• 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.
  • Do you have audio editing experience and want to help out with the Famiboards Discussion Club Podcast? If so, we're looking for help and would love to have you on the team! Just let us know in the Podcast Thread if you are interested!

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

Not necessarily. Nintendo can limit the frequency to match the og switch for bc. The Wii U did not have an improved Wii mode that could have fixed games like The Last Story.

As much as I’d love that to be wrong, it may end up being the case indeed, like most 3DS games on New 3DS (unfortunately). Well, I guess?

I don’t think there’s been much discussion on this specific topic, but I guess we’d need to know more about Nintendo’s plan for BC on a technical level.

Sadly I don’t have much technical insight, but I’d love some kind of confirmation and reassurance on this from some folks here.
Having a proper BC but without any improvements just like it was on New 3DS and Wii U would be a huge bummer, and even more so if we don’t get patches for the most notable games.
There are fairly specific preconditions for Nintendo's historical approach to BC that, as far as we can tell at this point, aren't met with Drake. It only really works when the system either includes processors with identical architecture to the original system, or something that can produce a very close facsimile thereof. Drake cannot provide that. There'd have to be a separate chip, which is highly unlikely.
 
Nate's main comment on its capability has always been DLSS capability (and what it implies). He's also occasionally mentioned A78, IIRC. So if there were comments about a 'modest' upgrade, they're not from Nate.
That’s a good point. And we also need to remember that a few weeks back he replied to one of Polygon’s posts discussing devkits with something along the lines of “not going to confirm specifics but a lot of this generally sounds like what I’ve heard partners say” (definitely me paraphrasing here).

(This wasn’t Polygon’s most recent informational post discussing 6gb RAM -> 12gb, but one before that)
 
Nate is correct that it alone does not imply mass production has started for a launch next spring. Not that it isn’t launching then. Which is fair because Mochizuki would have had an article out if that were the case.
Yeah we just know now that the SoC is done and can be mass produced. Switch's successor could be in production right now, but we have no proof that it is, just proof that all intended components are ready for mass production.
 
This is maybe a weird question, but what would a “withered technology” Drake look like?
Nintendo Switch - OLED Model /srs
This is true. 720p OLED panels were almost exactly like the GameBoy, a cheap panel that was mass manufactured but not really going anywhere.
I don't really think there is a way forward like that anymore. "Lateral thinking with withered technology" (which, according to the man who said it, Gunpei Yokoi, involves novel new uses for mature, cheap-to-produce and well understood technologies) doesn't work when mature technologies have had all the novelty wrung out of them; we have entered a period in technological development (in the entertainment industry especially)
Eh, I don’t actually buy this. It’s probably the end of doing so for graphics processing, but the whole existence of modern VR is “withered” technology when Palmer Lucky realized he could use a battery of low granularity cheap phone accelerometers to infer data ordinarily captured by more expensive equipment. The Joy-Con has an IR camera that is a mass produced sensor for motion activated lighting. The Ring-Con is as Old School a peripheral as it is possible to be, strapping injection molded plastic to existing controllers started was dead by the SNES before the Wii revived it, and the river of profit from RFA is enough to fund a couple Mario’s.

The point of Yokoi’s talk wasn’t the just withered technology, it was the lateral thinking. By definition the usage should be unexpected, and creative - not easily anticipated by the consumer.

On the left hand we have the planck length, beyond which silicon transistors cannot get more dense, and on the right hand we have graphics cards pursuing resolutions and frame rates beyond the limits of human perception. It’s not impossible that the home video console becomes “withered”, and Nintendo’s whole bet is that innovating the gameplay experience will yield longer returns than graphical investments
 
0
That’s a good point. And we also need to remember that a few weeks back he replied to one of Polygon’s posts discussing devkits with something along the lines of “not going to confirm specifics but a lot of this generally sounds like what I’ve heard partners say” (definitely me paraphrasing here).

(This wasn’t Polygon’s most recent informational post discussing 6gb RAM -> 12gb, but one before that)
IIRC, Nate's comments on Polygon's posts were related to:
The lack of certainty (as far as developer side is aware) on how Drake would be presented to the public.
&
Relatively recently, dev partners were briefed by Nintendo on... something.
 
Yeah we just know now that the SoC is done and can be mass produced. Switch's successor could be in production right now, but we have no proof that it is, just proof that all intended components are ready for mass production.
That’s rather soon. Too soon really.

