• 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| (New Staff Post, Please read)

Kinda disappointed about the leak, there is currently no game that is set to be released in 2024 at day one launch from a very reliable source.

Sorry just to be clear, this is about a lack of leaks right? Not that there’s been some disappointing leak recently?

Didn’t think anything had changed.
 
Yeah, I believe that to be the general consensus. You think that's too optimistic? Within 15% of Series S seems reasonable and grounded, at least to me anyway.

Also I believe that figure is before DLSS in the picture. And compared to Series S, Switch 2 would also have better features and overall more streamlined.
That depends on what you mean by "Within a 15%". The thing is, is that this is a very specific number with a precise mathematical definition. So, it is within 15% of rated Flops? This would be something that can be computed but I doubt it is realistic unless they massively increase the power consumption on desktop mode. Perhaps within 15% on actual performance? That means if you run a benchmark like Wild Life the score would be within 15%? Perhaps some kind of performance test a dev did is getting this "within 15" number? I also doubt this.
 
Speaking of Nate’s birds,

His voice is something I could listen to all day. Like birds chirping.

I just listened to the newest episode today on the way to work, and it’s my first time hearing his voice. Dude is soothing to listen to. Makes MVG’s Aussie accent feel tame by comparison, and I love a good Aussie accent.
Thanks.
 
That depends on what you mean by "Within a 15%". The thing is, is that this is a very specific number with a precise mathematical definition. So, it is within 15% of rated Flops? This would be something that can be computed but I doubt it is realistic unless they massively increase the power consumption on desktop mode. Perhaps within 15% on actual performance? That means if you run a benchmark like Wild Life the score would be within 15%? Perhaps some kind of performance test a dev did is getting this "within 15" number? I also doubt this.

I would just refer you to this excellent reply:

It's not a rumor, it's a compilation of speculation and expectations primarily from this thread.

I would also find it best not to overthink what RGT meant by within 15% of Series S. It could mean anything (like what you listed), but as has already been pointed out by others (and I myself knew this already too), RGT isn't reliable. RGT previously said Switch 2 was going to use Mediatek SoCs (lol), then 2 weeks later changed that and then shared that specs that we are familiar with here (the one that includes "clocks clocks" line).

However, RGT's reliability aside, Switch 2 performing within 15% of Series S seems somewhat reasonable based on what we gathered from nvidia leaks.
 
What chances do you think we have of GTA VI being released on Switch 2 (Next Gen) on the same day as the rest of the consoles?
Or at least, on a very very close date.

I get the feeling that RockStar doesn't want to but I wouldn't understand why.
Maybe they see Nintendo as an inferior place to bring new great games on day 1?
 
I still like "Super Nintendo Switch":
1. A call-back to one of the greatest Nintendo consoles of all time
2. The former of which has aged enough that pinning it to Switch would be like putting an adjective on a noun, like "Super Switch" ("Super Nintendo Switch" being its full name, just as how people shorten Nintendo Switch to just being called the Switch).
3. Not just "New", not just an "iterative successor" ("2"), but "Supercharged" with new "super features like DLSS/machine learning" and possible some enhanced form of gimmick.
4. If it has backwards compatibility support, would definitely convey this idea better than from GameBoy Advance to Nintendo DS did.

I think Nintendo is going to make it as unambiguous as possible and call it something utilitarian like “Nintendo Switch — Generation 2”
Nintendo Switch Gen 2 2x2
Switch4
 
What chances do you think we have of GTA VI being released on Switch 2 (Next Gen) on the same day as the rest of the consoles?
Or at least, on a very very close date.

I get the feeling that RockStar doesn't want to but I wouldn't understand why.
Maybe they see Nintendo as an inferior place to bring new great games on day 1?
Having the Switch 2 version launch alongside all other versions would be HUGE, but considering the game has been in development for a long while and the priority from the start would have been PS5 and Series X, I can't see it launching day in date on switch. Maybe 6 months down the line?
 
Last I had heard, MP2/3 wasn't getting the remastered treatment, just a port.

Which would suck ass. I want remastered 2.
Nobody ever said this. Jeff Grubb said that remasters were happening but "not getting all the love and care" of the Prime 1 remaster, which is a meaningless statement. HD ports à la Pikmin would contradict the one actually credible source we had (which is not Grubb) about what the plans were for the Prime games.

But this is not the right thread for this discussion, so I'm not sure why it's here.
 
How about Nintendo Switch ii?
Why not Nintendo EGG?

They had Wii, why not Piss?

Why anything? Why ever?



I KIND OF hope they extend the current naming scheme. All models so far have put the descriptor AFTER "Nintendo Switch" - so the logical conclusion to me is [GENERATION] Nintendo Switch [MODEL].

SUPER Nintendo Switch fits this best, which is why that still has my highest hope. Even if I think, from a business perspective, Nintendo Switch - Generation 2 would be better.
 
0
GTA VI announcement teased for October 24th

Will history repeat ?

Can you imagine the console warrior reactions if the first peak at GTA6 was within the Switch 2 Announcement video? I doubt R* would go for that but stranger things have happened. If it was a launch title (because the stars just happen to align) I could actually see that as a possibility.

OR perhaps (more likely?) there is a rockstar event earlier that week that shows the game off, doesn't mention a Switch 2 port at all, and then it just shows up in the Switch 2 Announcement video. That would be cool too.

But I just crave the meltdowns of it being shown on Nintendo first... Gimme.
 
Thanks for this.

Crazy how decent it looks (at least to me anyway) at 360p, DLSS'd to 1080p..



Watching other 360p DLSS to 1080p videos shortly

Edit: If anyone know where I can find those same games (in the ResetEra link) at 360p (not DLSS'd or cleaned up) as a reference, let me know. I'd like to compare.


