Switch 2 Speculation

If it wasn't a nintendo console, I'd think they would want to keep these. LTE tablets and laptops are already very wide spread. No reason a handheld can't have it as a feature.

Another console maker (e.g. Sony or Microsoft) might still prefer to leave the modem out of the SoC, so they could offer different SKUs with and without phone network capability.
As for the ISP and A.I. DSP, the ISP they'd probably shelve because I can't see how a mobile console would need the ability to take a 100MPx photo in less than a second. Perhaps they could find some use for the A.I. DSP, but only if they found a strong enough use case for it (which nvidia could not, for Turing.. despite all the DLSS marketing).
 
I think one strength that was o. Switch's side was that despite its large preformance diferential to the bone and ps4, it had feature parity with them. With that in mind, I think it would be highly desireable for switch 2 to have a feature-set that matches ps4 and xbsex, even if the brute horsepower is more like that of the base one.
They'll need to do something about the CPU though as Switch is already at a deficit in cores/threads, and even if A57s may have a bit better IPC, next to the midgen twins, that's < 50% clock speed to boot.
 
They'll need to do something about the CPU though as Switch is already at a deficit in cores/threads, and even if A57s may have a bit better IPC, next to the midgen twins, that's < 50% clock speed to boot.

I think I remember seeing the Cortex A57 getting very similar IPC to the Jaguar cores (at least in geekbench?)

Jaguar core at 1500MHz: https://browser.geekbench.com/geekbench3/search?q=a4-5000
Cortex A57 at 1912MHz: https://browser.geekbench.com/geekbench3/search?q=Pixel+C

On the Jaguar, ~900 score * (1600/1500) = 960
On the Cortex A57, ~1200 score * (1600 / 1912) = 1004

Then we need to consider that the A4-5000 uses at best 64bit DDR3 1600MT/s, for a total of 12.8GB/s, but the Pixel C is using 64bit LPDDR4 3200MT/s so it has twice the bandwidth 25.6GB/s. That ought to make some difference.


It's with Cortex A76 that IPC really sets it apart from the previous ARMv8 big cores.
 
It's with Cortex A76 that IPC really sets it apart from the previous ARMv8 big cores.

If I am reading correctly on the Arm website, the A76 still uses the same instruction set as the A57, so if backwards compatibility is desired, then it should a good option. Not only is the A76 a big step up in terms of IPC, but its clock speeds can be way higher without drawing to much power. The A57 cores powering the Switch are a meager 1Ghz ( really wish Nintendo would boost it to 1.4Ghz), so not only would the A76 cores have a much higher IPC, they would likely be clocked twice as high.
 
If I am reading correctly on the Arm website, the A76 still uses the same instruction set as the A57, so if backwards compatibility is desired, then it should a good option. Not only is the A76 a big step up in terms of IPC, but its clock speeds can be way higher without drawing to much power. The A57 cores powering the Switch are a meager 1Ghz ( really wish Nintendo would boost it to 1.4Ghz), so not only would the A76 cores have a much higher IPC, they would likely be clocked twice as high.

Yes, I'd say a 4-core Cortex A76 @ 2GHz would probably consume about almost as much power as the A57 @ 1GHz in the 20nm Tegra X1, while performing ~2.5x faster.
Though if what they want is CPU performance parity with the 8th-gen consoles then their best bet would be 8x A76 cores at 1.4 or 1.5GHz.
 
Now I wonder, that much power will allow for what new Nintendo feature?

With BC as given with a76. It seems really fitting for Nintendo. As they usually release new handheld (be it a new break or a "pro" model) with higher performance AND a new feature while keeping BC. Not just more perf with the same features
 
Contrary to popular belief (among TX1 fans), Qualcomm does semicustom solutions, and the Snapdragon SQ1 for the Surface Pro X has 8-channel LPDDR4X with 68GB/s, a 3GHz 4-Core Cortex A76 + 1.8GHz 4-Core Cortex A55 and GPU with 1.05 TFLOPs FP32 / 2.1 TFLOPs FP16.......
Not only that, they are not averse to making small volume special purpose versions of their SoCs, such as the VR/AR focused XR chips, which are produced and sold in tiny volumes compared to a Switch2 SoC.
As far as I understand, Nintendo isn’t married to Nvidia from a software standpoint either. Vulkan and industry standard middleware is is used for game production, so new titles should be reasonably straight forward, and reasonable backwards compatibility as well. If you require strict bug-for-bug backwards compatibility though, that is obviously a trickier proposition. It’s not even necessarily straightforward using the same SoC supplier.

