• 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!
  • General system instability
    šŸš§ We apologise for the recent server issues. The site may be unavaliable while we investigate the problem. šŸš§

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

Or, and hear me out on this:

SXd4KrX.jpg


Edit: holy shit @Skittzo how did you do that?
I miss y2k-era neon colors and transparent tech
 
Hah! I get where @Concernt is coming from, didn't mean to create a pile on. I probably have a little bit of a bug up my ass about it.

It's clear that no one here is defending Lapsus$, but that particular argument gets deployed by people who do, and I think it's wrong. And while we can all condemn someone publishing a bunch of random employees social security numbers online, the fact that we (and I include myself, totally) rabidly consumed the data, and were excited about what it revealed can create a perverse incentive I'm not entirely comfortable with.

Sorry about that.
I think ultimately we agree, regardless of linguistic, or philosophical, differences on the matter.

Forced data breach bad.

But also very fun for nerds like us.
 
Anyway, I have Drake's graphics drivers for Linux now.
  1. Drake definitely does not have Ada's OFA, which is unsurprising.
    • It is the same OFA as Orin though
    • Figuring this out was a giant PITA
  2. Drake definitely does have Ada's NVENC implementation
    • As does Orin, not surprising.
  3. Unlike Orin, Drake supports clock gating on the OFA, NVENC, NVDEC, TSEC and VIC.
    • This almost definitely the FLCG @LiC has discovered
    • Interestingly, Erista also support this level of clock gating, so it's not a back ported Ada feature
    • Also interesting: Orin at one point appears to have had this form of clock gating as well, but it was dropped
  4. Drake pulls all of the camera management hardware out of Orin
    • Not surprising but worth noting
  5. At some point, T239 might have been called T234D.
    • Lots of defines for T234 and then defines for T234D
    • At one point T234D is referred to as Orin in a comment
May be some other stuff in here, something more interesting, but my eyes are tired.
 
Anyway, I have Drake's graphics drivers for Linux now.
  1. Drake definitely does nothave Ada's OFA, which is unsurprising.
    • It is the same OFA as Orin though
    • Figuring this out was a giant PITA
  2. Drake definitely doeshave Ada's NVENC implementation
    • As does Orin, not surprising.
  3. Unlike Orin, Drake supports clock gating on the OFA, NVENC, NVDEC, TSEC and VIC.
    • This almost definitely the FLCG @LiC has discovered
    • Interestingly, Erista also support this level of clock gating, so it's not a back ported Ada feature
    • Also interesting: Orin at one point appears to have had this form of clock gating as well, but it was dropped
  4. Drake pulls all of the camera management hardware out of Orin
    • Not surprising but worth noting
  5. At some point, T239 might have been called T234D.
    • Lots of defines for T234 and then defines for T234D
    • At one point T234D is referred to as Orin in a comment
May be some other stuff in here, something more interesting, but my eyes are tired.
I'm not sure whether FLCG would necessarily mean the ability to gate certain new blocks -- wouldn't that just be ELCG or BLCG or whatever, implemented on a different side of the power tree? Whereas the "level" in "first level" implies a different granularity. But it's possible!

As for T234D and T239D -- those are just internal identifiers for T234 and T239 specifically for the Linux display driver. As I understand it, it's a consequence of Nvidia open sourcing only parts of the driver stack, so a distinct ID is necessary to recognize/expose only the display driver parts that they want to be public.
 
Last edited:
Isnā€™t there a minimum storage speed needed?
No, not any that's mentioned in documentation anyway. You're probably referring to brainchild's needs, but that might be specific to him. At worse, I assume you'll just be stuck with a lower fidelity model than flat out not working

There is this notice, but it's too vague to be of any use

Since Nanite relies on the ability to rapidly stream mesh data from disk on demand. Solid State Drives (or SSDs) are recommended for runtime storage.
 
No, not any that's mentioned in documentation anyway. You're probably referring to brainchild's needs, but that might be specific to him. At worse, I assume you'll just be stuck with a lower fidelity model than flat out not working

There is this notice, but it's too vague to be of any use
I wonder what the actual minimums speed is, since "SSD" is pretty vague. Couldn't eMMC count as "SSD"?