For a product that people here believe will release in May, which is 8 months out, starting now is highly unlikely.

If it were to start production, it would be in around January for May 2023 launch or around June/July for a November 2023 launch. Unless people believe it’s March 2023….


If something is starting now, it would most likely be a new shield product or a different nvidia product, not for a new switch who would start later.


All the Linux indicated is that nvidia is working on something with T239 that runs in Linux for it, it has 8 core in what seems like a homogeneous cluster or is the assumption.


It has no direct meaning for what Nintendo is doing though. It only has direct meaning of what Nvidia is doing.

I feel like we have to be careful with phrasing here since people will believe anything we say and get the wrong idea when things don’t pan out.
 
That’s rather soon. Too soon really.

For a product that people here believe will release in May, which is 8 months out, starting now is highly unlikely.

If it were to start production, it would be in around January for May 2023 launch or around June/July for a November 2023 launch. Unless people believe it’s March 2023….


If something is starting now, it would most likely be a new shield product or a different nvidia product, not for a new switch who would start later.


All the Linux indicated is that nvidia is working on something with T239 that runs in Linux for it, it has 8 core in what seems like a homogeneous cluster or is the assumption.


It has no direct meaning for what Nintendo is doing though. It only has direct meaning of what Nvidia is doing.

I feel like we have to be careful with phrasing here since people will believe anything we say and get the wrong idea when things don’t pan out.
Is it fair to assume that the chip has been taped out by this point?
 
That’s rather soon. Too soon really.

For a product that people here believe will release in May, which is 8 months out, starting now is highly unlikely.

If it were to start production, it would be in around January for May 2023 launch or around June/July for a November 2023 launch. Unless people believe it’s March 2023….


If something is starting now, it would most likely be a new shield product or a different nvidia product, not for a new switch who would start later.


All the Linux indicated is that nvidia is working on something with T239 that runs in Linux for it, it has 8 core in what seems like a homogeneous cluster or is the assumption.


It has no direct meaning for what Nintendo is doing though. It only has direct meaning of what Nvidia is doing.

I feel like we have to be careful with phrasing here since people will believe anything we say and get the wrong idea when things don’t pan out.
Switch begun initial production at Foxconn in September and grew to 20k units a day by the beginning of October 2016. Holiday months have double the mass production potential over all other months, which is why a lot of new products release in Spring actually... I'm not suggesting that Switch's successor is already in mass production, I'm saying that we know that it CAN be, and if they want say 5M units ready for sale in May, instead of the 2.7M Switch had, they actually would need to start production right now to get 4-4.5M by the end of May.

Note that it takes about 8 weeks to ship a product from Asia to North America too.
 
jesus christ without T239 we're back at square one
What is "square one"? Does this mean nothing?

FMwgyRxXsAEOhnr
 
What is "square one"? Does this mean nothing?

FMwgyRxXsAEOhnr
According to MVG, that could be just the ARM T239 running Windows, for an entirely separate product, unrelated to Nintendo.

(To anyone who thinks I might be being a bit rough on MVG, he is misleading 10s of thousands of viewers via Spawnwave, and he should know better than to say NVN2 is supported in Windows, as if it is being used to create Windows programs and not inside of a development environment for other devices.)
 
According to MVG, that could be just the ARM T239 running Windows, for an entirely separate product, unrelated to Nintendo.

(To anyone who thinks I might be being a bit rough on MVG, he is misleading 10s of thousands of viewers via Spawnwave, and he should know better than to say NVN2 is supported in Windows, as if it is being used to create Windows programs and not inside of a development environment for other devices.)
Unless MS is finally going to come out with another NVidia powered device akin to the Surface 1 and 2 or even Zune HD.
 
Is it fair to assume that the chip has been taped out by this point?
Only assume they are working on it. As per @oldpuck from what they found, they had engineering samples in April 2022 (though I’d like to see the link to that), from engineering to tape out it can take a few quarters, 2-3.


And that assumes all goes perfect.


PS5 engineering samples happened in April 2019, we saw the product a whole year and a half later.

Drake could be taped out by December for all we know.

Or was taped out a while ago.

