendo Switch model since it didn't work out for the New Nintendo 3DS (except for Minecraft since Minecraft sells no matter the hardware)
• Jon thinks people who want the next Nintendo Switch model wants Nintendo Switch games to load faster and run better, which is the case for Nintendo 3DS games running on the New Nintendo 3DS
• Jon says he could see Nintendo increasing the bus width from 64-bit to 128-bit, increasing the RAM amount from 4 GB to 8 GB, and slightly overclock the CPU and GPU on the Tegra X1, since it's cheaper than designing a brand new SoC from scratch (I have no idea if the Tegra X1 supports a 128-bit bus width; if the Tegra X1 doesn't support a 128-bit bus width, I don't think his argument stands)
• Jon thinks Nintendo will continue to use a 720p display since he thinks Nintendo doesn't want to spend money on things where Nintendo doesn't have to
• Jon thinks that Nintendo shouldn't go for a mini LED display for the next Nintendo Switch model, but rather Nintendo should go for the mini LED display for the successor; instead wait for Apple to incentivise more companies to manufacture more mini LED displays before Nintendo starts using mini LED displays
• Jon thinks that Nintendo would price the next Nintendo Switch model at $349.99 and thinks the next Nintendo Switch model would eventually replace the revised Nintendo Switch model
• Jon thinks that Nintendo might release a TV-only model for $149.99 at the end of the Nintendo Switch's life cycle
• Jon thinks that Nintendo is going to use the same form factor and same heat sink for the next Nintendo Switch model since it needs to use the same dock and use the same Joy-Cons (I think Nintendo needs to redesign and reengineer the Joy-Cons to make the Joy-Cons less susceptible to drift)
• Jon thinks it's going to be very difficult to change the look of the next Nintendo Switch model; he thinks that changing the logo and having new housing colours might help differentiate the next Nintendo Switch model with the Nintendo Switch (I think new housing colours are more helpful in terms of differentiating between the next Nintendo Switch model and the Nintendo Switch than changing the logo since I doubt people are going to notice the logo)
• Jon predict Nintendo's going to release the next Nintendo Switch model on 14 May 2021; he also mentions that an autumn 2021 release is also possible
• Jon thinks Nintendo's probably going to announce the next Nintendo Switch model in April, potentially giving consumers and retailers around 40 days to pre-order and to process pre-orders respectively
  • Jon mentions Nintendo doesn't tell retailers about any new hardware until after Nintendo announces new hardware
• Jon thinks the next Nintendo Switch model simply needs to be better than the Nintendo Switch (I don't think that's enough of a reason for me to buy the next Nintendo Switch model)
• Jon thinks third party developers might release a native version of games on the next Nintendo Switch model whilst releasing a cloud version of games on the Nintendo Switch

Where have you heard about the new Snapdragon 86x SoC? I've only heard of the Snapdragon 865 and the Snapdragon 865+.

Click to shrink...

Generally he seems to be expecting a traditional "Pro" revision, which is reasonable, but definitely not certain at this point. My main objections are that exclusives are probably happening, regardless of how much some people seem to have convinced themselves that they won't, and the whole "native on Pro, cloud on base Switch" thing seems kinda unlikely to actually to actually happen to any significant degree.

I feel the current Nintendo Switch generation is wrapping up. We've had all the big mainline games hit the Switch, now it's time for the next big thing with Zelda bridging the gap. Gaming wise the future still seems to be X86, so I can see the next gen hybrid being backwards compatible with Switch games.

All the "Switch Pro" talk I'm betting is now getting rolled into the next console. Winter 2020 would of been the time to do the pro move but covid bringing on supply, engineering, planning disruptions said otherwise.

We're nearing 4 years of say a 5 year life cycle. The Switch U is real.

Click to shrink...

x86 is dying. I wouldn't really call it the "future" of anything, especially not devices that have to run off a battery.

Apples move, and the amazing performance they demonstrated on the m1 chip will light a fire under their asses, don't you think?

Click to shrink...

Apple seems to be basically cheating on their emulation performance by using custom instructions. It's not going to be the easiest thing to respond to quickly, though it is an approach that I think some of the higher end standard ARM cores could start to copy.

Though I think the more immediate issue is that MS apparently still doesn't sell licenses for ARM Windows directly to consumers. It is absolutely embarrassing that it's not even possible to legally run Windows on these new ARM Macs. If the issues were just missing drivers, I could understand, but they have no excuse for how difficult it currently is to obtain a license for ARM Windows.