I actually still think eMMC is a possibility for Drake, even if UFS is reasonably affordable, since I doubt many games demand speeds higher than what it can provide. Meanwhile it's cheap and they could provide a lot more storage for the same price, in theory.
 
I know we are all pretty sure the switch dock won't change much, and there is a million reasons not to do this, but I think a redesign that works like an old VCR would be hella cool.

A mechanism whereby you remove the joy con's and slide the console into a flap on the front and a mechanism then proceeds to swallow the device and feed it into the body.

The mechanism would probably fail, trapping your drake switch, it would be a pain to keep taking your controllers off, it serves no functional purpose, but damn it would be cool and nostalgic.

Wonder if I can get some joycon rails, an old VCR and a cheap dock and make one?
 
I know we are all pretty sure the switch dock won't change much, and there is a million reasons not to do this, but I think a redesign that works like an old VCR would be hella cool.

A mechanism whereby you remove the joy con's and slide the console into a flap on the front and a mechanism then proceeds to swallow the device and feed it into the body.

The mechanism would probably fail, trapping your drake switch, it would be a pain to keep taking your controllers off, it serves no functional purpose, but damn it would be cool and nostalgic.

Wonder if I can get some joycon rails, an old VCR and a cheap dock and make one?
Or like the original NES in the US where you put it in connector first then press down on it. Interesting but entirely absurd. I do kind of wish the next dock redesign would make allowances for mounting the dock horizontally like Sakurai does, as it stands the Dock with LAN Port's hinges slightly tilt it up at the rear, so when horizontal the console has a small risk of slipping out. Even if it's just a rubber ring on the back and a slight forward angle to the part that covers the screen so it faces slightly up when horizontal. Not much they can do about this position turning the logo upsidedown, but Playstation and Xbox cope fine with their logos going sideways in some orientations.
 
While researching when the lite backplate leaked, I found an interesting quote from a reddit thread that I thought applied here as well:
People here have a mindset that speculation even on extremely likely things is inappropriate because someone could get disappointed. Its a really bizarre mindset to me. Speculation is fun, if you get disappointed once or twice so what get over it.
 
Last edited:
So this means no DLSS 3.0 then, right?
Unfortunately I believe so, yes.
I think it's the final nail in the coffin, yeah. Honestly, with Drake's size and likely clock speeds, it seems like it was probably always off the table. DLSS 3.0 is not a great fit for 30fps->60fps games anyway, so I think on the whole have that extra power and transistor budget used elsewhere is probably a win.
 
Mentionning 2024 or even end of 2023 used to be haram opinions, but I see that we are slowly, but surely, entering that liminal phase where plans must have surely changed and a lot must have happened behind closed doors.



I personnally have no issue with coasting through 2023 with the current Switch, and if that's a indeed Nintendo's intention as I believe it, I have no doubt that we will have a solid year of videogames despite the aging hardware. Additionally, the new generation of console is expected to last much longer than usual due to a combination of factors such as covid and diminishing returns, so Nintendo can in my opinion perfectly release their new console in 2024 without it feeling late/outdated.
While this post was likely said in jest, this type attitude leads to abuse and harassment when plans do change. When plans change (because they can and do so often), it's not used as an excuse. It's used to explain what happened in a situation.
 
If this thing does get a delay to holiday 2023 or beyond, I just hope there's a good reason for it. Like I would be fine waiting a couple more months after May to hear about the system, if it was because, for example, they delayed it so it could release on a better node. Maybe the battery life concerns among other things made them feel like they could get a much better device if they waited a little (half a year or so) longer. Maybe they'll launch it with the next 3D Mario after Zelda, which is why they've kept it hidden for so long? Who knows, but I would hope they didn't choose to delay it for some inane reason.
 
0
So this means no DLSS 3.0 then, right?

Unfortunately I believe so, yes.

I think it's the final nail in the coffin, yeah. Honestly, with Drake's size and likely clock speeds, it seems like it was probably always off the table. DLSS 3.0 is not a great fit for 30fps->60fps games anyway, so I think on the whole have that extra power and transistor budget used elsewhere is probably a win.
Ehhhhhhh I wouldn't say so
Orin is a part for AI Computing and specifically AI Automation.