It’s not easy to pinpoint.
Switch begun initial production at Foxconn in September and grew to 20k units a day by the beginning of October 2016. Holiday months have double the mass production potential over all other months, which is why a lot of new products release in Spring actually... I'm not suggesting that Switch's successor is already in mass production, I'm saying that we know that it CAN be, and if they want say 5M units ready for sale in May, instead of the 2.7M Switch had, they actually would need to start production right now to get 4-4.5M by the end of May.

Note that it takes about 8 weeks to ship a product from Asia to North America too.
You aren’t suggesting it but someone will take it as though that’s what you’re saying. The start of late September lines up for a very early March launch. They don’t need to start this now for what people speculate is a May release.
 
but who to trust?! I can't choose between a Youtube news channel and someone who has shown after time again that he has great technical insight when it comes to these devices that I know nothing about and even passionately defends any wrong information by backing it up with cold harsh facts.

I jest ofcourse, but it has been a delight to be here so far. Stay awesome.
 
You aren’t suggesting it but someone will take it as though that’s what you’re saying. The start of late September lines up for a very early March launch. They don’t need to start this now for what people speculate is a May release.
The Switch had 2M units available at launch before covid, in a world that wasn't hit by a semiconductor shortage. They might not have to start production right now, but they may have to, we won't know either way with the current leaks, rumors and public info, but if production lines do start (which would line up with the OLED Splatoon model leaker saying that they were tightening security for these types of leaks) there will eventually be a leak about production, if Apple can't stop them, Nintendo doesn't stand a greater chance of doing so.
 
NVN is the Switch API.
As long as it is not officially confirmed that NVN2 is for Switch Next, it could obviously be anything, but come on. Of course it's for Switch Next.

Nate and MVG are only saying that nothing is confirmed so we should not treat anything as a fact just how they do it it's not good communicated and confusing.

Alas, no real news to find in the podcast.
 
Was NVN used in other Nvidia products or was it a specific graphics API for Switch?
"The development encompassed 500 man-years of effort across every facet of creating a new gaming platform: algorithms, computer architecture, system design, system software, APIs, game engines and peripherals. They all had to be rethought and redesigned for Nintendo to deliver the best experience for gamers, whether they’re in the living room or on the move."
https://archive.ph/45GgZ#selection-2093.171-2093.342
"NVIDIA additionally created new gaming APIs to fully harness this performance. The newest API, NVN, was built specifically to bring lightweight, fast gaming to the masses."

And yeah, it is only used in Nintendo devices so far, though you can use it in SDKs for the Switch on Windows.
NVN is the Switch API.
As long as it is not officially confirmed that NVN2 is for Switch Next, it could obviously be anything, but come ob. Of course it's for Switch Next.

Nate and MVG are only saying that nothing is confirmed so we should not treat anything as a fact just how they do it it's not good communicated and confusing.

Alas, no real news to find in the podcast.
I think the problem is MVG coloring Spawnwave's report about the leak, specifically that T239 doesn't link to Nintendo, of course Jon did point out that Kopite7 linked T239 as the custom built chip for Nintendo before it was ever even known to exist, because Jon/Spawnwave does an incredible job IMO. However if you want updates on Leaks, Rumors, public information about the Switch "2", this thread on this forum is the primary source of information, and that is because of the incredible work of Dakhil first and foremost, as well as the tech heavy minded people in this thread that just enjoy discussing Nintendo used Technology.
 
Only the switch.
To add, NX i.e. the Switch is specifically mentioned from that breach.

FMxwhDGXsAEIu37


To clarify - an NVN Windows "reference implementation" is like a proof-of-concept. It does not mean that NVN is specifically used to develop Windows software, but that a Windows implementation is available as a reference and can be used to develop NX software. (someone please add on / correct if I am missing something) .

In this instance it even says "Ampere-based GPUs are preferable [because] they are more compatible with the NX implementation". Implying NX is the target.
 
To add, NX i.e. the Switch is specifically mentioned from that breach.

FMxwhDGXsAEIu37


To clarify - an NVN Windows "reference implementation" is like a proof-of-concept. It does not mean that NVN is specifically used to develop Windows software, but that a Windows implementation is available as a reference and can be used to develop NX software. (someone please add on / correct if I am missing something) .

In this instance it even says "Ampere-based GPUs are preferable [because] they are more compatible with the NX implementation". Implying NX is the target.
This is specifically why MVG's reasoning was so disappointing to hear. He de-links it from Switch and specifically mentions that the API is used in Windows, but he should know why it's misleading to say so.
 