 

[ShadowFox08]

I got a quick question for the tech savvy. Are the A72 and A73 (Armv-8A) cores identical to the A57s on performance per Hz, with the only difference being that they are smaller and can run higher clocks (up to 2.8-3Ghz)? Or do they actually perform better per clock?

Edit: they have more cache.

Nevermind, a switch pro with TX1 really isn't going to go far. Nintendo isn't going to break CPU compatibility. If they could remove the A53s and add 2-4 more A57 more cores, that would help tremendously like the the new 3ds, but that ain't happening obviously.

this is probably the best showcase of why the Switch Pro/2 needs DLSS (5:40 if timestamp doesn't work). also, an 8-core Pro/2 will be a godsend for many games

Click to shrink...

If we got tx1 with pro, maybe we could get some of that Nvidia up scaling tech... Which is better than nothing.

 

[NineTailSage]

I got a quick question for the tech savvy. Are the A72 and A73 (Armv-8A) cores identical to the A57s on performance per Hz, with the only difference being that they are smaller and can run higher clocks (up to 2.8-3Ghz)? Or do they actually perform better per clock?

Edit: they have more cache.

Nevermind, a switch pro with TX1 really isn't going to go far. Nintendo isn't going to break CPU compatibility. If they could remove the A53s and add 2-4 more A57 more cores, that would help tremendously like the the new 3ds, but that ain't happening obviously.


If we got tx1 with pro, maybe we could get some of that Nvidia up scaling tech... Which is better than nothing.

Click to shrink...

Seems like this is the best summary of A72 in comparison to A57

ARM Cortex-A72 Architecture Deep Dive

ARM's Cortex-A72 CPU adds power and performance optimizations to the previous A57 design. Here's an in-depth look at the changes to each stage in the pipeline, from better branch prediction to next-gen execution units.

www.tomshardware.com

Just this quote from the article shows how inefficient the A57 was and is on 20nm.

"The Cortex-A72 is an evolution of the Cortex-A57; the baseline architecture is very similar. However, ARM tweaked the entire pipeline for better power and performance. Perhaps the A57's biggest weakness was its relatively high power consumption, especially on the 20nm node, which severely limited sustained performance in mobile devices, relegating it to short, bursty workloads and forcing SoCs to use the lower-performing Cortex-A53 cores for extended use."

 

[ILikeFeet]

If we got tx1 with pro, maybe we could get some of that Nvidia up scaling tech... Which is better than nothing.

Click to shrink...

We're not getting DLSS without dumping TX1. 1.9 is never coming back

 

[Lelouch0612]

We're not getting DLSS without dumping TX1. 1.9 is never coming back

Click to shrink...

I don't think they are refering to DLSS but to another AI scaling technique.

 

[ILikeFeet]

I don't think they are refering to DLSS but to another AI scaling technique.

Click to shrink...

What would be the point? Nvidia has an agenda to push so they're not gonna devote that kind of time to it, I think. Not to mention any other AI-based method might still take up computational power. Maybe if Nvidia adapted CAS for the switch?

 

[Dekuman]

Since the Switch chipset has been used in parallel in nvidia's own shieldTvs, where do we expect the next-generation Shield to land on? That may be a good hint on what the Switch Pro can do.

Is there a huge imperative the next-gen shield TV be on a brand new SoC?

I don't feel very confident they can double clockspeeds for everything and have it at 15 watts in docked mode on 8nm. They could probably hit 750-800 GFLOPs GPU (1.5GHz) with 8GB RAM at 50GB/, but 2GHz CPU also? I just don't know Bout reaching both those GPU and CPU numbers.

It's funny because this is what the TX2 was meant for at 16nm ages ago (not sure about 2Ghz CPU). Has TX2 clocks ever been tested 🤔

Click to shrink...

Is it absolutely vital they hit that 15 watt target? I suppose they could just trade off some battery life.

 

[Mr_F_Snowman]

Considering the unreal sales of the Switch right now and the many statements directly from Nintendo that the Switch is only now halfway through its life its curious why anyone would believe a successor comes out anytime before late 2022 / early 2023

 

[NineTailSage]

Considering the unreal sales of the Switch right now and the many statements directly from Nintendo that the Switch is only now halfway through its life its curious why anyone would believe a successor comes out anytime before late 2022 / early 2023

Click to shrink...

