• Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
  • Furukawa Speaks! We discuss the announcement of the Nintendo Switch Successor and our June Direct Predictions on the new episode of the Famiboards Discussion Club! Check it out here!

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

The MX570 isn’t an SoC. It’s just a GPU. The multi core CPU, IO, and other system controllers are going to use a huge chunk of Dane’s die. IIRC, the GPU only accounted for a third of the X1’s size.
Assuming that the preliminary annotation of the press die shot of Orin is a mostly accurate representation of what the actual die shot for the Jetson AGX Orin probably looks like, the GPU seems to take up the most space.
NVIDIA_Orin_press_-_die_-_annotated.jpg
 
0
There’s also 5500mAh batteries that can be used. I believe that the 4310 in the switch was the only one like that at the time.

@Mercury_Sagit
“I'm aware, but at least the information gives us a ballpark since we have TX1 die shot (not sure about later Tegra SoC) and extrapolate from that the area allocated for GPU logic.”

Tegra Xavier if you’re curious:
nvidia_xavier_die_shot.png

Annotated:
900px-nvidia_xavier_die_shot_%28annotated%29.png



nvidia-xavier.png


AGX-die.jpg



According to this: https://en.wikichip.org/wiki/nvidia/microarchitectures/carmel#CPU_Complex_2

The 12nm Xavier has 8 CPU cores and 4 CPU Core complexes take up ~62.25 mm² die size area

GPU on the other hand takes up ~89.2 mm² of silicon area

So on the 8nm it should be smaller especially for the CPU. I’m not so sure with respect to the GPU though.
 
Last edited:
There’s also 5500mAh batteries that can be used. I believe that the 4310 in the switch was the only one like that at the time.
The single 5500 mAh batteries on the Xiaomi Mi Max 3 and the Doogee BL5500 Lite seem to be considerably larger than the OLED model's 4310 mAh battery.



And the Lenovo Legion Phone Duel 2 uses two 2750 mAh batteries to achieve a total battery capacity of 5500 mAh.
 
The single 5500 mAh batteries on the Xiaomi Mi Max 3 and the Doogee BL5500 Lite seem to be considerably larger than the OLED model's 4310 mAh battery.



And the Lenovo Legion Phone Duel 2 uses two 2750 mAh batteries to achieve a total battery capacity of 5500 mAh.

It should be noted that the switch is in fact a thicker device and has padding that helps the battery to be protected. If they do manage to make it slightly thicker it can increase the density to that.


Granted I think 5000mAh is the sweet spot.
 
It should be noted that the switch is in fact a thicker device and has padding that helps the battery to be protected. If they do manage to make it slightly thicker it can increase the density to that.


Granted I think 5000mAh is the sweet spot.
I should mention that I was referring to the size, not the thickness, when talking about the single 5500 mAh batteries used on the Xiaomi Mi Max 3 and the Doogee BL5500 Lite, since the batteries seem to have a very similar thickness to the OLED model's battery.

But I do agree that 5000 mAh is a sweet spot in terms of battery capacity, and is probably relatively inexpensive, especially since there are many smartphones that are using 5000 mAh batteries.
 
I think the discussing is MX570 being the closest GPU to Dane. But the spec sheet says 4GB of VRAM, I assume the Switch 2 GPU will have access to more than that, even in the worst case 8 GB configuration.

Two things that stand out
This has 1028 cores, i'm not sure we're expecting that many cores for Switch2(?)
25w TDP, which is probably too high so I assume we're expecting further reductions in performance to hit 15w (target) for them.


But VRAM is dedicated video RAM thats in the GPU card already, and doesn't count for on system ram. I don't think any of the Tegras have VRAM. 🤔

Also it's going to be really hard to imagine because Tegras also come with their ARM CPUs, while MX570 on laptops will likely come with AMD or Intel CPU processors as well.

i can totally see 1028 cores on switch though on what theoretically could be a 2 TFLOP machine at 2GHz max clocks. But I suspect Nividia and Nintendo will only reach like 80-90% in docked mode like switch for thermals. So 1.6-1.8 ampere like flops.