To add, NX i.e. the Switch is specifically mentioned from that breach.

FMxwhDGXsAEIu37


To clarify - an NVN Windows "reference implementation" is like a proof-of-concept. It does not mean that NVN is specifically used to develop Windows software, but that a Windows implementation is available as a reference and can be used to develop NX software. (someone please add on / correct if I am missing something) .

In this instance it even says "Ampere-based GPUs are preferable [because] they are more compatible with the NX implementation". Implying NX is the target.
I initially assumed "NX" there was a reference to their Tegra NX line (like Orin NX) but in context that really makes no sense. The only reasonable conclusion as far as I can tell is that it refers to the Switch code name NX, yeah.

Unless this chip works better with the Lexus NX specifically for some reason.
 
I initially assumed "NX" there was a reference to their Tegra NX line (like Orin NX) but in context that really makes no sense. The only reasonable conclusion as far as I can tell is that it refers to the Switch code name NX, yeah.

Unless this chip works better with the Lexus NX specifically for some reason.
occums razor and all that lol
 
Nate mentioned that T239's linked to the NVN2 leaks. And that the only takeaway from the Linux kernel is that T239 has 8 CPU cores per cluster. Nate said that he had conservations with individuals asking is there any direct link to the Nintendo Switch (I assume Nate meant Nintendo's new hardware here) mentioned in the NVN2 leaks (with T239), with the answer being no. And Nate thinks that outside of confirming T239 has 8 CPU cores per cluster, the Linux kernel means nothing to him.

MVG said that outside of confirming T239 has 8 CPU cores per cluster, and that Nvidia added T239 support to the Linux kernel, there's nothing of note in the Linux kernel. MVG said that Nvidia adding T239 support to the Linux kernel could mean that Nvidia's simply began to test T239 now and spending at least a year to fix bugs. And MVG mentioned that although T239 could be Nintendo related, T239 could also be related to a NDA project nobody knows about. MVG said that if there's a Nintendo email address in the Linux kernel, then he could say T239 could be related to Nintendo's new hardware. But since there's only a Nvidia email address in the Linux kernel, T239 isn't necessarily linked to the Nintendo Switch (I assume MVG meant Nintendo's new hardware).

And Nate and MVG said that T239 being in the Linux kernel doesn't confirm T239's in production.
This is a far better summary of what was said vs the bulletpoint version that omits context. We are viewing this from a vacuum and in that space one cannot make assumptions beyond what was shown -- which is strictly 8 CPU core per cluster spec. There is nothing else provided there.
 
To add, NX i.e. the Switch is specifically mentioned from that breach.

FMxwhDGXsAEIu37


To clarify - an NVN Windows "reference implementation" is like a proof-of-concept. It does not mean that NVN is specifically used to develop Windows software, but that a Windows implementation is available as a reference and can be used to develop NX software. (someone please add on / correct if I am missing something) .

In this instance it even says "Ampere-based GPUs are preferable [because] they are more compatible with the NX implementation". Implying NX is the target.
Oh right, NVN is windows based for tegra devices.

I forgot about that bit.
 
0
This is a far better summary of what was said vs the bulletpoint version that omits context. We are viewing this from a vacuum and in that space one cannot make assumptions beyond what was shown -- which is strictly 8 CPU core per cluster spec. There is nothing else provided there.
Supposedly elsewhere it indicates engineering samples in like April.
 
This is a far better summary of what was said vs the bulletpoint version that omits context. We are viewing this from a vacuum and in that space one cannot make assumptions beyond what was shown -- which is strictly 8 CPU core per cluster spec. There is nothing else provided there.
So based on all the available evidence and not just this Linux kernel update, you do believe it's likely that T239 is the chip Nintendo is planning to use, right?

It seems MVG explicitly does not believe this? I think this is where most of the confusion here is coming from.
 
This is a far better summary of what was said vs the bulletpoint version that omits context. We are viewing this from a vacuum and in that space one cannot make assumptions beyond what was shown -- which is strictly 8 CPU core per cluster spec. There is nothing else provided there.
"MVG said that Nvidia adding T239 support to the Linux kernel could mean that Nvidia's simply began to test T239 now and spending at least a year to fix bugs. And MVG mentioned that although T239 could be Nintendo related, T239 could also be related to a NDA project nobody knows about. MVG said that if there's a Nintendo email address in the Linux kernel, then he could say T239 could be related to Nintendo's new hardware. But since there's only a Nvidia email address in the Linux kernel, T239 isn't necessarily linked to the Nintendo Switch (I assume MVG meant Nintendo's new hardware)."