A part of that nature needs to be incredibly precise, especially in it's awareness of the world around it in automotive or mechanical applications, so the OFA in Orin likely is vastly superior to the OFA in Ampere at the very least to facilitate that (Can't have a self-Driving car or a Drone/Automated Machine careen into a thing because it's OFA is too slow to give it a sense of awareness of the world around it)

So Drake, sharing that OFA, would inherit that increase in speed/precision.
The question is, what is the Orin OFA's relation to the Lovelace OFA? Is the Lovelace OFA faster than the Orin one? Was it built off of the Orin one, or were they designed in parallel for the different parameters/targets of the architecture they were targeting?

Now, none of this confirms DLSS Frame Generation as being viable, it does make it more likely than if it had the Ampere OFA, but even if the Orin OFA could run DLSS-FG just fine, Nintendo may want to go the route of using that performance to drive the DLSS cost as close to 0 as they could rather than using Frame Generation to make up the cost-difference.

Either way, there is no "Nail in the coffin" yet IMHO, and to assume such is jumping to conclusions without considering the scope of what exactly Drake having Orin's OFA means.
And again, even if it could run DLSS-FG just fine like Lovelace could, Nintendo may want to use the headroom to optimize DLSS's Upscaler instead, so there is a lot of uncertainty there.
 
It's silly to say Nintendo is targeting a certain market with a console that hasn't been announced yet. One has no idea what the console's price will be, how it will be marketed/positioned, or even really how it will perform (not that the latter actually implies a target market). Those are all assumptions.
I kinda disagree on one thing, and that's the pricing. Based off the fact that Switch is still selling quite well (for now), and that we have yet to see a price drop for any of the models... I think we have reason to believe that Drake can easily be priced at $400 for a base model with the lowest storage offered (128 GB lol). And I say $400 under the assumption/scenario that OLED and regular V2 model gets a $50 price drop to $250 and $300.

That and chip parts being more expensive because of the transistor shortage.

I hope Nintendo offers two skus for Drake. I would buy the more expensive one for more storage.
Not at all. I just expect Drake to run Hogwarts at 4k DLSS/60fps versus the Switch version being 720p/20-30fps when docked. XBO will probably be 900p/30fps and PS4 1080p/30fps.
I dunno. I have a really hard time believing Drake will be that big of a jump for Hogwarts version, even if it's based off the Switch version, and DLSS is involved.

720p to 2160p requires 9x as many pixels alone, which is like 9x GPU power. And then we're talking about doubling framerate to boot. This is also assuming Drake won't get any upgrades in fidelity, and we are talking about an optimistic 3 tflops in docked (7.6x faster GPU). I guess we will have to see how much of the framerate is CPU vs GPU. Maybe it is doable if the fidelity is switch level, but at 4k?

I hope Hogwarts on Drake is based off better specs though. Maybe we'll get 1080p 60 fps without DLSS and 2K with, if it's xbone/PS4 quality/fidelity?
 
Last edited:
I thought for like 6 full pages of this thread, why is the general Nintendo thread full of talk about the next Switch and it bugged me when there is a dedicated thread about it. I was just about to ask why the discussion moved here, before I was done typing I realized it was all me, because I entered the wrong thread :S

I was the baddy here.
 
I wonder what the actual minimums speed is, since "SSD" is pretty vague. Couldn't eMMC count as "SSD"?

I actually still think eMMC is a possibility for Drake, even if UFS is reasonably affordable, since I doubt many games demand speeds higher than what it can provide. Meanwhile it's cheap and they could provide a lot more storage for the same price, in theory.
There most likely isn't a hard minimum, just practical limits on what you can do with particular hardware without hitching.
 
I know we are all pretty sure the switch dock won't change much, and there is a million reasons not to do this, but I think a redesign that works like an old VCR would be hella cool.

A mechanism whereby you remove the joy con's and slide the console into a flap on the front and a mechanism then proceeds to swallow the device and feed it into the body.

The mechanism would probably fail, trapping your drake switch, it would be a pain to keep taking your controllers off, it serves no functional purpose, but damn it would be cool and nostalgic.