I still find it hard to assume 8nm for what would be the successor to ampere (Orion). It doesn't add up. Not that I doubt it would be 8nm, but that Nvidia would use 8nm for all their Orion products. it will be interesting to see if Orion really has any architectural efficiency gains over ampere as many have speculated here.
 
I don't think any of the Tegras have VRAM.
They do and don’t, because they are SoCs :p

It’s shared unified memory. That’s their VRAM. But also their system RAM.
I still find it hard to assume 8nm for what would be the successor to ampere (Orion). It doesn't add up. Not that I doubt it would be 8nm, but that Nvidia would use 8nm for all their Orion products. it will be interesting to see if Orion really has any architectural efficiency gains over ampere as many have speculated here.
I don’t see why, nVidia hasn’t split it like that before.

I should mention that I was referring to the size, not the thickness, when talking about the single 5500 mAh batteries used on the Xiaomi Mi Max 3 and the Doogee BL5500 Lite, since the batteries seem to have a very similar thickness to the OLED model's battery.

But I do agree that 5000 mAh is a sweet spot in terms of battery capacity, and is probably relatively inexpensive, especially since there are many smartphones that are using 5000 mAh batteries.
Now let’s hope they actually do go for 5000mAh :p

Though their track record shows they aren’t shy to increasing the battery density when push comes to shove.
 
But VRAM is dedicated video RAM thats in the GPU card already, and doesn't count for on system ram. I don't think any of the Tegras have VRAM. 🤔

Also it's going to be really hard to imagine because Tegras also come with their ARM CPUs, while MX570 on laptops will likely come with AMD or Intel CPU processors as well.

i can totally see 1028 cores on switch though on what theoretically could be a 2 TFLOP machine at 2GHz max clocks. But I suspect Nividia and Nintendo will only reach like 80-90% in docked mode like switch for thermals. So 1.6-1.8 ampere like flops.

I still find it hard to assume 8nm for what would be the successor to ampere (Orion). It doesn't add up. Not that I doubt it would be 8nm, but that Nvidia would use 8nm for all their Orion products. it will be interesting to see if Orion really has any architectural efficiency gains over ampere as many have speculated here.

1024 cuda cores @ 2Ghz would be 4Tflops by the way, so Nintendo wouldn't need the max clocks that high in order to hit 2Tflops...
The interesting thing to find out will be what transistor density are they getting for this MX570 since it also is suppose to have a max clock of 1155Mhz and kopite7kimi says its called GA107S. Can they achieve better yields on 8nm with this new mobile GA107 variant over the Ampere desktop cards and what correlation to Orin being capped at 1Ghz as well for the GPU does this have relevance to?
 
Any possibility that Nintendo goes with a smaller yet same density battery as the Switch has? They could probably use the extra space for a tiny bit bigger SoC?
 
0
i can totally see 1028 cores on switch though on what theoretically could be a 2 TFLOP machine at 2GHz max clocks. But I suspect Nividia and Nintendo will only reach like 80-90% in docked mode like switch for thermals. So 1.6-1.8 ampere like flops.

I still find it hard to assume 8nm for what would be the successor to ampere (Orion). It doesn't add up. Not that I doubt it would be 8nm, but that Nvidia would use 8nm for all their Orion products. it will be interesting to see if Orion really has any architectural efficiency gains over ampere as many have speculated here.
There are 128 CUDA cores per SM for GA102, so 1024 CUDA cores for 8 SMs, not 1028 CUDA cores. (GA100 has 64 CUDA cores per SM.)

I imagine Orin uses the Ampere architecture as the base, and then add Lovelace features, such as AV1 encoding (consumer Ampere GPUs only support AV1 decoding), etc. And I believe that's what Nvidia did with the Tegra X1, which is basically use the Maxwell architecture as the base, and then add Pascal features, such as FP16 operations, etc.

Are they still using the 3ds battery for the pro controller?
Yes.