Linux public kernal, and Nvidia's internal work is very different, as another user here discovered that T239 has been in linux via "t23X" through Orin and in it's own internal use since January, and that engineering samples of T239 seem to have been completed since April. This is where bugs would be worked out in a developers environment before they push an update "Upstream" to the public Linux kernal, suggesting that the bugs have been ironed out to a point where other companies could use T239 in Linux, and not just internally at Nvidia.

Because NVN2 links to "NX" specifically in the API as posted here:
To add, NX i.e. the Switch is specifically mentioned from that breach.

FMxwhDGXsAEIu37


To clarify - an NVN Windows "reference implementation" is like a proof-of-concept. It does not mean that NVN is specifically used to develop Windows software, but that a Windows implementation is available as a reference and can be used to develop NX software. (someone please add on / correct if I am missing something) .

In this instance it even says "Ampere-based GPUs are preferable [because] they are more compatible with the NX implementation". Implying NX is the target.
and that NVN2 only lists support for T239 as the only Ampere SoC with ray tracing as of the end of February this year... It's extremely unlikely that the people you've mentioned targeting 1H next year as a release date with Switch "2" exclusive software, could be using any other SoC in the final devkits that would have to exist around this time for software to be completed ~6 months from now.

Basically the info MVG says isn't there in the Linux public info, is there in the Nvidia Hack from March 1st.
 
So based on all the available evidence and not just this Linux kernel update, you do believe it's likely that T239 is the chip Nintendo is planning to use, right?

It seems MVG explicitly does not believe this?
It's very likely but can't say it with 100% certainty -- even if I'd wager it is 99.99% for Nintendo.

Making a definitive claim on this would be a fool's errand.
 
That’s rather soon. Too soon really.

For a product that people here believe will release in May, which is 8 months out, starting now is highly unlikely.

If it were to start production, it would be in around January for May 2023 launch or around June/July for a November 2023 launch. Unless people believe it’s March 2023….


If something is starting now, it would most likely be a new shield product or a different nvidia product, not for a new switch who would start later.


All the Linux indicated is that nvidia is working on something with T239 that runs in Linux for it, it has 8 core in what seems like a homogeneous cluster or is the assumption.


It has no direct meaning for what Nintendo is doing though. It only has direct meaning of what Nvidia is doing.

I feel like we have to be careful with phrasing here since people will believe anything we say and get the wrong idea when things don’t pan out.
I still think a May release is pretty absurd. It doesn't have any special significance. This device doesn't NEED to sell alongside Zelda, and never has. This isn't 2016 anymore, Nintendo's hardware sells well on its own. I'd go so far as to say Nintendo would probably want the device to be out ahead of time to get it in more hands before Zelda released.

What we HAVE heard (not confirmed, all rumours):

Late 2022, early 2023. Seems reasonable. May is neither of these. Late 2022 or early 2023 always implied "this fiscal year". As I've said before, hardware missing its quarter is normal. Slipping an entire fiscal year? With something they've been planning for so long, I simply don't believe it.

A factory rumour detailed the Splatoon 3 OLED Model and a new model with a different backplate with a new shape, using the same logo as always. This implies this device is very much entering production, if not assembly, or already has.

Sparks of Hope looks "too good" to be on the original Nintendo Switch, and it's coming out soon.

2022 was mentioned by Grubb in a tweet, and more interesting to me, MVG, a developer for Switch, with MVG's comments from this year.

Developers were briefed recently by Nintendo, as I've heard a rumour, not sure it was here, that Nintendo around E3 time was flying in journalists to Washington for. Reasons we do not yet know. This happened in 2016 with press and developers getting a look at Nintendo Switch during E3.

What we KNOW:

T239 is used in NVN2, NVN1 was the API for Nintendo Switch. T239 has DLSS capabilities, is finalised and likely has production sampling and testing finished as drivers are now being finalised and distributed.

Nintendo has had a HUGE glut of valuable inventory for several months starting Q1 of this Fiscal Year with no valid explanation, the new accounting methods they claimed caused this accounting for less than 30% of the increased value. Leaving them with more raw materials, value wise, now, than when they were about to launch Switch Lite.

