• 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.

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)

Is the "real" question

"Would Nintendo just add dock-cooling for good performance bumps using 8nm?"

or is it

"Would Nintendo 'gimp' portable mode this much using 8nm, while beefing dock play up way too much?"
to go along with this why increase the cooling considerably for docked if it's 5nm (4N)? maybe some of our technical guys can chime in but could heat be an issue at the higher clocks if the chip is on a smaller process.

i'm not arguing for 8nm but it seems baffling if the cooling system had to be souped up to accomodate a bigger, hotter chip. the cost of doing this surely takes it far beyond what it would for using a smaller node to begin with.

2 scenarios - it's 4N and Nintendo is going balls out to guarantee maximum potential performance in docked. this would be amazing.

8nm because the chip was finalized such a long time ago and the process can't be changed. improved cooling solution added later as way to squeeze more out of it in docked.
 
Last edited:
If that's the case then that's perfect!
Yeah, MicroSD Express is about 800MB/s - 1000MB/s, which is close enough to allow similar performance. Old SD cards are more like 80 MB/s, so clearly in a different league. Micro SD Express cards will start to enter the mass market in this calendar year, so there should be some options available by the time Switch 2 launches.
 
You can say that.


Zen 2 cores seem to be beat it in single core performance by a good amount, or at least the A78s in Orion. While multi core is better on A78s. Will be interesting to see how A78cs perform.


Best case scenario for us I think is 2GHz CPU cores. I dunno if they can do that with a 1.6 handheld 4 docked GPU TFLOPs. Safe to say current gen x series and PS5 will be at least twice as fast in CPI

Alright fine...

DBZ Cell Saga layman terms

SSJ2 Gohan: PS5 Pro
Super Perfect Cell: X Series X
Cell Jr: Xbox Series S
Ascended SSJ Goku: Switch 2
Ascended Vegeta: Xbone X
Ascended Trunks: PS4 Pro
Piccolo: PS4
Android 16: Switch

I am not sure if i should me impressed or terrified that the DBZ scale kinda makes sense and how we can still use it to compare consoles horsepowers since the Wii U days :ROFLMAO:
 
But his source is Famiboards...

Hot take:

If any reports by major outlets like Eurogamer pop up today, there's generally a fat chance that Famiboards will be the source.

Ideally, they would add corroboration from their own sources, though.

But i feel like these infos are too "big" to not report on them, even without having that corroboration.
 
actually no thats entirely my doing. thraktor estimated the SEC5nm formula as 73% of 4132, so i wondered what 73% of 3016 would be, which gives me 2201. its a little more than half the power consumption of SEC8nm, which is just a silly lil "what if" for something lower than SEC5nm. i didnt know what to call it and i dont want to say "this is my TSMC4nm" cus eh its nowhere near as well thought out as thraktors work

Hehe I was initially going to call it TSMC N4, because the curve made sense through some extrapolation when I looked at some relative comparisons between ampere and ada.

Look i might be totally wrong here, but i think the current SD Express cards can somewhat match the internal storage speed?

Yup, moreover with the constrained memory bandwidth you'd likely not even need such high read (/write) speeds.
FDE + IO API with an external storage medium whose theoretical sequential read speeds can top out at 800MB/s (as per samsung's press release). Thus micro SD Express seems like a natural solution for the switch (the physical dimensions are likey not as different compared to microSD). I think it's more than sufficient for internal and external storage speed parity.

The PS5 very likely does not use all the bandwidth it can use, so in a sense it's overprovisioned to allow for features such as downloading games while playing, store recording while playing, but also for future parity with the development of an custom i/o unit, api and for developers to have a breadth of bandwidth to capitalize on high asset quality, audio etc.
From most PC games with Directstorage it's already been investigated how much raw bandwidth makes a difference.



Necrolipe confirms Moji as the Switch 2 codename


Very odd to present yourself as a source of confirmation, when you've never dropped any hints or comments on it 😂
 
to go along with this why increase the cooling considerably for docked if it's 5nm (4N)? maybe some of our technical guys can chime in but could heat be an issue at the higher clocks if the chip is on a smaller process.

i'm not arguing for 8nm but it seems baffling if the cooling system had to be souped up to accomodate a bigger, hotter chip. the cost of doing this surely takes it far beyond what it would for just using a smaller node to begin with.

2 scenarios - it's 4N and Nintendo is going balls out to guarantee maximum potential performance in docked. this would be amazing.

8nm because the chip was finalized such a long time ago and the process can't be changed. improved cooling solution added later as way to squeeze more of the chip in docked.