The interesting thing to find out will be what transistor density are they getting for this MX570 since it also is suppose to have a max clock of 1155Mhz and kopite7kimi says its called GA107S. Can they achieve better yields on 8nm with this new mobile GA107 variant over the Ampere desktop cards and what correlation to Orin being capped at 1Ghz as well for the GPU does this have relevance to?
There's a good chance nobody's going to have enough information to calculate GA107's transistor density since the die size alone is not enough information to calculate GA107's transistor density, since the number of transistors in GA107 is also needed to calculate GA107's transistor density, which probably requires the destruction of the die.

And I imagine little relevance since Orin's GPU's probably considerably smaller compared to GA107, considering Orin's a SoC and GA107's simply a GPU.
 
0
I don't seem to understand why the Switch 2 couldn't have a TDP higher than 15 watts in docked mode maybe around 25 to 30. If they wanted to do so couldn't they just design a new dock that have external fans to cool down the switch to manage the extra heat? Couldn't the switch 2 have more open areas for air intake like around the charging port at the bottom
 
Last edited:
I don't seem to understand why the Switch 2 couldn't have a TDP higher than 15 watts in docked mode maybe around 25 to 30. If they wanted to do so couldn't they just design a new dock that have external fans to cool down the switch to manage the extra heat? Couldn't the switch 2 have more open areas for air intake like around the charging port at the bottom
I’ve mentioned this before, but essentially you can do active cooling (funneling heat away from sources via heat pipes and creating a vacuum effect to push it out with fans) or you can do passive cooling (constant drawing air in and out to do circulative cooling), but you can’t effectively do both for a single SoC design. And between the two, active cooling is more efficient for this form factor.

That being said, they CAN go a bit higher on the TDP, so long as the thermals can accommodate the target TDP in both modes without massive battery drain, obnoxious levels of fan noise or a size increase to accommodate a larger blower for heat management.
The highest you would ever likely see for that to be the case is somewhere between 10-20W TDP when docked. Higher than that and at least one of those three negative factors I mentioned becomes unavoidable.
 
Based on anantech, OG 20nm X1 Switch docked and not charging joycons or battery is only drawing 11w. I think there's a lot of room here, even just hitting 15w would be nearly a 40% increase.
 
0
Do you think this is realistic information? 2024 is a long way off. 👀


That's just not how SoC development works for something like this

(Not to mention it is pretty much against all of the reliable rumours we've had from people with highly accurate track records)
 
Do you think this is realistic information? 2024 is a long way off. 👀


Samus Hunter is not a reliable source and its our own @NateDrake who initially reported kits going out way back in 20220, so I'd wait for him to comment on just a radical and weird speculation, over some rando clout chaser on twitrter.
 
We use the 15W TDP as the base expectance for the device in docked mode, as we know Nintendo won't clock it really high and were really pushing it with the OG switch that was already hitting the ceiling as is. The Tegra X1 had thermal issues as a result of the node it was on and the CPU cores it has were inefficient on top of that. Samsung 8nm does not suffer from the same thermal issues as the TX1 and neither does the A78 compared to the A57. Both are dramatically more advanced and efficient for their roles. One being denser the other being more performant.

+20W in docked mode and you likely run into issues where the fan cannot keep up with the cooling and over time you will need to replace that tiny fan as it would be pushed closer to the limit. There is a possibility of throttling here and should be taken into account which would mess with performance. There's also a battery to consider, as those do get affected by heat, and the OLED display too.


ORIN NX has as low of 10W and a max of 25W in TDP, but it also has a bunch of other elements, and that's with a CPU clocked as high as 2GHz and a GPU as high as 1GHz. The GPU is likely what makes up most of the power consumption of the the device, so clocking it say to around the same as the switch in docked and portable could perhaps bring power draw down by a few watts in docked and portable mode. This includes the RAM which goes to 3200MHz.

In docked mode we have to account for other things too that consume power. So, Wifi, BT, thermal detector, things like that. They are likely small, but point being that they consume a power draw.