The Nintendo Switch Dock with LAN Port supports 4K output and can be updated, but neither feature has yet been put to use, or even been announced to have a use, a year out from launch.



Given all this I fully expect this thing to fall into March, and comfortably at that. From what I can glean from these facts, the hardware is finalised and the raw materials are in Nintendo's hands, with developers and press briefed on its existence and capabilities. There's no benefit to waiting, nor for reveal, not for release, and I doubt they will. They will reveal it when it is ready, and release it when it is ready, which, if I were one for betting, I'd say in the coming weeks for reveal, and March for release.

Holiday sales of Nintendo Switch would not be meaningfully impacted by the announcement, because the sort of people to buy Nintendo Switch NEXT are likely already to have a Nintendo Switch, and people getting into it this late in the generation aren't likely to be the kind that care about the best performance. Furthermore, Switch remains constrained in sales by supply, rather than demand. Lowering demand a little by revealing a new system would result in more money for Nintendo long term, because existing Switch stock would no doubt sell out anyway, especially with the glut of great games this holiday, while those "scared off", are scared IN to buying their more expensive new hardware.

My final point is one of economics. The holiday season is a huge shopping season. But, tax season in America and Golden Week in Japan are too, which is what the original Nintendo Switch targeted with its March release. If they are able to release it in March, before their fiscal year ends, when tax returns and Golden Week are on the horizon, there's no doubt in my mind that they will. May by comparison holds no real value to a launch other than being the month before E3 and related shows, and so having the spotlight to itself. Something March also has.
 
This is a far better summary of what was said vs the bulletpoint version that omits context. We are viewing this from a vacuum and in that space one cannot make assumptions beyond what was shown -- which is strictly 8 CPU core per cluster spec. There is nothing else provided there.
That is true but it also means you are (purposefully?) ignoring the NVN2 leak from the Nvidia ransomware attack.
Seen in a vacuum, there is no proof that T239 is Switch Next but connecting the dots from kopite to ransomware leak to Linux NVN2 leak gives a quite clear picture.
Would you agree that it is highly probable that this is the Switch Next chip?
Not saying there is definitely final proof because of course there is none yet. Just saying based on everything it is very probable.

Edit: ha, already answered while I was typing this, thanks ;)
 
Only assume they are working on it. As per @oldpuck from what they found, they had engineering samples in April 2022 (though I’d like to see the link to that), from engineering to tape out it can take a few quarters, 2-3.

To clarify, it's not definitive, just suggestive. In January and February, they're clearly just trying to get a system booted on the virtualizer, and are definitely not working with actual hardware.


It's pretty dead after that, but then work picks up in March with a rash of updates to the IO drivers, which could be nothing, but this one -


- involves fixing timing quirks that are specific to T239's PCI bus. It's not definitive, but this kind of bugfix implies actual hardware.
 
It's very likely but can't say it with 100% certainty -- even if I'd wager it is 99.99% for Nintendo.

Making a definitive claim on this would be a fool's errand.
Yeah of course we can't say with certainly but I do believe everything is pointing to it being likely, as it seems you do too.

I think that was really the source of confusion.
 
It's very likely but can't say it with 100% certainty -- even if I'd wager it is 99.99% for Nintendo.

Making a definitive claim on this would be a fool's errand.
This I completely understand. Until we hear that T239 is literally inside Switch "2" anything could potentially happen, as unlikely as it is.
 
There are fairly specific preconditions for Nintendo's historical approach to BC that, as far as we can tell at this point, aren't met with Drake. It only really works when the system either includes processors with identical architecture to the original system, or something that can produce a very close facsimile thereof. Drake cannot provide that. There'd have to be a separate chip, which is highly unlikely.
I don’t think historical conditions really apply this time, because things like hardware agnostic apis weren’t a thing in the past, and they never had a partner like Nvidia.

If Nvidias ceo was serious about a 20 year partnership, he better have a plan for future bc in place.
 
Yeah of course we can't say with certainly but I do believe everything is pointing to it being likely, as it seems you do too.

I think that was really the source of confusion.

They’re looking at this in a vacuum…geez why is this a discussion going for so long with people when the viewpoint was a vacuum
Which is why it was said when viewed in a vacuum, the info isn't anything substantial. I think I said vacuum three or four times to emphasize that point.

