StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)

Given that several PS5-XSS only games are being hinted at for the system -- Flight Simulator 2024, Final Fantasy 7 Rebirth, Assassin's Creed Shadows, Gotham Knights, Metal Gear Solid Delta ... I have to think performance must be decent.

Maybe 7nm needs to be looked at again.
 
Basically now we just have to accept the basic premisse that Nintendo would not give us 30 minutes of battery life to arrive in 1 of 2 conclusions.

1. If it is 8nm, Nvidia made an engineering miracle and the Orin calculator was way off.

2.It is between 7-5nm.
 
Given that several PS5-XSS only games are being hinted at for the system -- Flight Simulator 2024, Final Fantasy 7 Rebirth, Assassin's Creed Shadows, Gotham Knights, Metal Gear Solid Delta ... I have to think performance must be decent.

Maybe 7nm needs to be looked at again.
True, Flight Sim was rumored to arrive to the Switch 2 a couple of days ago, if that's true it will clear all my concerns about the Cpu because that's a heavy cpu game
 
dang did I miss some entertainment porpoises?
giphy.gif

"entertainment porpoises"
 
After the launch we should have a new thread about figuring out the speeds and nodes. I'm sure someone will xray the thing and someone dev will leak out some numbers.
 
Well, I know for one that MS kind of jumped the gun on DX12, and ended up missing a couple really big features that came to Metal and Vulkan, for example, the subpasses concept and the function constants concept are both missing from DX12 and both could provide a decent performance uplift if used effectively (subpasses by allowing GPU work to be scheduled in larger and more tightly synchronized batches, and function constants by enabling more optimization during shader linking and side-stepping uniform registers and uniform buffer locking). At least DX12 got push constants which helps a lot but that covers a slightly different use case.

On the RT and Neural side of things, DF just touched on this, that Nvidia’s architecture actually allows for a lot more direct mixing between compute, RT, and neural processing within a given shader, but the APIs gated it off because other GPUs wouldn’t be able to conform to it easily, leaving the neural cores mostly only accessible through DLSS or through discrete workloads scheduled with CUDA or DirectML.

A custom API like NVN2 would have no such problem. In fact according to those at CES (including DF) who got the hands on demo of neural rendering, part of the magic there is a proposed expansion of DXR to fix this limitation, and that expansion of DXR is what enabled the “neural materials” demo, which is essentially embedding a ML model inline into the DXR material shader.

It's the very definition of working smarter and not harder because Nintendo will have to use every bit of resources to their advantage. I remember reading somewhere that DLSS only uses about 1% of the tensor cores or in some cases depending on the work load as much as 9% of the tensor cores. This has me thinking what about the remaining cores? If Nintendo develops a game using DLSS to upscale 540p to 1440p or even 4k what would the remaining cores be doing? Would they just be idling and sipping some power? . If the NVN2 api does have a different way of scheduling it's work load we can probably see RT effects taxing the console less than the traditional methods.
 
Honestly, when I read it I did think this was just a joke, because of people freely speculating about clocks and then adding +1 (why would it be 561MHz from all possible numbers).

I was going to make some smartass remark about how 561 is a prime number, and thus it is a “prime” clock speed, but turns out I am wrong. The closest prime number to 561 is 563.

So I’m going with 563Mhz. 😜
 
If these numbers are accurate, then my graphical expectations with the S2 have officially been reached!

9go24x.jpg

Not to question Thor, Son of Odin, but Switch 2 Portable mode will handily beat the Steam Deck.

Incidentally, the only advantage the PS4 and PS4 Pro has is literally just the teraflops. The Switch 2 is more modern and efficient in virtually every way:

+The Jaguar CPUs on the PS4/Pro were atrocious. The CPU on the Switch 2 alone runs circles around it.
+50% more RAM than what's available on either console. The Switch 2 has 12gigs of the newer LDRR5X RAM with a lean OS which means the RAM is more than the XSS and very close the PS5.
+UFS 3.1 internal storage + decompression hardware = near-instant loadtimes. The PS4/Pro uses a super slow HDD.
+The overall features baked into the GPU, not the least of which the tensor cores (enabling DLSS 3.5 sans Frame Generation), RT cores for lighting and audio, mesh shading, and so-forth blow away what the PS4 and Pro do. It's modern-day Nvidia and all it's industry-leading tech in a single box.