Going with Nvidia would require Nintendo to shoulder the cost of a ground-up 5nm SoC design (for instance). Even though that is certainly possible with Switch volumes it still is quite a commitment, if repurposing for instance a Qualcomm SoC is a viable alternative. If I were Qualcomm/Mediatek/.. I would take notice of the situation, and approach Nintendo.
 
Going with Nvidia would require Nintendo to shoulder the cost of a ground-up 5nm SoC design (for instance).

This is an expense that Nintendo has always paid in the past. The Tegra X1 is the first off the shelf chip Nintendo has used. Hell, they paid IBM a good dollar to transform the PPC750 CPU into a custom 3 core processor for the Wii U. If the partnership with Nvidia is healthy, it would be a very Nintendo thing to do by sticking with them. I find it hard to believe that it would break the bank for Nvidia to pair some ARM CPU cores with as many Volta GPU cores as possible within the power draw constraints of the platform. Nintendo doesn't prioritize having the most powerful hardware, so even if its a weaker SOC than Qualcom could offer, I doubt that would be a deal breaker for Nintendo. Many of their most successful platforms have had weaker processing power compared to the competition. Anything that exceeds base PS4 power will be sufficient for the Switch successor. It allows third parties the opportunity to rerelease PS4/X1 software on the Switch 2.
 
This is an expense that Nintendo has always paid in the past. The Tegra X1 is the first off the shelf chip Nintendo has used. Hell, they paid IBM a good dollar to transform the PPC750 CPU into a custom 3 core processor for the Wii U. If the partnership with Nvidia is healthy, it would be a very Nintendo thing to do by sticking with them. I find it hard to believe that it would break the bank for Nvidia to pair some ARM CPU cores with as many Volta GPU cores as possible within the power draw constraints of the platform. Nintendo doesn't prioritize having the most powerful hardware, so even if its a weaker SOC than Qualcom could offer, I doubt that would be a deal breaker for Nintendo. Many of their most successful platforms have had weaker processing power compared to the competition. Anything that exceeds base PS4 power will be sufficient for the Switch successor. It allows third parties the opportunity to rerelease PS4/X1 software on the Switch 2.
You’re right. Of course there are advantages to staying with Nvidia. Cost is not likely to be one of them though.
Nintendo will have a list of desires, and a target pricepoint that is likely to be quite firm. So what can Nvidia deliver at a given node and price?
It will be really interesting to see the design decisions going into the successor to the Switch.
So the next typical Nintendo thing to do is have 2 TX1s ducttaped together for their NextGen.
Nah. That would be the PS4Pro.
 
This is an expense that Nintendo has always paid in the past. The Tegra X1 is the first off the shelf chip Nintendo has used. Hell, they paid IBM a good dollar to transform the PPC750 CPU into a custom 3 core processor for the Wii U. If the partnership with Nvidia is healthy, it would be a very Nintendo thing to do by sticking with them. I find it hard to believe that it would break the bank for Nvidia to pair some ARM CPU cores with as many Volta GPU cores as possible within the power draw constraints of the platform. Nintendo doesn't prioritize having the most powerful hardware, so even if its a weaker SOC than Qualcom could offer, I doubt that would be a deal breaker for Nintendo. Many of their most successful platforms have had weaker processing power compared to the competition. Anything that exceeds base PS4 power will be sufficient for the Switch successor. It allows third parties the opportunity to rerelease PS4/X1 software on the Switch 2.

Not going to happen (ps4 level power)

Current switch is like 1/4 of xbox one fp32 perf docked and ps4 is 40% faster than xbox one so nintendo needs something like 6-8x perf of switch to be above ps4.

My realistic assumption would be 1.5-2.5x switch for switch pro, max. (Gpu, cpu power is easier as switch cpu is so fucking slow)

Nintendo isnt high-tech company(performance wise) so they just make something that is cheap to make and easy to sell for huge profits.
 
Nah. That would be the PS4Pro.

I was more thinking N64 to GameCube, GameCube to Wii, and Wii to WiiU. Or did Nintendo use 4 items and not 2 for the ducttaping?
 