I simply acknowledged the NVN2 leak and public speculation/information but when viewed alone there is little here beyond the 8 core cluster in terms of useful information.
 
Which is why it was said when viewed in a vacuum, the info isn't anything substantial. I think I said vacuum three or four times to emphasize that point.

I simply acknowledged the NVN2 leak and public speculation/information but when viewed alone there is little here beyond the 8 core cluster in terms of useful information.
It does suggest final T239 hardware has been produced and thus ready for mass production
 
It does suggest final T239 hardware has been produced and thus ready for mass production
Yeah, primarily because to be put in the Linux Kernel in the way it was, the SoC does need to be finalized to well, be safely put in the Kernel primarily for discretionary reasons.

Otherwise you'd see a catalog of thousands of never-finished CPUs in the Linux Kernel
 
It does suggest final T239 hardware has been produced and thus ready for mass production
I think regardless or not these leaks indicate something, I think even in a vacuum, Nintendo needs to show new upgraded hardware lest they lose the momentum they built with the Switch.
 
0
It does suggest final T239 hardware has been produced and thus ready for mass production
I'd assume he knows but since it suggests but doesn't proof it, he mentioned it only proofed the 8 core CPU.

From what I see Nate is very explicitly trying not to indulge in Switch Next rumors but only comment on actual findings / facts right now.

Which is what is leading to confusion, because it gets interpreted as him negating anything that is not confirmed but conclusive.

I think he can only make up for it by leaking the release date now!
 
Yeah, primarily because to be put in the Linux Kernel in the way it was, the SoC does need to be finalized to well, be safely put in the Kernel.

Otherwise you'd see a catalog of thousands of never-finished CPUs in the Linux Kernel
I would like to gently correct and say that isn't quite true in this case. This is a generic Tegra driver that supports both Orin and Drake. Linux development strongly encourages drivers to live in-tree for technical reasons that aren't worth going into. Which means, because these drivers are shared, the update either had to bubble up or Nvidia devs would have had to maintain a really ugly multiple-internal-branch-cross-merging scenario which is highly mistake prone.

If I had to put money down, yeah, I'd say that hardware exists already. But because of the particular nature of this driver, it absolutely could have gone into the kernel without the hardware being finalized. If it breaks Tegra drivers for other customers, that's really NVidia's problem anyway, not the larger Linux scene
 
I'd assume he knows but since it suggests but doesn't proof it, he mentioned it only proofed the 8 core CPU.

From what I see Nate is very explicitly trying not to indulge in Switch Next rumors but only comment on actual findings / facts right now.

Which is what is leading to confusion, because it gets interpreted as him negating anything that is not confirmed but conclusive.

I think he can only make up for it by leaking the release date now!
Nate isn't suggesting that T239 isn't Switch "2", for some reason MVG is, but Nate isn't trying to say anything for or against, his own opinion is being muted to reflect confirmed reality, I appreciate that. MVG is mudding up reality and has been all week, it's not that he is lying, just misleading people with his comments about the API. Linux info in a vacuum is fine. Nate will eventually get to do his podcast with confirmed info, and we are all here for that, even hearing MVG's insights to it in that podcast. MVG is just telling a bunch of people who listen to these places for our insights, the wrong/misleading info... Of course NVN is for Nintendo, Nvidia said as much, the hack in NVN2 confirms it is targeting NX as a platform and it's only ever been used for Switch software.
I would like to gently correct and say that isn't quite true in this case. This is a generic Tegra driver that supports both Orin and Drake. Linux development strongly encourages drivers to live in-tree for technical reasons that aren't worth going into. Which means, because these drivers are shared, the update either had to bubble up or Nvidia devs would have had to maintain a really ugly multiple-internal-branch-cross-merging scenario which is highly mistake prone.

If I had to put money down, yeah, I'd say that hardware exists already. But because of the particular nature of this driver, it absolutely could have gone into the kernel without the hardware being finalized. If it breaks Tegra drivers for other customers, that's really NVidia's problem anyway, not the larger Linux scene
This is why I use the word suggests, but given that we know that T239 was in testing before tape out in spring 2021 via brainchild, and we know that it has been used in linux since at least January, that the public addition here is very likely final hardware, especially because you spotted engineering samples from back in April.
 
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