Wonder if I can get some joycon rails, an old VCR and a cheap dock and make one?

I'm with you thinking this sounds cool, but the realist in me sees a variety of reasons that this would be a bad idea, from too expensive to too failure prone.

The OLED dock is good enough, they should keep it like that.

While this post was likely said in jest, this type attitude leads to abuse and harassment when plans do change. When plans change (because they can and do so often), it's not used as an excuse. It's used to explain what happened in a situation.

One day, suits in companies will understand that there's only two viable ways for a Plan B, as taught by the A-Team and the Ghostbusters. It's either:

- There is no Plan B
or
- Plan B is running away

^^ j/k

I thought for like 6 full pages of this thread, why is the general Nintendo thread full of talk about the next Switch and it bugged me when there is a dedicated thread about it. I was just about to ask why the discussion moved here, before I was done typing I realized it was all me, because I entered the wrong thread :S

I was the baddy here.

You either die a hero or scroll long enough to be the thread's baddie. ;D
 
I wonder what the actual minimums speed is, since "SSD" is pretty vague. Couldn't eMMC count as "SSD"?

I actually still think eMMC is a possibility for Drake, even if UFS is reasonably affordable, since I doubt many games demand speeds higher than what it can provide. Meanwhile it's cheap and they could provide a lot more storage for the same price, in theory.
Nanite has other benefits than just throwing in million poly meshes. Even for low poly meshes, you can get a more compressed file out of it, allowing you to send more data. So there's no real lower bound here
 
I wonder what the actual minimums speed is, since "SSD" is pretty vague. Couldn't eMMC count as "SSD"?

I actually still think eMMC is a possibility for Drake, even if UFS is reasonably affordable, since I doubt many games demand speeds higher than what it can provide. Meanwhile it's cheap and they could provide a lot more storage for the same price, in theory.
technically thereā€™s no difference between an SSD and eMMC since both are flash storage, one can be faster than the other and vice versa.

But SSD has taken a hold under marketing as meaning one type of flash storage, a very specific one at that. And it can have nothing to do with the actual speed in the marketing angle, just has to be NVMe. And thatā€™s a protocol.

UFS, CFExpress, eMMC, SATA, etc., are all SSDs and they all function for a similar purpose but are different enough.


This is all to say that, people asking for the next switch to have an SSD donā€™t realize it already has an SSD.

If people ask for a faster SSD thatā€™s a particular request, rather than asking for an SSD at all. As SSD is equated as being the same thing as ā€œfaster specific storage mediumā€.

and that there is not specific medium that is ā€œSSDā€.

But, it is used in marketing especially in consoles whereā€¦ no one knows what it is lol. Usually anyway.

Like if you have any smartphone or tablet of the last decade, it has an SSD already. But 2020 PS5 and XBox marketing made it even more of a tool for console wars ā€¦šŸ¤·šŸ¾ā€ā™‚ļø

which is lameā€¦ but anyway, people should change their language for it from ā€œI want a switch with an SSDā€ to ā€œI want a switch with a faster SSDā€. Mean two different things.

Had convos where people did actually think in the first quote, and thought the switch had a HDD. Explained it doesnā€™t and they only retorted with ā€œbut itā€™s not NVMe, so how can it have an SSD?ā€
 
Isnā€™t there a minimum storage speed needed?
Iā€™m sure Iā€™ve read that hardware will hit a CPU bottleneck long before they hit a drive speed bottleneck when it comes to UE5 and Nanite.

Itā€™s going to be one of the reasons imo that graphically intense, open World PS5/XBSS/X only games will probably target 30fps instead of 60fps despite the large optimisations Epic has made to the engine, Nanite and Lumen in the past year. Most AAA developers vision (going for realism) will almost always push visual fidelity over killing themselves trying to squeeze everything into a 16ms frame to hit 60fps.

As @ILikeFeet has said in the past targeting 30fps instead of 60fps means devs can use up to 4x extra effects like higher quality shading, depth of field, motion blur, subsurface scattering and more precise Lumen all in game (as opposed to locked to cut scenes when targeting 60fps in game) and all at higher base resolutions before the UE5 temporal upscaling which are the things that will make most people go ā€œwow now thatā€™s next genā€.