It's all about keeping the Switch base energized, many owners have had Switch since it launched and no longer play on it as consistently as they use to and want something shiny and new. This happens all of the time in technology where retention is the hardest thing to gage and then we are amazed when a company like Apple figures out how to sustain its ecosystem.

It's not just about the current sells of the hybrid and lite consoles, but how do we keep the enthusiast gamers hooked every few years not to look anywhere else because they know something new is coming from Nintendo. We would not only be talking about massive sells from the late adopters next year, but the surge in gamers that want something new, (they already own a Switch) but maybe can't get a PS5/XSX or they would just like the next greatest thing out...

 

[Dekuman]

Considering the unreal sales of the Switch right now and the many statements directly from Nintendo that the Switch is only now halfway through its life its curious why anyone would believe a successor comes out anytime before late 2022 / early 2023

Click to shrink...

It's all about keeping the Switch base energized, many owners have had Switch since it launched and no longer play on it as consistently as they use to and want something shiny and new. This happens all of the time in technology where retention is the hardest thing to gage and then we are amazed when a company like Apple figures out how to sustain its ecosystem.

It's not just about the current sells of the hybrid and lite consoles, but how do we keep the enthusiast gamers hooked every few years not to look anywhere else because they know something new is coming from Nintendo. We would not only be talking about massive sells from the late adopters next year, but the surge in gamers that want something new, (they already own a Switch) but maybe can't get a PS5/XSX or they would just like the next greatest thing out...

Click to shrink...

Their biggest mistake with Wii was not refreshing it in 2010. Wii U type device launched in 2010 (perhaps with a different lower power but still HD capable GPU) may have worked allowing them to launch a successor same time as PS4/XONE. But knowing its development history they were not in a position to even launch that hypothetical device.

Based on leaks, at least we know something is being done to refresh Switch.

 

[Look over there]

iirc per DF's Switch review at launch when they were building a Switch equivalent PC, they commented Switch's A57 cores had better performance compare to Jaguar cores, but Switch's A57 were also only clocked at 1Ghz, meaning per core it matched an equivalent Jaguar core. Had they been clocked higher, the per-core performance on Switch would have exceed the per core performance of the CPU cores in the XONE /PS4. It wouldn't make up for the fewer cores overall and not everything is CPU bound on Switch.

Click to shrink...

Wait, Digital Foundry said that 1Ghz A57 matched 1.6Ghz Jaguar? Is Jaguar really that much weaker than I thought (which was A57 ~= Jaguar on per clock basis)?

I got a quick question for the tech savvy. Are the A72 and A73 (Armv-8A) cores identical to the A57s on performance per Hz, with the only difference being that they are smaller and can run higher clocks (up to 2.8-3Ghz)? Or do they actually perform better per clock?

Click to shrink...

A57->A72: +16% integer, +26% floating point, more power efficient to some degree, and ~10% area reduction apparently
A72->A73: ~+4% integer, ~-4% floating point, more power efficient to some degree, and supposedly up to 25% area reduction on the same process. IPC-wise, it treaded water because this particular design had the core being more narrow, I believe.
I don't know how much more efficient they are, but I am convinced that there are some improvements there, as A7_ series architectures have no problem going above 2 ghz at 14 nm generation or later.

 

[NineTailSage]

Their biggest mistake with Wii was not refreshing it in 2010. Wii U type device launched in 2010 (perhaps with a different lower power but still HD capable GPU) may have worked allowing them to launch a successor same time as PS4/XONE. But knowing its development history they were not in a position to even launch that hypothetical device.

Based on leaks, at least we know something is being done to refresh Switch.

Click to shrink...

I agree and if they don't learn from that mistake, they could find themselves in an odd position again...
There's an ebb and flow to these transitions and Sony are pretty great at their hardware roll outs for the most part.
It's definitely something Nintendo has to get a grasp on, how to lure, maintain and migrate their player base during generations.

 

[ILikeFeet]

modern phone processors regularly hit over 2GHz, some getting close to 3GHz. so sustaining that won't be a problem

hell, a quick search brought up this 2019 processor made on Samsung's 8nm (speak of the devil) hitting 2.2GHz

Snapdragon 730G - Qualcomm - WikiChip

Snapdragon 730G is a mid-range 64-bit ARM LTE system on a chip designed by Qualcomm and introduced in early 2019. Fabricated on Samsung's 8nm LPP process, the 730G features six Kryo 470 Silver high-efficiency cores operating at 1.8 GHz along with two high-performance Kryo 470 Gold operating at...

en.wikichip.org

.