Oh and the fan also consumes power, can't forget that

NV has this with respect to ORIN AGX, thus it also applies to ORIN NX:
"Like other Jetson Modules, Jetson AGX Orin is designed with a high efficiency Power Management Integrated Circuit (PMIC), voltage regulators, and power tree to optimize power efficiency. Jetson AGX Orin supports three optimized power budgets: 15W, 30W, and 50W. Each power mode caps various component frequencies, and the number of online CPU, GPU TPC, DLA, and PVA cores. Customers can leverage the nvpmodel in Jetson Linux to use one of these pre-optimized power modes or to customize their own power mode within the design constraints provided in our documentation."

source: https://www.nvidia.com/content/dam/...in/nvidia-jetson-agx-orin-technical-brief.pdf


I'm of the few that think they should go crazier with the CPU clock speed and remain modest with the GPU Clock speed as this will gain more GPU cores to handle a task, but CPU can determine the level of support in the longterm IMO. Engine support is crucial for such a device. It likely already supports everything XB1 supports engine wise for the hardware. So keeping the same clocks that the current switch has but for the GPU, and while increasing the amount of CPU cores, they also increase the CPU clock speed, just not over 2GHz since it seems like the cap.


Engines such as Unity, UE4+5, Frostbite, RAGE, CryEngine, RE Engine, Luminous Engine, Dawn Engine, etc., can allow for games of old to be brought to the device. Those that will continue supporting these engines in use would be able to include this Dane model in their SKUs, perhaps.

Also, the CPU cores would consume a lot less than the GPU would, docked or not, and would be locked to a frequency, GPU and RAM speed would be fluid and change between docked and handheld mode.

Dakhil and me spoke earlier about hoping for a larger battery as it lets them stretch their legs a little more, and 16% more battery allows them to add extra time to what they can work with.



I think, 5-9W portable and 10-20W docked is reasonably what I would want, but I do expect 15W if it can't reach it without issues cropping up. Like throttling.
 
Last edited:
I honestly didn't know if she was trustworthy. I saw the retweet, that she was being followed by several people and thought that new information might have "arrived".

I will admit that 2024, as a release year, kind of scares me 😅. I hope she's completely wrong.
 
I honestly didn't know if she was trustworthy. I saw the retweet, that she was being followed by several people and thought that new information might have "arrived".

I will admit that 2024, as a release year, kind of scares me 😅. I hope she's completely wrong.
Like, 2024 could be when a Switch 2+ or Switch 2 Lite comes out considering a 2022 release date is looking more and more likely for the base Switch 2?

Following Nintendo's 2 year release for new models for the OG Switch line
 
I honestly didn't know if she was trustworthy. I saw the retweet, that she was being followed by several people and thought that new information might have "arrived".

I will admit that 2024, as a release year, kind of scares me 😅. I hope she's completely wrong.
They aren't reliable whatsoever. They're not even a leaker or an insider, they are an attention seeker. They are like Zippo tier lol
 
Do you think this is realistic information? 2024 is a long way off. 👀


The reports we've received about developers having dev kits say that the games they're working on are aiming for a 2022 release. Samushunter saying 2024 launch flies in the face of that.
 
I honestly didn't know if she was trustworthy. I saw the retweet, that she was being followed by several people and thought that new information might have "arrived".

I will admit that 2024, as a release year, kind of scares me 😅. I hope she's completely wrong.
if SamusHunter says the sky is blue, doubt them
 
Samus Hunter is not a reliable source and its our own @NateDrake who initially reported kits going out way back in 20220, so I'd wait for him to comment on just a radical and weird speculation, over some rando clout chaser on twitrter.
Devs give feedback on early devkits and Nintendo can adjust some things here & there -- RAM/clocks if the adjustment doesn't have a negative impact. You don't send early prototype hardware to developers years in advance. Developers are actively working with new devkit hardware & have a window for their games to reach completion.

Nikkei and Bloomberg haven't been reporting on prototype hardware that devs got to test for a few weeks.
 