There will be inevitable PS5 Proā€™s and Series XX for those that want the AAA games at nearer 4k/60fps imo even if it means they arrive in year 4 (2024) rather than after 3 years (2016) PS4 Pro.

What the above means for Drake is anyoneā€™s guess. The good thing about UE5 is that itā€™s incredibly scalable (even more so than other modern engines which already are) so when you combine that with a potential 2tflop handheld / 4tflop docked console which also supports DLSS it opens the doors to a lot of possibilities that just wouldnā€™t be there were it not for Drakeā€™s power, UE5 and DLSS.
 
0
Iā€™m not sure what AMD can offer for SONY/MICROSOFT for a console targeting 2024 as revisions.

Sony and Microsoft are the major customers for AMD that fund their R&D for their GPUs.

Whatever they want, theyā€™ll find AMD to deliver it and AMD will also deliver it for their PC graphics.

However, the consoles are limited in that even if they get a new additions, it has to be a Frankenstein approach like with the PS4 Pro and XBox One X.

So, letā€™s say RDNA 4 has a new feature, the PS5 Pro still has to use a lot of the hardware features that are present in its custom RDNA implementation. Likewise the Series X Pro or whatever it is called.


Ps5 doesnā€™t use mesh shaders for example, and RDNA3 dropped the legacy NGG which was present in RDNA1 and 2 (and older). If they were to use something, it would have to have the legacy RDNA1/2 implementation with something else bolted from the RDNA(x) uArch.

Mesh Shaders wouldnā€™t be present, but NGG would. Why? Because PS5 doesnā€™t use that. Mesh shaders are more efficient and a better form of something the geo engine in the Oberon aims to do. Itā€™s still good but MS is later down the path.


Seires X is more deeply wedded to the RDNA2, and carries with it other issues with such, like memory. A faster GPU would require higheR memory bandwidth. SX is already using a huge SOC for a console, to increase effective BW, youā€™d need to increase the size or the complexity and MS already discussed some of the issues with cost there.


Unless people are fine with say, 700/800-1000 dollar consoles, Iā€™m not sure what they can improve much upon. Theyā€™ll get more of what they already have right now is the most realistic approach.



So, I guess thatā€™s my comment on all of this Pro stuff for the other consoles. Maybe Iā€™m wrong and they blow my expectations out of the water, but they just seem likeā€¦ theyā€™ll be more of the same. Maybe Iā€™m too conservative here for them. But I wonā€™t exclude the possibility, just recognize that itā€™s not high on my list. Pretty low.




Not even a chiplet approach Iā€™m confident right now, and donā€™t think it makes particularly a lot of sense hereā€¦
 
It's silly to say Nintendo is targeting a certain market with a console that hasn't been announced yet. One has no idea what the console's price will be, how it will be marketed/positioned, or even really how it will perform (not that the latter actually implies a target market). Those are all assumptions.

But I don't even think it makes sense to frame it like "targeting casual gamers," or not, at all. I think they're going to do the same thing they've been doing since 2019: targeting "Nintendo Switch owners" and "potential Nintendo Switch owners."

There's also the fact that the timeline from GPU architecture, to Tegra, to Switch model is not actually going to end up all that different from the original Switch, contrary to the apparent belief that the new model is going to be so cutting edge that it must be positioned for enthusiasts only.
To put a exclamation point on this very fine point, I believe the OLED Switch was Nintendo's first step into the upper market. The numbers clearly reflect that. With that said, the product and brand is strong/mature enough to allow it to appeal to all audiences provided the new hardware can deliver.
 
Ehhhhhhh I wouldn't say so
Orin is a part for AI Computing and specifically AI Automation.

A part of that nature needs to be incredibly precise, especially in it's awareness of the world around it in automotive or mechanical applications, so the OFA in Orin likely is vastly superior to the OFA in Ampere at the very least to facilitate that (Can't have a self-Driving car or a Drone/Automated Machine careen into a thing because it's OFA is too slow to give it a sense of awareness of the world around it)

So Drake, sharing that OFA, would inherit that increase in speed/precision.
The question is, what is the Orin OFA's relation to the Lovelace OFA? Is the Lovelace OFA faster than the Orin one? Was it built off of the Orin one, or were they designed in parallel for the different parameters/targets of the architecture they were targeting?