 

[z0m3le]

Jon from Spawn Wave uploaded a YouTube video today talking about what he thinks the next Nintendo Switch model releasing in 2021 would be. I'm sure many people would disagree with Jon, but I thought it was worth discussing. (The bolded is my opinion.)

• Jon thinks Nintendo would name the next Nintendo Switch model releasing in 2021 the "New Nintendo Switch" since he thinks it worked for the New Nintendo 3DS, but he wouldn't mind it being called the "Super Nintendo Switch" (I disagree with Jon about the "New" moniker working for the New Nintendo 3DS)
• Jon thinks there are two things Nintendo doesn't want to do with the next Nintendo Switch model
  • Jon thinks Nintendo doesn't want to spend a lot of money on the next Nintendo Switch model since Nintendo would prefer to spend a lot of money on the successor as technology continues to improves throughout the years leading up to the successor
  • Jon thinks Nintendo doesn't want the performance jump between the Nintendo Switch and the next Nintendo Switch model to be too big since doing so would make the performance jump from the next Nintendo Switch model to the successor seem like less of a jump, which he argues is the case when going form the PlayStation 4 Pro and the Xbox One X to the PlayStation 5 and the Xbox Series X (I completely disagree with Jon about the performance jump between the PlayStation 4 Pro and the Xbox One X to the PlayStation 5 and the Xbox Series X seeming like less of a jump since the CPU on the PlayStation 5 and Xbox Series X is significantly better than the CPU on the PlayStation 4 Pro and the Xbox One X)
• Jon thinks Nintendo isn't too interested about third party developers developing games exclusive to the next Nintendo Switch model since it didn't work out for the New Nintendo 3DS (except for Minecraft since Minecraft sells no matter the hardware)
• Jon thinks people who want the next Nintendo Switch model wants Nintendo Switch games to load faster and run better, which is the case for Nintendo 3DS games running on the New Nintendo 3DS
• Jon says he could see Nintendo increasing the bus width from 64-bit to 128-bit, increasing the RAM amount from 4 GB to 8 GB, and slightly overclock the CPU and GPU on the Tegra X1, since it's cheaper than designing a brand new SoC from scratch (I have no idea if the Tegra X1 supports a 128-bit bus width; if the Tegra X1 doesn't support a 128-bit bus width, I don't think his argument stands)
• Jon thinks Nintendo will continue to use a 720p display since he thinks Nintendo doesn't want to spend money on things where Nintendo doesn't have to
• Jon thinks that Nintendo shouldn't go for a mini LED display for the next Nintendo Switch model, but rather Nintendo should go for the mini LED display for the successor; instead wait for Apple to incentivise more companies to manufacture more mini LED displays before Nintendo starts using mini LED displays
• Jon thinks that Nintendo would price the next Nintendo Switch model at $349.99 and thinks the next Nintendo Switch model would eventually replace the revised Nintendo Switch model
• Jon thinks that Nintendo might release a TV-only model for $149.99 at the end of the Nintendo Switch's life cycle
• Jon thinks that Nintendo is going to use the same form factor and same heat sink for the next Nintendo Switch model since it needs to use the same dock and use the same Joy-Cons (I think Nintendo needs to redesign and reengineer the Joy-Cons to make the Joy-Cons less susceptible to drift)
• Jon thinks it's going to be very difficult to change the look of the next Nintendo Switch model; he thinks that changing the logo and having new housing colours might help differentiate the next Nintendo Switch model with the Nintendo Switch (I think new housing colours are more helpful in terms of differentiating between the next Nintendo Switch model and the Nintendo Switch than changing the logo since I doubt people are going to notice the logo)
• Jon predict Nintendo's going to release the next Nintendo Switch model on 14 May 2021; he also mentions that an autumn 2021 release is also possible
• Jon thinks Nintendo's probably going to announce the next Nintendo Switch model in April, potentially giving consumers and retailers around 40 days to pre-order and to process pre-orders respectively
  • Jon mentions Nintendo doesn't tell retailers about any new hardware until after Nintendo announces new hardware
• Jon thinks the next Nintendo Switch model simply needs to be better than the Nintendo Switch (I don't think that's enough of a reason for me to buy the next Nintendo Switch model)
• Jon thinks third party developers might release a native version of games on the next Nintendo Switch model whilst releasing a cloud version of games on the Nintendo Switch

Where have you heard about the new Snapdragon 86x SoC? I've only heard of the Snapdragon 865 and the Snapdragon 865+.