Bare in mind Takashi Mochizuki said a Switch Pro would launch in 2019 and again in 2021 and was wrong both times
Did he actually say there would be a pro in 2019 though? There was an upgraded Switch model in 2019. As far as I know its never actually been disproven that Mariko switches are using igzo display tech.
 
Bare in mind Takashi Mochizuki said a Switch Pro would launch in 2019 and again in 2021 and was wrong both times
No. I suspect that he and the Bloomberg pubilcation didn't actually do much wrong. However, a lot of people made the mistake of conflating what they read about the next Generation of Nintendo systems with the idea of a "Pro" Switch (which was NEVER alluded to by Nintendo, quite rightly, and which was never going to be thing for all sorts of sensible, practical reasons). The publication has been spot on for the main part.
 
No. I suspect that he and the Bloomberg pubilcation didn't actually do much wrong. However, a lot of people made the mistake of conflating what they read about the next Generation of Nintendo systems with the idea of a "Pro" Switch (which was NEVER alluded to by Nintendo, quite rightly, and which was never going to be thing for all sorts of sensible, practical reasons). The publication has been spot on for the main part.
They specifically said the DLSS system would be announced when the Oled was actually announced. So they did get that one wrong or conflated.

I don't think they did anything wrong in 2019 though.
 
They specifically said the DLSS system would be announced when the Oled was actually announced. So they did get that one wrong or conflated.

I don't think they did anything wrong in 2019 though.
In 2019 they said there would be a smaller version (the Lite) in addition to a new upgraded version geared towards "enthusiasts".

They were clearly referring to the Mariko revision and they likely assumed since it was more advanced it would be more powerful and marketed as such. Their info they get from sources is spot on but the assumptions they make based on that info have been shown to be flawed a few times.
 
They specifically said the DLSS system would be announced when the Oled was actually announced. So they did get that one wrong or conflated.

I don't think they did anything wrong in 2019 though.
Clearly there was a lot of conflation, although I suspect they have a heads-up on development for the next Generation of Nintendo systems. It was obvious to me that DLSS was for the next system, because a "Pro" system would've been a hard sell, a botching of the Switch brand, AND a developer's nightmare. Nintendo also told us they were halfway through the Switch cycle last year, and there's something in taking that at face value, because it was from the horse's mouth. All the more so because they're not going to tell such a lie to investors, no matter the level of "Because Nintendo" some places want to peddle.
 
Bare in mind Takashi Mochizuki said a Switch Pro would launch in 2019 and again in 2021 and was wrong both times
Takashi Mochizuki only said in 2019 that "one version will have enhanced features targeted at avid videogamers, although it won’t be as powerful as Sony Corp.'s PlayStation 4 Pro or Microsoft Corp.'s Xbox One X, according to parts suppliers and software developers for Nintendo who have access to a prototype of the machine", which seems to describe the Nintendo Switch (2019), considering that the Tegra X1+ used was fabricated using TSMC's 16FF process node instead of TSMC's 20 nm** process node for the Tegra X1 used on the Nintendo Switch (2017).

In 2019 they said there would be a smaller version (the Lite) in addition to a new upgraded version geared towards "enthusiasts".

They were clearly referring to the Mariko revision and they likely assumed since it was more advanced it would be more powerful and marketed as such. Their info they get from sources is spot on but the assumptions they make based on that info have been shown to be flawed a few times.
That's what Hermii said in the second paragraph.

Hermii's referring to the article Takashi Mochizuki wrote for Bloomberg on 22 March 2021 for the first paragraph.
 
Takashi Mochizuki only said in 2019 that "one version will have enhanced features targeted at avid videogamers, although it won’t be as powerful as Sony Corp.'s PlayStation 4 Pro or Microsoft Corp.'s Xbox One X, according to parts suppliers and software developers for Nintendo who have access to a prototype of the machine", which seems to describe the Nintendo Switch (2019), considering that the Tegra X1+ used was fabricated using TSMC's 16FF process node instead of TSMC's 20 nm** process node for the Tegra X1 used on the Nintendo Switch (2017).