Now, none of this confirms DLSS Frame Generation as being viable, it does make it more likely than if it had the Ampere OFA, but even if the Orin OFA could run DLSS-FG just fine, Nintendo may want to go the route of using that performance to drive the DLSS cost as close to 0 as they could rather than using Frame Generation to make up the cost-difference.

Either way, there is no "Nail in the coffin" yet IMHO, and to assume such is jumping to conclusions without considering the scope of what exactly Drake having Orin's OFA means.
And again, even if it could run DLSS-FG just fine like Lovelace could, Nintendo may want to use the headroom to optimize DLSS's Upscaler instead, so there is a lot of uncertainty there.

There's a possibility of Nintendo having their own custom upscaling technology Based on a few patients nintnedo put up in 2021. Some of them had something to do with using AI to upscale images. I am sure DLSS will be the default upscaler used but I wouldn't question Nvidia drive to implementing their own technology for Nintendo sake and potentially calling a custom version of frame generation "Nintendo's secret stuff" or something like that. Will it be as good as Nvidia current frame generation? Probably not but it will allow that feature to be present in future games. There's only 35 games that support DLSS 3.0 and FG so I am sure Nvidia wants more developers to adopt that feature.

I wouldn't rule out the possibility of Nintendo receiving a custom version of FG that works only with Drake and the 40 series Nvidia cards
 
By inflation, the $300 launch Switch would be ~$370 now. Ignoring the SOCs for a second, the Switch was built off of industry standard parts that were mostly already at stable prices, and likely have been affected by inflation same as everything else.

I have no sense of what Drake will cost Nintendo, because TX1 was an existing chip and Drake is customized. But I think the best case scenario is that by reusing some of Orin's arch, and letting Nvidia sell the product elsewhere, Nintendo has managed to get the cost of their custom chip back down to approximately the same price.

I believe Nintendo will launch their next Switch at $399. I believe the OLED model launching at $349 was testing the waters to see how a higher price would impact sales, and consumers have shown they prefer the more expensive OLED model. I also believe releasing the OLED model allowed them to start sourcing OLED displays with the idea that the next Switch will launch with the same OLED display thus allowing them to get better pricing thanks to the high volume purchasing.

The Xbox Series S uses an APU with a die size of 197mm2 and sells for $299. I dont think the actual cost of Drake will be problematic for Switch 2 at a $399 price, but the bigger question is just how big can the SOC be and still fit within the Switch form factor? The Tegra X1 was originally 118mm2, but when looking at tear downs it looks like there is plenty of board space to spare compared to something more crowded like the Aya Neo Air. The full fat Orin is 450mm2, comparing die photos to something like the PS4 APU suggest that the Orin uses about 10-15% of its die space on hardware not related to gaming. So will the believed specs be able to fit in a SOC smaller then 200mm2?

The Switch OLED launched with a dock that has a built in control board capable of outputting 4K, and essentially has no use with the current Switch. Why put this in there, it certainly cost more money than the previous dock. Unless of course they wanted the OLED dock to be compatible with the next Switch, but this would mean the Switch 2 would not be able to be any thicker than the current model.
 
The Switch OLED launched with a dock that has a built in control board capable of outputting 4K, and essentially has no use with the current Switch. Why put this in there, it certainly cost more money than the previous dock. Unless of course they wanted the OLED dock to be compatible with the next Switch, but this would mean the Switch 2 would not be able to be any thicker than the current model.
I assumed it was parts availability for HDMI converter chips, with 4K 60 being a max default. May have been cheaper to use an existing one.

I could be wrong - the Aula datamine mentioned a 4K realtek chip which is the one on the OLED, maybe 4K output was enabled at one point? Not like it would do much for the current chip.

I am curious if they could fit 8nm Drake into the current OLED shell, I asked that question here before and was told it could be a tight fit.
 