TLDR? It's extremely likely that the actual in-game experience of playing the Switch 2 will greatly outperform what the PS4/Pro can do, regardless of whatever the final numbers will be. It's a slaughter.
 
Last edited:
I think we may also be underestimating that top USB port. It's entirely possible Nintendo is more comfortable with higher power draws because the system is designed to have easier access to plug in power and external battery packs (maybe even an official one from Nintendo that is a snap on thing that doesn't need to be removed all the time).

That might be something that's being overlooked. I think the top USB port is mainly for power purposes (undocked). Sure they'll probably have a wacky add-on or two (camera?) for a game or two, but they didn't go bonkers with things like exotic, wild Joycons for the Switch 1 like people were predicting, I think that USB port is mainly for power plug in.
 
For a history of vitriolic comments against insiders, you have been banned for 3 days - MarcelRguez, Zellia, Phendrift
What clocks were people expecting when we first learned it was likely an 8nm node? Well below 560mhz it sounds like?
470 was the minimum that I think T239 could run so we saw a few posts saying 480-500ish. It's a good number. I doubt they'll have multiple handheld clocks this time because the 561 might be a good balance between performance and battery drain.
 
I have hard some concerns shared that the initial trailer may be entirely hardware focused, and show little to no gameplay. Even if that were to end up being true, once Switch 2 has been officially announced and shown in, third parties are going to start announcing and showing games. Personally, I expect a teaser trailer that does indeed focus on the hardware, but in a similar way to the 2016 teaser trailer for Switch. In the 2016 trailer, we saw Mario Odyssey, Splatoon 2, Skyrim and son on, but none of them were titled by name. I remember people were like, WTF, was that Skyrim?

We have gone round and round with clock speeds and nodes. I have basically settled on they are both good enough to make 1536 GPU cores and 8 CPU cores make sense for performance, battery life and cost. The GPU cores get a lot of attention as it relates to node and clock speeds, but a reduction in CPU cores to a cluster of 4 could have still been on the table if needed (obviously not needed). So regardless if this is 8nm and Nvidia found a way to make that work, or if we are looking at Samsung 5nm at a lower than typical density for reasons Z0m3le has laid out. All the passionate debating for one or the other is quickly coming to an end once we can see in what the system is actually capable of.
 
Dang we really had dug ourselves into a hole then. I think it might be time to start writing off that calculator before we start the 5nm hopium
We don't really need to be huffing too much hopium. The primary concern with 8nm was going to be the clocks. Assuming they at the very least have Switch V1 battery life, there's nothing else to worry about.
 
We don't really need to be huffing too much hopium. The primary concern with 8nm was going to be the clocks. Assuming they at the very least have Switch V1 battery life, there's nothing else to worry about.

There's no law that says they can't go a bit lower than Switch V1 anyway.

Like I'm sorry for people who are going to be really upset about that, but everyone and their grandma knows you're going to buy the Better Battery Switch 2 that comes out in 2027 or 2028 anyway. Nintendo knows it.
 
These clock numbers are pretty confusing (if true) but maybe Nvidia has an engineering marvel on their hands with this chip
There were, I think, rumours that they were "proud" of it, which could be true on an 8nm 561Mhz chip - extraordinary efficiency at a low cost, or an advanced Samsung node, bringing all this technology from different lines onto a new node to suit the usecase.
 
Not to question Thor, Son of Odin, but Switch 2 Portable mode will handily beat the Steam Deck.

Incidentally, the only advantage the PS4 and PS4 Pro has is literally just the teraflops. The Switch 2 is more modern and efficient in virtually every way:

+The Jaguar CPUs on the PS4/Pro were atrocious. The CPU on the Switch 2 alone runs circles around it.
+50% more RAM than what's available on either console. The Switch 2 has 12gigs of the newer LDRR5X RAM with a lean OS which means the RAM is more than the XSS and very close the PS5.
+UFS 3.1 internal storage + decompression hardware = near-instant loadtimes. The PS4/Pro uses a super slow HDD.
+The overall features baked into the GPU, not the least of which the tensor cores (enabling DLSS 3.5 sans Frame Generation), RT cores for lighting and audio, mesh shading, and so-forth blow away what the PS4 and Pro do. It's modern-day Nvidia and all it's industry-leading tech in a single box.