Click to shrink...

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

What makes more sense, selling an average of 18 Million units a year for the next 3 years (25M, 20M, 10M with a new Platform launching that year) or an average of 25 Million for the next 5 years? Because if next year's Switch is where we are speculating, there won't be a need for another upgraded Switch until 2025, and they will see sales for the next 5 years at ~30M, 25M, 25M (new lite model), 20M, 25M (switch 3)

 

[ILikeFeet]

the good news is that there will still be some pretty big games from JP third parties coming in 2022 for PS4. as long as the Switch Pro and hit that level (and have an as good or better CPU), they could probably sustain a Pro for a while. though that would risk letting games be exclusive. even a 4-core A78 at 1.78GHz would allow for many games but put it out of the range of the OG switch. might be risky

 

[NineTailSage]

They probably know that gaming cards as MCM will be a dumpster fire. Or you can argue that random guys on Twitter know better how to design GPUs than Nvidia.


There is nothing new in this tweet.
I've said already a few pages back - we know jack shit about NV's plans.
We don't know what Hopper is and how it is related to Ampere.
We don't know if Hopper is the next gaming architecture or not.
We don't know what NV plans are for the next architecture for either gaming or DC.
NV is using "MCM" in the form of GPU+HBM integration since Pascal but only for DC products.
Will they expand this into several logic dies in one package with Hopper?
Is Hopper a name of the architecture or the MCM with several logic dies on it? Which architecture do these dies use?
How will this affect gaming side of their lineup? It's been almost 5 years since GP100 came out with HBM and yet there are no gaming GPUs from NV with HBM still.
Basically we don't know anything on future NV plans.
And it's fairly clear that kopite7kimi doesn't either.

Click to shrink...

modern phone processors regularly hit over 2GHz, some getting close to 3GHz. so sustaining that won't be a problem

hell, a quick search brought up this 2019 processor made on Samsung's 8nm (speak of the devil) hitting 2.2GHz

Snapdragon 730G - Qualcomm - WikiChip

Snapdragon 730G is a mid-range 64-bit ARM LTE system on a chip designed by Qualcomm and introduced in early 2019. Fabricated on Samsung's 8nm LPP process, the 730G features six Kryo 470 Silver high-efficiency cores operating at 1.8 GHz along with two high-performance Kryo 470 Gold operating at...

en.wikichip.org

.

Click to shrink...

The thing about the Snapdragon 730 though only 2-A76 cores were @2.2Ghz +6-A55 cores @1.8Ghz.
Even in 2019 the Adreno 618 gpu put out 400Gflops(even though I know this is considered a low budget phone SoC).

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

What makes more sense, selling an average of 18 Million units a year for the next 3 years (25M, 20M, 10M with a new Platform launching that year) or an average of 25 Million for the next 5 years? Because if next year's Switch is where we are speculating, there won't be a need for another upgraded Switch until 2025, and they will see sales for the next 5 years at ~30M, 25M, 25M (new lite model), 20M, 25M (switch 3)

Click to shrink...

This is what I don't understand either and why there are people who don't seem to get why Nintendo would want to maintain continuous growth and platform retention than to go the withered cycle routine and then release.

 

[Pokemaniac]

Since the Switch chipset has been used in parallel in nvidia's own shieldTvs, where do we expect the next-generation Shield to land on? That may be a good hint on what the Switch Pro can do.

Is there a huge imperative the next-gen shield TV be on a brand new SoC?

Click to shrink...

I'll admit that I don't really follow what's going on with the Shield all that closely, but it kinda seems to just be following what the Switch does at this point, at least when it comes to the SoC.

I've never gotten the impression that the Shield is super popular as a device to play games natively on, so the main impetus to upgrade (aside from just higher output resolutions) would probably be for newer/better media codecs. In that regard, Ampere did notably add AV1 decoding support, so that could be a reason to upgrade.

Considering the unreal sales of the Switch right now and the many statements directly from Nintendo that the Switch is only now halfway through its life its curious why anyone would believe a successor comes out anytime before late 2022 / early 2023

Click to shrink...

The lifespan of a system doesn't just end the instant that its successor releases. When I predict that Switch 2 might release next year, what I'd expect to happen in that scenario is for the drop-off in Switch 1 support (at least for first party) to be very gradual, to the point where Nintendo will probably keep releasing at least some games for it basically right up until Switch 3 launches.