There's a possibility of Nintendo having their own custom upscaling technology Based on a few patients nintnedo put up in 2021. Some of them had something to do with using AI to upscale images. I am sure DLSS will be the default upscaler used but I wouldn't question Nvidia drive to implementing their own technology for Nintendo sake and potentially calling a custom version of frame generation "Nintendo's secret stuff" or something like that. Will it be as good as Nvidia current frame generation? Probably not but it will allow that feature to be present in future games. There's only 35 games that support DLSS 3.0 and FG so I am sure Nvidia wants more developers to adopt that feature.

I wouldn't rule out the possibility of Nintendo receiving a custom version of FG that works only with Drake and the 40 series Nvidia cards
AMD's frame generation will probably be open source like FSR and work on everything, so it would be a good proof of concept, on top of previous interpolation experiments

 
AMD's frame generation will probably be open source like FSR and work on everything, so it would be a good proof of concept, on top of previous interpolation experiments


That's the wonderful thing about AMD FSR and future implementation of their FG it will be open sourced. It's why we saw some first party Nintendo games use FSR because it allowed them to save some resources and squeeze out more performance out their dated hardware for games like XC3.
 
That's good, at least. I recall something about an expected new SoC that would arrive in 2025/2026, and while I wouldn't expect Nintendo to go for that bleeding edge tech right away, it's still a mild concern of mine to have new hardware on the horizon that would feel outdated almost immediately after the fact lol
There's always a bigger fish.
I dunno. I have a really hard time believing Drake will be that big of a jump for Hogwarts version, even if it's based off the Switch version, and DLSS is involved.

720p to 2160p requires 9x as many pixels alone, which is like 9x GPU power. And then we're talking about doubling framerate to boot. This is also assuming Drake won't get any upgrades in fidelity, and we are talking about an optimistic 3 tflops in docked (7.6x faster GPU). I guess we will have to see how much of the framerate is CPU vs GPU. Maybe it is doable if the fidelity is switch level, but at 4k?
You mention DLSS you don't seem to account for it. DLSS isn't strictly limited to these options, but going with the labels they use for PC settings, to achieve 2160p60 in the various modes, they'd need to start with...
Quality: 1440p60
Balanced: 1252p60
Performance: 1080p60
Ultra Performance: 720p60

Hard to say what will be considered feasible or best practice. Getting a 720p30 Switch game to at least those lower two seems like it should be trivially easy in most cases. Less so if it's really "720p30, but drops both resolution and frame rate a lot".
 
That's the wonderful thing about AMD FSR and future implementation of their FG it will be open sourced. It's why we saw some first party Nintendo games use FSR because it allowed them to save some resources and squeeze out more performance out their dated hardware for games like XC3.
XC3 uses an in-house temporal upsampler. the other Nintendo games that did list FSR don't actually show signs of usage
 
There's always a bigger fish.

You mention DLSS you don't seem to account for it. DLSS isn't strictly limited to these options, but going with the labels they use for PC settings, to achieve 2160p60 in the various modes, they'd need to start with...
Quality: 1440p60
Balanced: 1252p60
Performance: 1080p60
Ultra Performance: 720p60

Hard to say what will be considered feasible or best practice. Getting a 720p30 Switch game to at least those lower two seems like it should be trivially easy in most cases. Less so if it's really "720p30, but drops both resolution and frame rate a lot".
I try to account for it. I have no idea how CPU extensive the game is, but considering it's open world, perhaps Drake can bring it from 30 on last gen consoles to 60fps. So 1080p 60fps might be doable with PS4 fidelity? without DLSS?

I think we have to consider if Drake will be a port of the switch version or something else.. like PS4 or a port of series/PS5. Because that does change things quite a bit.

i've only used DLSS as a 2x resolution ball park estimate.

These are one of the e games I'm not gonna really bother deep dive predicting from an arm chair. I would love to see 720p 1st party switch games bring upgraded first on Drake with the maximum improved performance.
 
0
forgive me if this has come up but i donā€™t seem to remember, but would it be plausible for the next switch to have the capacity to have an nvme ssd added to it as expandable storage? i was just looking at how cheap the lower tier nvme drives have become, youā€™re looking at Ā£45 here in the UK for 1tb, thatā€™s basically half the price of a 1tb microsd card.
 