If 1.32GHz on 8nm has a power draw of ~35W (going by the Jetson Tool), then 4N would be half of that? 17.5W for the GPU only... Maybe a fan on the dock would still be useful, even if they went with 4N.
 
the demo sounded impressive but devils advocate it could still reach impressive performance levels on 8nm in docked with the improved cooling system (lets presume the 'target hardware' was maxed out). we don't know what resolution the Matrix demo was running at, what the base resolution was (if DLSS was employed) etc. lots of variables vs a hard number if we knew the final clocks.
The 1024 core Orin uses 25 watts running at 918 MHz , with 1.6 GHz memory clock.
Don't see how it could be 8nm.

source:
https://www.techpowerup.com/gpu-specs/jetson-orin-nx-16-gb.c4086
 
If 1.32GHz on 8nm has a power draw of ~35W (going by the Jetson Tool), then 4N would be half of that? 17.5W for the GPU only... Maybe a fan on the dock would still be useful, even if they went with 4N.
i don't know the numbers but this sounds like more than Mariko, though the Jetson tool doesn't include any efficiency savings in the custom T239. i'm genuinely curious about the second fan thing and whether it points either way to 8nm or the smaller node (or maybe something else) or it's just inconclusive.
 
Fun mind cinema:

When Nintendo provided near-final or final dev-kits to Capcom along with the finalized tech details of the retails units, they hopefully added a personal note to the package reading:

"Here, go WILDS with that ... oh and don't ask us for anything ever again."

(This is of course just a joke scenario.)

Wilds is the one game I'm really hoping and praying gets a Switcheroo version. Please please please. Rise/Sunbreak was great and all and I'm sure a new title like it made specifically with the successor in mind would be just as good, but it sucked so bad to see the MH community split apart like that last gen.
 
The 1024 core Orin uses 25 watts running at 918 MHz , with 1.6 GHz memory clock.
Don't see how it could be 8nm.

source:
https://www.techpowerup.com/gpu-specs/jetson-orin-nx-16-gb.c4086
And this is without all the other chips on the board of a Switch console because it's just a computing board, not a whole system. Doing this with 1536 Cuda Cores would need more wattage than the Wii U and wouldn't even work for the docked console in my opinion.
 
Hehe I was initially going to call it TSMC N4, because the curve made sense through some extrapolation when I looked at some relative comparisons between ampere and ada.



Yup, moreover with the constrained memory bandwidth you'd likely not even need such high read (/write) speeds.
FDE + IO API with an external storage medium whose theoretical sequential read speeds can top out at 800MB/s (as per samsung's press release). Thus micro SD Express seems like a natural solution for the switch (the physical dimensions are likey not as different compared to microSD). I think it's more than sufficient for internal and external storage speed parity.

The PS5 very likely does not use all the bandwidth it can use, so in a sense it's overprovisioned to allow for features such as downloading games while playing, store recording while playing, but also for future parity with the development of an custom i/o unit, api and for developers to have a breadth of bandwidth to capitalize on high asset quality, audio etc.
From most PC games with Directstorage it's already been investigated how much raw bandwidth makes a difference.



Very odd to present yourself as a source of confirmation, when you've never dropped any hints or comments on it 😂
??????

Necrolipe is a famiboard user. And both here on this forum and on his social networks when rumors of the alleged codename have surfaced, he has been confirming that they are fake because he knew the real codename.


 
??????

Necrolipe is a famiboard user. And both here on this forum and on his social networks when rumors of the alleged codename have surfaced, he has been confirming that they are fake because he knew the real codename.




Oh well this is embarrassing, I just put 1 and 1 together haha. I never looked at the twitter handle 😵 .
I'm not as tuned in with twitter, so I thought this was a different person that the one that's in the thread.
Mb 🫡
 
Finally caught up from the beginning of this ride to my post from yesterday. (Btw. the spelling mistakes in those are a reason why i usually only participate here when i'm im front of a keyboard, and not via mobile phone...)

Again for the 8nm vs 4nm situation, see it this way ... we learned from the findings of the threads detectives that the situation looks better in some parts than even the regulars here were speculating / hoping for, or at least matching the speculations the folks would think are "good".

Wouldn't it be strange to have Nintendo really "go all out" in so many things, but then totally skimping on the node?

The rumored/speculated dimensions of the device, the things from above ... i'd say theres a lot more hints / points for 4nm than there are for 8nm.
I know i wrote that earlier this week already, but now there's even more hints / points imo.

Wilds is the one game I'm really hoping and praying gets a Switcheroo version. Please please please. Rise/Sunbreak was great and all and I'm sure a new title like it made specifically with the successor in mind would be just as good, but it sucked so bad to see the MH community split apart like that last gen.

Be more bold, hope for it to be a launch game! ^^
 
wasn't there another (8nm) Ampere GPU posted recently (a laptop chip) that used surprisingly little power, max of 15W if i remember correct. enough that it made 8nm seem possible.

MX570 I believe. TDP of 15W, but it has 16 SM and GDDR6. I just don't know if that TDP is real. It's completely different from the power draw we got from the Jetson tool.
 
The English translation is interesting, particular this question.

Q: President Furakawa, you have commented previously on how Nintendo is always looking for new ways to use technology to bring new and innovative gameplay to your fans. Can you reveal how you are going to continue this tradition with the switch successor?

A:
magnets.jpg
 
It doesn't seem clear that 8nm actually is cheaper per completed SOC, considering the volume advantage of 4N, with more SOCs per wafer.