Not going to happen (ps4 level power)

Current switch is like 1/4 of xbox one fp32 perf docked and ps4 is 40% faster than xbox one so nintendo needs something like 6-8x perf of switch to be above ps4.

My realistic assumption would be 1.5-2.5x switch for switch pro, max. (Gpu, cpu power is easier as switch cpu is so fucking slow)

Nintendo isnt high-tech company(performance wise) so they just make something that is cheap to make and easy to sell for huge profits.

I wasn't speaking about a Switch Pro, but the true successor that is at least 3 years away from release. By then, base PS4 capabilities will be within reach with a mobile SOC.
 
It's Nintendo, so it's going to be pretty conservative but at the same time you'd have to think they'd at least get close enough performance wise to XBO so it can get AAA ports from this gen.
 
Why would you think that? They've abandoned their console, it's a handheld+ now so think DS to 3DS.

Oh. In that situation I dont know about Switch 2, but I have the next next-one down. They'll add 3D and say you're holding it wrong.
 
I think it's being overthought here.
Just take the Snapdragon 855 performance from Galaxy S10 phones (Spring 2019) and stuff it into Switch 2 via Nvidia's Ampere architecture in 2022.

CPU:
A76 quad core @ 2GHz + A55 quad core @ 1.5GHz (for A57 performance).
GPU:
768 Cuda Cores @ 614MHz for 943GFLOPs portable and 1GHz for 1.536TFLOPs
96 Tensor cores for DLSS AI upscaling
RAM:
8GB LPDDR5 128bit for 100GB/s
Display:
1080p IGZO

Basically a 4x performance increase across the board, with DLSS they could target 540p and AI upscale it to 1080p, when docked they can target 720p and upscale to 1440p. At those rendering resolutions, it would be a next generation performance, but be a very cheap SoC with a low power consumption, likely below launch Switch, as the Snapdragon 855 only uses about ~5 watts during gaming in the Galaxy S 10.
 
The more I read from Nintendo's investor meetings, the further out I believe the Switch successor is. It sounds like they are doubling down on the current Switch, and are full steam ahead on developing new software for it. This is good news for current Switch users who have little interest in upgrading. We already know Zelda BoTW 2 is in development, as well as Metroid Prime 4, but I am starting to think that perhaps a new Mario Kart and Splatoon might be rolling out in the next couple years. Square Enix will be full steam ahead with new titles from their RPG Factory, and Nintendo could green light IP's like F-Zero, Wave Race, and 1080 to potentially make a return. A new 2D Zelda similar to Links Awakening could be in the cards. Nintendo is looking at the Switch knowing its not a direct competitor to the PlayStation and Xbox, Phil Spencer himself says this, and I believe they will sell the Switch for years to come on the back of its exclusive software.

After watching the DF video on Nvidia's DLSS, it certainly makes me think that could be part of a Switch Pro model, if it really exist. The ability to improve image quality on sub native resolutions would be a welcome feature to a new Switch model, and it wouldn't really create a new performance profile that developers would have to contend with.
 
The more I read from Nintendo's investor meetings, the further out I believe the Switch successor is. It sounds like they are doubling down on the current Switch, and are full steam ahead on developing new software for it.
Why would Nintendo want to rock the boat? The Switch has outsold the Xbox in under half the time it's been on the market, it's wildly successful. When you are this successful, it would be silly to make noises about this product's successor.

Your list of anticipated games - retreads of decades-old IP - fills me with dread. It's like Microsoft with Halo. I want Nintendo to do something new. :yep2:
 
Why would Nintendo want to rock the boat? The Switch has outsold the Xbox in under half the time it's been on the market, it's wildly successful. When you are this successful, it would be silly to make noises about this product's successor.

Your list of anticipated games - retreads of decades-old IP - fills me with dread. It's like Microsoft with Halo. I want Nintendo to do something new. :yep2:
Well, stuff like ARMS, LABO, Ring Fit Adventure, Astral Chain and of course less high profile stuff like Snipperclips and 1-2-Switch, are new stuff that they’ve introduced in less than three years. (And arguably Breath of the Wild was a completely new game set within the context of existing IP.) If we are complaining abuot lack of creativity in console space, I really don’t think that Nintendo is the publisher to single out.....
 
Back
Top