 

[ILikeFeet]

The thing about the Snapdragon 730 though only 2-A76 cores were @2.2Ghz +6-A55 cores @1.8Ghz.
Even in 2019 the Adreno 618 gpu put out 400Gflops(even though I know this is considered a low budget phone SoC).

Click to shrink...

those A55s still probably outperform the switch cpu by a decent margin

and yea, it's midbudget. here's a new phone at 360 euro-bucks with a 7mAh battery, 6GB or 8GB ram

Samsung Galaxy M51 review

7Ah. That's the battery capacity of the Samsung Galaxy M51 we have for you today. Sure, you can call that 7,000mAh (thanks, SI), but at such large and...

www.gsmarena.com

 

[FernandoRocker]

those A55s still probably outperform the switch cpu by a decent margin

and yea, it's midbudget. here's a new phone at 360 euro-bucks with a 7mAh battery, 6GB or 8GB ram

Samsung Galaxy M51 review

7Ah. That's the battery capacity of the Samsung Galaxy M51 we have for you today. Sure, you can call that 7,000mAh (thanks, SI), but at such large and...

www.gsmarena.com

Click to shrink...

I have an even cheaper phone (Poco X3 NFC) with the new Snapdragon 732G. It was only $250 USD with 6GB of RAM.

It's an 8nm processor, and the G in the name is because it has some gaming enhancements:

• Faster Kryo 470 CPU
• Up to 15% faster Adreno 618 GPU
• Up to 3.6 AI TOPS performance

Snapdragon 732G Mobile Platform | Qualcomm

The Snapdragon 732G Mobile Platform brings select Snapdragon Elite Gaming features and a 15% graphics boost to gaming-ready smartphones.

www.qualcomm.com

Qualcomm Snapdragon 732G Processor - Benchmarks and Specs

Benchmarks, information, and specifications for the Qualcomm Snapdragon 732G

www.notebookcheck.net

 

[Deleted member 79256]

snip

 

[BlackTangMaster]

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

What makes more sense, selling an average of 18 Million units a year for the next 3 years (25M, 20M, 10M with a new Platform launching that year) or an average of 25 Million for the next 5 years? Because if next year's Switch is where we are speculating, there won't be a need for another upgraded Switch until 2025, and they will see sales for the next 5 years at ~30M, 25M, 25M (new lite model), 20M, 25M (switch 3)

Click to shrink...

Especially considering what chip a $299 device can have in 2020 (Xbox series S - 190mm2 7 nm chip). I expect Nintendo to go cheap with either a <100mm2 7 nm chip or a <130mm2 8 nm chip. Both with a 2.5-3x transistor jump from TX1 which should be way above what a 'pro' model could bring.

The problem I have with a 2x more powerful pro model is that it could have already been possible with Mariko on 12 nm. Moreover, a > 2GHz chip would be really affected by power binning even on 8 nm.

Since the Switch chipset has been used in parallel in nvidia's own shieldTvs, where do we expect the next-generation Shield to land on? That may be a good hint on what the Switch Pro can do.

Is there a huge imperative the next-gen shield TV be on a brand new SoC?

Is it absolutely vital they hit that 15 watt target? I suppose they could just trade off some battery life.

Click to shrink...

Shield TV really needs to have a more powerful chip especially on the CPU side. And it already runs at 2GHz most of the time.

 

[Onix555]

Pretty excited for next year.

Thats said, first I need to get back into playing the games I bought. I hate having autism, procrastinating everything.

 

[NineTailSage]

Especially considering what chip a $299 device can have in 2020 (Xbox series S - 190mm2 7 nm chip). I expect Nintendo to go cheap with either a <100mm2 7 nm chip or a <130mm2 8 nm chip. Both with a 2.5-3x transistor jump from TX1 which should be way above what a 'pro' model could bring.

The problem I have with a 2x more powerful pro model is that it could have already been possible with Mariko on 12 nm. Moreover, a > 2GHz chip would be really affected by power binning even on 8 nm.

Shield TV really needs to have a more powerful chip especially on the CPU side. And it already runs at 2GHz most of the time.

Click to shrink...

I don't think a chip on 7 or 8nm being 100-130mm makes it cheap, Microsoft are losing money on the Series S so that may not be a great comparison. I also don't think a 2-3x transistor increase would only equate to double the power of the current Switch. Efficiency gains from a newer nodes like 7 or 8nm, plus newer ARM CPU cores and GPU architecture would have to be factored into these performance gains.