TLDR? It's extremely likely that the actual in-game experience of playing the Switch 2 will greatly outperform what the PS4/Pro can do, regardless of whatever the final numbers will be. It's a slaughter.
I hate the naming of DLSS in general but the Switch 2 will be capable of DLSS 4 all RTX cards will benefit form the new transformers that will power DLSS upscaling to improve image quality and temporal stability. We just won't see multi FG that is only dedicated to Blackwell cards and the regular FG found on ADA cards.
 
Maybe there's just one portable clock this time. I was always under the impression that the higher profiles existed because Nintendo wasn't thrilled with battery life but knew devs needed to squeeze whatever they could out of the Switch 1.

561 MHz is high enough that they don't need higher, and if it gets decent battery life anyway there's no need to make a lower one.
I hope you are correct, and I must say your explanation for how that could be the case makes sense to me.
 
I hate the naming of DLSS in general but the Switch 2 will be capable of DLSS 4 all RTX cards will benefit form the new transformers that will power DLSS upscaling to improve image quality and temporal stability. We just won't see multi FG that is only dedicated to Blackwell cards and the regular FG found on ADA cards.
I don't think we can say whether DLSS4 is realistic on T239 until we see how it performs on existing Ampere cards.
 
0
There were rumours that they were "proud" of it, which could be true on an 8nm 600Mhz chip - extraordinary efficiency at a low cost, or an advanced Samsung node, bringing all this technology from different lines onto a new node to suit the usecase.

It's also quite plausible that South Korean business paper, which is apparently a very mainstream outlet was correct on 7nm. Samsung is a massive company in South Korea, it doesn't really stretch it to think one of the bigger business papers would have a source there.
 
I have hard some concerns shared that the initial trailer may be entirely hardware focused, and show little to no gameplay. Even if that were to end up being true, once Switch 2 has been officially announced and shown in, third parties are going to start announcing and showing games. Personally, I expect a teaser trailer that does indeed focus on the hardware, but in a similar way to the 2016 teaser trailer for Switch. In the 2016 trailer, we saw Mario Odyssey, Splatoon 2, Skyrim and son on, but none of them were titled by name. I remember people were like, WTF, was that Skyrim?
Sort of offtopic, but Nintendo always proclaim themselves as an "Integrated hardware and software" company. There's no way they will reveal their new console with only hardware.

It might be like the Switch reveal footage, where the total game footage only added up to a few seconds. But it was there.
 
Maybe there's just one portable clock this time. I was always under the impression that the higher profiles existed because Nintendo wasn't thrilled with battery life but knew devs needed to squeeze whatever they could out of the Switch 1.

561 MHz is high enough that they don't need higher, and if it gets decent battery life anyway there's no need to make a lower one.
If first level clock gating really can cut power from unused GPU portions, one clock mode for handheld is all Switch 2 needs.
 
There's no law that says they can't go a bit lower than Switch V1 anyway.

Like I'm sorry for people who are going to be really upset about that, but everyone and their grandma knows you're going to buy the Better Battery Switch 2 that comes out in 2027 or 2028 anyway. Nintendo knows it.
I won't be buying a better battery. I actually don't know a single person who's bought a better battery for any of their V1 switches.
 
We are at the point that the only thing that could kill the excitement for me would be Switch 2 being region locked outside of BC.
 
0
I won't be buying a better battery. I actually don't know a single person who's bought a better battery for any of their V1 switches.

Well you bought a Switch 1 then regardless of fairly mediocre battery life. Nintendo knows they can get away with poor battery life. People will whine and moan about the battery number and then quietly pull out their wallet when they see Mario Kart 9, Mario 3D, Final Fantasy VII Rebirth, new Pokemon, Metroid Prime 4 in 4K 60, etc. etc.

The majority of Switch owners own one of the 16nm models too, the early adopting 20nm crowd is the minority.
 
It's also quite plausible that South Korean business paper, which is apparently a very mainstream outlet was correct on 7nm. Samsung is a massive company in South Korea, it doesn't really stretch it to think one of the bigger business papers would have a source there.
I'm not sure 7nm would give us the right efficiency at such numbers, but it could be a combination of a huge lift to a suitable but cheap node, THEN extensive work to get the performance per watt required out of it.
 
Last edited:
Please read this new, consolidated staff post before posting.
Last edited:


Back
Top Bottom