Drake will almost certainly use a different colour to differentiate it from the OLED model. The base console is Black, OLED is White how about Nintendo Red for Drake?

Both OG, V2 and the OLED Switch tablets are black. Only the dock and Joy Con are different. Then OLED/premium model however was the only version with white accessories. I would take out the white version for the OLED Model and offer white only with the premium modelā€¦ then the Drame Model.

Other idea, ditch the neon for Drake and go with matte/pastel colors, like with the animal crossing V2 or mixed neon, like with the Splatoon 3 edition.
 
forgive me if this has come up but i donā€™t seem to remember, but would it be plausible for the next switch to have the capacity to have an nvme ssd added to it as expandable storage? i was just looking at how cheap the lower tier nvme drives have become, youā€™re looking at Ā£45 here in the UK for 1tb, thatā€™s basically half the price of a 1tb microsd card.
It's a rather fascinating discussion that I have seen it come up, but I think in short that while it would be appreciated, 1) Nintendo does currently sit on the board of the SD Association, and it wouldn't make sense to use a competing storage format, and 2) space constrictions might prove difficult to fit inside the current form factor of the Switch if no changes to the thickness of the console are being made
 
forgive me if this has come up but i donā€™t seem to remember, but would it be plausible for the next switch to have the capacity to have an nvme ssd added to it as expandable storage? i was just looking at how cheap the lower tier nvme drives have become, youā€™re looking at Ā£45 here in the UK for 1tb, thatā€™s basically half the price of a 1tb microsd card.
Possible? Yes, but very unlikely. I believe Nintendo wouldnā€™t consider that ā€œPlug and Playā€ enoughā€¦ maybe they could design around that but it would be surprising if they tried, I think.
 
Nintendo should replace the kickstand design with an NVME drive that just hangs off the back of your switch. They could do Nintendo branded ones with different designs and storage volumes.

I don't see anything that could go wrong with this approach.
 
0
forgive me if this has come up but i donā€™t seem to remember, but would it be plausible for the next switch to have the capacity to have an nvme ssd added to it as expandable storage? i was just looking at how cheap the lower tier nvme drives have become, youā€™re looking at Ā£45 here in the UK for 1tb, thatā€™s basically half the price of a 1tb microsd card.
Post in thread 'Future Nintendo Hardware & Technology Speculation |ST|' https://famiboards.com/threads/future-nintendo-hardware-technology-speculation-st.55/post-486284
 
Ehhhhhhh I wouldn't say so
Orin is a part for AI Computing and specifically AI Automation.

A part of that nature needs to be incredibly precise, especially in it's awareness of the world around it in automotive or mechanical applications, so the OFA in Orin likely is vastly superior to the OFA in Ampere at the very least to facilitate that (Can't have a self-Driving car or a Drone/Automated Machine careen into a thing because it's OFA is too slow to give it a sense of awareness of the world around it)

So Drake, sharing that OFA, would inherit that increase in speed/precision.
The question is, what is the Orin OFA's relation to the Lovelace OFA? Is the Lovelace OFA faster than the Orin one? Was it built off of the Orin one, or were they designed in parallel for the different parameters/targets of the architecture they were targeting?

Now, none of this confirms DLSS Frame Generation as being viable, it does make it more likely than if it had the Ampere OFA,
It probably does have the OFA from Ampere, because Orin probably has the same OFA from Ampere.

Both Drake and Orin have OFA 1.2. What version Ampere has, I dunno, because it's not documented in the drivers anywhere - Tegra chips have an alternate driver structure than the pure GPU does, which is why I have this data at all.

The optical flow SDK however, documents 3 OFA types. The Turing OFA, the Ampere OFA, and the Ada OFA. The Ada OFA removes features from its Ampere predecessor, which is how I know that the Orin OFA isn't Ada's - the example code for Orin in the SDK uses those features.

The truth is, it probably doesn't matter. It seems very likely that there isn't magic in the OFA design on Ada. Nvidia documents a 2x performance uplift for Ada's OFA but... that's the same as the clock speed increase. It's probable that OFA needs excellent performance to be able to generate flow data fast enough for DLSS 3, but it gets that speed not from some design change or features, but just the upped clock that 4nm provides.
 
0
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