Not to mention the potential extra battery/coming costs that might come with 8nm.

I have my doubts about this, considering there is a 30% raster performance gap between both switch2 and steam deck in portable mode, I don't think docking will allow drake to have raster performance that is comparable to ps4pro, there is a specification gap here on a physical level.

Steam Deck uses RDNA2 which is way beyond Polaris. Also I think the raster advantage of RDNA2 comes partly from Infinity Cache which the Deck and consoles do not have.
 
The English translation is interesting, particular this question.

Q: President Furakawa, you have commented previously on how Nintendo is always looking for new ways to use technology to bring new and innovative gameplay to your fans. Can you reveal how you are going to continue this tradition with the switch successor?

A:
magnets.jpg
I can’t see the picture.
 
Is "Muji" really the proper spelling of this codename? Because it kind of... means... a couple of things in japanese that I'm not sure Nintendo would want associated with the console unless they were having a laugh in the office.
 
Is "Muji" really the proper spelling of this codename? Because it kind of... means... a couple of things in japanese that I'm not sure Nintendo would want associated with the console unless they were having a laugh in the office.
The only meaning of Muji in Japanese that I know of is "无印", a word that has appeared in a number of Japanese game series, meaning "original".
 
Muji can be used to indicate the number 610(or 62) using goroawase(japanese wordplay with numbers) but i can't imagine the meaning of.
 
Man, when I look at those specs I really can't see the Switch 2 being 399. Much higher price point.
SD is 399, has 16 gb memory, and valve would get worse prices than Nintendo cause they're a smaller player.

It also doesn't have separate controllers or a dock though.
 
I've been lurking in the shadows for ages, congrats on the debunked details about the Switch 2, made my day!

Do we have any information regarding the battery?
 
Yeah, MicroSD Express is about 800MB/s - 1000MB/s, which is close enough to allow similar performance. Old SD cards are more like 80 MB/s, so clearly in a different league. Micro SD Express cards will start to enter the mass market in this calendar year, so there should be some options available by the time Switch 2 launches.
Currently
 
I've been lurking in the shadows for ages, congrats on the debunked details about the Switch 2, made my day!

Do we have any information regarding the battery?

Might check the first post of this thread, maybe there's a link when there has been news.

Personally, i don't remember any info about battery, but i think as a general rule of thumb Nintendo would want at least the same playtime estimates as OG launch Switch had.
 
Man, when I look at those specs I really can't see the Switch 2 being 399. Much higher price point.
Switch 1 launch price adjusted for inflation would suggest a $380,- price. That price gets you an LCD screen at 8 inch and 1080p, a chip that has been in a finalised state since somrthing like 2022, a bog-standard RAM specification, and a storage type that has been commonly used in phone for several years as well. As such, I don't think $399,- is that unlikely. I could see maybe $449,- happen as well, but I don't think the price will go "much higher", personally.

One thing that might skew the perception is that from WiiU to Switch, the absolute change in performance was not quite a full generational leap because of the vastly diminished size of the device and the necessity of it being portable. This time around, however, we are looking at a full-fat generational leap in performance, and changes like the ones we are seeing are to be expected with that. This combines with the massive change in software-hardware combined rendering approaches (ray tracing and DLSS) that allow for even better results with the same silicon. So it's a great moment for Switch 2 to showcase the potential of its form factor without going into ultra-premium price territory.
 
Thank you, Famiboards.

At the end of the day, it doesn't matter where the info ends up, who makes a video about it (and comes to outrageous conclusions), etc.

You found it all first, and your continued insight keeps my interest in gaming alive when I thought it way dying.

Seriously, thank you.
 
Yup true. I also didn't want gimmick in Switch 2, just power up all things from Switch 1. But in Switch 3 or new concept then Nintendo can start again their innovative gimmick
Not gonna lie, I wanted a gimmick.
But the president’s words from the latest investors meeting and now this latest leak are the final nails in the coffin for #teamgimmick.
 
It hasn’t to do that, Switch 2 has more than enough advantages against it to top PS4 Pro (pre DLSS) on most Gen 8 games.
What are these great number of advantages that it has that would have it outperform the PS4 Pro before DLSS? Is it low level API? Developer support? We removed DLSS from the picture since it’s pre-DLSS, so that means the tensor cores aren’t being used for their intended purpose, what are the updated features of T239 that would have it outperform the PS4 Pro?


Not tile based rendering as you’d said that already, so I’m assuming it’s something else. The difference between ampere and the GCN Frankenstein hybrid that exists in the PlayStation 4 Pro is about four years or so of change.

It has advantages*, but now I’m curious on what the great number of advantages are to top the PS4 Pro, and since the focus is of gen 8 games I’m also assuming it’s not even using ray tracing atp.




*I don’t think I have enough space to detail how many advantages exist, I already know it but this post is removing the aforementioned major advantages and now it has me curious if I missed some aspects of it.
 
Please read this staff post before posting.

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


Back
Top Bottom