Unless you are speculating that they may redesign Maxwell cores into a new chip design?

 

[BlackTangMaster]

I don't think a chip on 7 or 8nm being 100-130mm makes it cheap, Microsoft are losing money on the Series S so that may not be a great comparison. I also don't think a 2-3x transistor increase would only equate to double the power of the current Switch. Efficiency gains from a newer nodes like 7 or 8nm, plus newer ARM CPU cores and GPU architecture would have to be factored into these performance gains.

Unless you are speculating that they may redesign Maxwell cores into a new chip design?

Click to shrink...

I expect a 2 5-3x bump in transistor count with a bump in CPU/GPU clocks.

That would mean a 1024 cuda core ampere GPU @1.4GHz docked (closer to 768 cuda core on 8 nm with probably less than a x2 bump in transistor count). And the transistor count of a 4+4 A57/A53 is probably close to the transistor count of a 4+4 A76/78 + A55).

Microsoft is not loosing more than 60/70$ on every series S.

 

[Mr_F_Snowman]

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

What makes more sense, selling an average of 18 Million units a year for the next 3 years (25M, 20M, 10M with a new Platform launching that year) or an average of 25 Million for the next 5 years? Because if next year's Switch is where we are speculating, there won't be a need for another upgraded Switch until 2025, and they will see sales for the next 5 years at ~30M, 25M, 25M (new lite model), 20M, 25M (switch 3)

Click to shrink...

You're assumption is simply historically incorrect - no platform refresh generates the kind of sales curve you are predicting where sales drop off in such a minimal way, it would be completely unprecedented. The only time we have seen a curve resembling anything like that was when the Kinect launched for Xbox 360 (the complete opposite of the kind of refresh you are expecting)

The Pro console models have not really altered the sales curve units wise - they will have allowed hardware revenues to curve downloads more gently and served as an alternative to deep ongoing price cuts but what you are predicting unit wise is just a guess that does not follow the data

 

[YolkFolk]

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

What makes more sense, selling an average of 18 Million units a year for the next 3 years (25M, 20M, 10M with a new Platform launching that year) or an average of 25 Million for the next 5 years? Because if next year's Switch is where we are speculating, there won't be a need for another upgraded Switch until 2025, and they will see sales for the next 5 years at ~30M, 25M, 25M (new lite model), 20M, 25M (switch 3)

Click to shrink...

I think people argue that because it's what every console manufacturer in the history of the industry has done during each and every generation transition in the history of the industry.

 

[NineTailSage]

I expect a 2 5-3x bump in transistor count with a bump in CPU/GPU clocks.

That would mean a 1024 cuda core ampere GPU @1.4GHz docked (closer to 768 cuda core on 8 nm with probably less than a x2 bump in transistor count). And the transistor count of a 4+4 A57/A53 is probably close to the transistor count of a 4+4 A76/78 + A55).

Microsoft is not loosing more than 60/70$ on every series S.

Click to shrink...

I'm still not following, with those specs that would equal a 2Tflop GPU roughly which would be a decent upgrade for Switch.

You're assumption is simply historically incorrect - no platform refresh generates the kind of sales curve you are predicting where sales drop off in such a minimal way, it would be completely unprecedented. The only time we have seen a curve resembling anything like that was when the Kinect launched for Xbox 360 (the complete opposite of the kind of refresh you are expecting)

The Pro console models have not really altered the sales curve units wise - they will have allowed hardware revenues to curve downloads more gently and served as an alternative to deep ongoing price cuts but what you are predicting unit wise is just a guess that does not follow the data

Click to shrink...

I still don't think it's hard to understand, so since people are making it harder than it needs to be let's equate it to physics.

Law of inertia, also called Newton's first law, postulate in physics that, a body in motion would remain in motion unless a force (such as friction) caused it to come to rest.

It just seems like people put more time and energy into not understanding something than including it as a possibility of something that could happen.

I think people argue that because it's what every console manufacturer in the history of the industry has done during each and every generation transition in the history of the industry.

Click to shrink...

That doesn't make it perpetual though, there were always cassette decks in cars until there wasn't.
The same could be said for CD players, DVD players and soon consoles won't ship with disc drives anymore either. It really does come down to evolution or die and any living organism or company that has failed to do this ends up in the history books.