That's what Hermii said in the second paragraph.

Hermii's referring to the article Takashi Mochizuki wrote for Bloomberg on 22 March 2021 for the first paragraph.
My point is that they were indeed wrong about how it would be positioned, even if they were right about the hardware itself.
 
0
As I've said in the past, no developers were aware of the OLED revision for 2021. Devkits for the OLED only went out a bit prior to launch, as they are not necessary for development, though the extra RAM helps with debugging. When you talk with suppliers and developers, it's easy to see how OLED and Switch 4K became conflated. One is saying they are preparing production for new hardware and the other is prepping software for new hardware. When the latter has no devkit for the revision, it's easy to believe the two were one in the same.
 
Curious question, what are the odds that Dane will have a OLED screen? Is it possible they would go back to a LCD screen to lower the costs?
 
Curious question, what are the odds that Dane will have a OLED screen? Is it possible they would go back to a LCD screen to lower the costs?
If OLED model supply in Japan is any indication after its rocky start, I’d personally say it’s pretty likely to continue on with an OLED screen.

But I’ve also had the sneaking suspicion that OLED model was designed with parts they intended to order for Dane anyways and started early to work out production hiccups and work towards higher volumes of these parts with a model that they could afford shortages and hiccups with as opposed to at Dane launch.
 
But I’ve also had the sneaking suspicion that OLED model was designed with parts they intended to order for Dane anyways and started early to work out production hiccups and work towards higher volumes of these parts with a model that they could afford shortages and hiccups with as opposed to at Dane launch.

totally my view on it. When you look at the V2 changes, they were minor compared to the OLED changes. With the OLED being slimmer I think this is a bit too much redesign just for the better tabletop stand.
 
If OLED model supply in Japan is any indication after its rocky start, I’d personally say it’s pretty likely to continue on with an OLED screen.

But I’ve also had the sneaking suspicion that OLED model was designed with parts they intended to order for Dane anyways and started early to work out production hiccups and work towards higher volumes of these parts with a model that they could afford shortages and hiccups with as opposed to at Dane launch.
Yep that's more or less my theory too.
 
0
with the OLED model having a different body, it does ask how will Dane look visually distinct from it
I really would not be surprised if Dane has the same chassis. Nintendo often saying things like "the Switch is halfway through its lifespan" makes me feel like they'll try to position and visually design Dane as "just another Switch" as opposed to making it look distinctly different from the Mariko, so they can carry on like it's the "same system" for sales numbers sake.
 
with the OLED model having a different body, it does ask how will Dane look visually distinct from it
New controllers. That's by far the best way to make it visually distinct and very in line with Nintendo, which is always looking into new forms of input.
 
0
I really would not be surprised if Dane has the same chassis. Nintendo often saying things like "the Switch is halfway through its lifespan" makes me feel like they'll try to position and visually design Dane as "just another Switch" as opposed to making it look distinctly different from the Mariko, so they can carry on like it's the "same system" for sales numbers sake.
if they're gonna do that, they might as well cut production of the Mariko hybrids. would offer a seemless transition for new buyers, though it would complicate things for potential upgraders whom aren't into spec talk. that's what a new marketing campaign would be for, but if Nintendo games keep coming to Mariko, they can hold back on it
 
if they're gonna do that, they might as well cut production of the Mariko hybrids. would offer a seemless transition for new buyers, though it would complicate things for potential upgraders whom aren't into spec talk. that's what a new marketing campaign would be for, but if Nintendo games keep coming to Mariko, they can hold back on it
That's aaalmost what I expect. I see Nintendo discontinuing the standard hybrid Mariko at least, but putting the OLED in its price slot as the new "standard" system. Dane will then be the new "premium" model, and the Lite will basically be the same as an entry-level model until they're ready for the Dane SoC to be the new standard across the board.
 
0
Please read this new, consolidated staff post before posting.

Furthermore, according to this follow-up post, all off-topic chat will be moderated.
Last edited by a moderator:


Back
Top Bottom