 

[BlackTangMaster]

You're assumption is simply historically incorrect - no platform refresh generates the kind of sales curve you are predicting where sales drop off in such a minimal way, it would be completely unprecedented. The only time we have seen a curve resembling anything like that was when the Kinect launched for Xbox 360 (the complete opposite of the kind of refresh you are expecting)

The Pro console models have not really altered the sales curve units wise - they will have allowed hardware revenues to curve downloads more gently and served as an alternative to deep ongoing price cuts but what you are predicting unit wise is just a guess that does not follow the data

Click to shrink...

I think people argue that because it's what every console manufacturer in the history of the industry has done during each and every generation transition in the history of the industry.

Click to shrink...

Pro consoles didn't bring more game to their respective platform. The lead platform was the base PS4. Something OG switch never benefited.

 

[BlackTangMaster]

I'm still not following, with those specs that would equal a 2Tflop GPU roughly which would be a decent upgrade for Switch.

Click to shrink...

2TFLOP is a little bit lower than (120mm2 128bit bus) apple A12X/Z had in 2018 on TSMC's old N7 DUV node. I expect more than 2TFLOP from Nvidia on 7 nm in 2021. Maybe 1.2-1.5TFLOP on 8 nm.

Nintendo may heavily underclock their chip in order to maintain the best possible yields the way they did with OG switch. That's how I see them 'going cheap' instead of asking for a die shrunk 40mm2 TX1 in 2021 on a 4 years old node.

 

[Mr_F_Snowman]

I'm still not following, with those specs that would equal a 2Tflop GPU roughly which would be a decent upgrade for Switch.



I still don't think it's hard to understand, so since people are making it harder than it needs to be let's equate it to physics.

Law of inertia, also called Newton's first law, postulate in physics that, a body in motion would remain in motion unless a force (such as friction) caused it to come to rest.

It just seems like people put more time and energy into not understanding something than including it as a possibility of something that could happen.



That doesn't make it perpetual though, there were always cassette decks in cars until there wasn't.
The same could be said for CD players, DVD players and soon consoles won't ship with disc drives anymore either. It really does come down to evolution or die and any living organism or company that has failed to do this ends up in the history books.

Click to shrink...

Quoting Newtons laws of physics as oppose to looking at the historical and very recent data we have of comparable video game systems sure is one hell of a head scratcher lol

 

[Mr_F_Snowman]

Pro consoles didn't bring more game to their respective platform. The lead platform was the base PS4. Something OG switch never benefited.

Click to shrink...

I don't really understand your point considering a Switch Pro will also not be the lead platform for third parties. Any third party actually interested in putting games on Switch is already doing it. If third parties haven't been tempted by the 80 million install base of the OG Switch what makes you think a Switch Pro with a small share of future sales (25% if it follows Xbox and Playstation Pro models market share) would suddenly swamp it with games?

 

[ILikeFeet]

I don' think we'll get over 1TFLOP for a switch pro. gotta save such large jumps for a Switch 2

 

[Hermii]

If next years revision is indeed getting a new up to date chip, and is coming out by mid next year I think its high time we get a reliable leak or news story saying so. Every day that doesn't happen, I believe it less.

 

[ILikeFeet]

If next years revision is indeed getting a new up to date chip, and is coming out by mid next year I think its high time we get a reliable leak or news story saying so. Every day that doesn't happen, I believe it less.

Click to shrink...

I think we would have heard about something if covid hadn't happened. if it's coming after H2, then I can see updated dev kits and whatnot going out early next year

 

[RennanNT]

I just think it's amazing people can argue that the business makes sense to roll the dice again in 3 years with a Switch 2, rather than to maintain the momentum of the current platform and just try to keep this platform going for another 3 to 5 years with peak sales instead.

Click to shrink...

The thing is, overlapping generations worked amazingly well for phones, but we have no guarantee it will work for gaming systems. It's an uncharted territory as far as dedicated gaming segment goes.

Also, PS consoles, XBox and Nintendo handhelds traditional gens have been quite consistent, they had one screw up each and even then it wasn't that much as a disaster.

Short and overlapping generations has the potential for heights traditional gens can't get, but it's also more risky than going traditional. The right call isn't nearly as clear as both sides in this argument try to make them be.

 

[K Samedi]

If next years revision is indeed getting a new up to date chip, and is coming out by mid next year I think its high time we get a reliable leak or news story saying so. Every day that doesn't happen, I believe it less.

Click to shrink...

There already are plenty of leaks.