They definitely drastically improved something in the DLSS algorithm over the years because Control never had this kind of image quality using 360p to 1080p on version 2.0.

I have to admit I've been thinking the very same thing, but thought it might get a little creepy receiving a bunch of messages salivating over the tone of your voice... lol

Hey Nate, I heard you're returning to the Spawncast soon. Good news!

I gotta admit I haven't really watched since he's been on hiatus.
 
I gotta admit I haven't really watched since he's been on hiatus.

His critical takes are surely missed, haha. Josie Woah has been a great addition though. I like hearing opinions on established things from the younger generations.
 
0

I must be behind in my internet slang if I had to look that one up. :confused:
If it makes you feel better based has been a thing for about a decade it only became widely popular recently
 
Waiting for someone to correlate Jim Ryan quitting to the Switch 2 coming sooner
I have no clue how to even begin spinning such a web but all I'll say is I hope whoever is in charge next makes the push for a real portable console...
 
jim ryan dodging the switch 2
TRUE. that's the angle we needed. Jim Ryan is fucking TERRIFIED of DLSS technology. The ransomware attack leaked some of his emails where he used numerous expletives after being told a tablet could achieve visuals close to the PS5 and ray tracing that may even SURPASS the PS5. On multiple occasions he described it as "unfair horseshit software magic". Him stepping down is a clear sign of weakness that the company is worried about the future of gaming where the T239 reigns supreme. There are also reports that he tried downloading an SNES emulator on the Playstation Portal and it "exploded".
 
Now I have to follow up on this. It's been an outstanding question before in this thread whether the different quality modes actually affect the results of DLSS, or if they just define allowable scaling factors and such. My belief has been that they must be different algorithms, because otherwise the introduction of DRS support should have made them pretty much redundant as a concept. The comment I'm quoting here was made with the assumption that they are different.

However, I poked around the source code a bit, and I can't actually find a place where the quality mode does anything within the DLSS evaluation. If all the modes are really the same, that would explain why the minimum and maximum values for DRS are always 1/2 to 100%, and not relative to the mode's supposed optimal scale factor.

So, if that's the case, then to the quoted comment, I would amend that it doesn't matter which "mode" you're using. The choice is simply whether to stick between 1/2 to 100% scaling where DRS can be used, or to go down to 1/3 where it can't. In the former case you can't get up to 1080p in the 540p-360p DRS example; in the latter, you can do so by locking the resolution at 360p.

This would suggest that the behavior of not allowing DRS below 1/2 scale is a limitation of DLSS, not just one particular mode. Something to do with the history of frames it has to keep track of. The question is whether it's a limitation caused by the small scaling factor itself, or the range of scaling. For example, if dynamic input between 1/2 and 100% is allowable, but 1/3 to 100% isn't, what about 1/3 to 2/3? If you're using a 1/3 lower bound scaling (i.e. the so-called "Ultra Performance"), it's very unlikely you're going to natively render any frames at 100% output res anyway, so it's okay if the ceiling is 2/3 or maybe even 1/2.

If it's something along those lines (or if it was just a totally arbitrary restriction, borne of adding DRS after the fact and also treating Ultra Perf as an afterthought), then they may be able to enable DRS for a 1/3 bottom end scaling factor. But it's not encouraging that they still haven't done so to date if it was going to be useful in Nintendo's upcoming hardware.

Note that because the "modes" still exist and basically act as labels for a collection of otherwise independent DLSS settings, it may still be relevant which of them is being used in benchmarks and comparison videos and such, and it could appear as though they are even indirectly affecting the output or performance due to their link to a default "preset" (a euphemism Nvidia uses for the model weights); see screenshot below. Internally I imagine these are actually mapped using the scaling factor, not the meaningless "mode," but I can't check that since the presets feature was added after the leak.

image.png


@Anatole might be interested in all this.
Ah yes, thank you for tagging me!

For those newer to the thread: a couple of years ago (god, has it been that long? yikes), one of the main debates about DLSS was whether computational cost depended on the ratio of output resolution to input resolution or just the number of pixels in the output resolution. The internal architecture of DLSS is a type of neural network called a convolutional autoencoder. I've written more about the details of how those work here. One of the main distinctions between a convolutional neural network and other kinds of neural networks, like the multilayer perceptron, is that you can use the same convolutional architecture on images with different resolutions. This would imply that the cost of the neural network portion of DLSS depends only on output resolution...

...as long as the assumption held true that each of the modes (Quality, Performance, etc.) used the same architecture, anyway. It would have been possible, for example, to use a neural network with more layers or more channels in each layer in Performance mode than in Quality mode to marginally increase the final image quality. It would hardly be noticeable on an RTX 4090, but on the much less performant Switch 2, it could have been an issue.

LiC's post shows that the different modes do indeed share the same architecture, which confirms that the cost of the neural network scales with output resolution. I think that's become the predominant opinion in the thread anyway, thanks to finally having some good primary sources measuring cost in the DLSS programming guide. But still, it's nice to have confirmation.
 
Hey Nate, I heard you're returning to the Spawncast soon. Good news!
Still a ways off from making a return. But before end of the year.

I have to admit I've been thinking the very same thing, but thought it might get a little creepy receiving a bunch of messages salivating over the tone of your voice... lol

Haha, one can say whatever they want about the tone of my voice. It's nice to see.
 
As was pointed out by necrolipe, RDR1 was already rated weeks ago. So the initial tweet was incorrect about the listing be about RDR1 (assuming Necrolipe is correct).



Ratings when they inadvertently leak a game, if the publisher requests it, they delete it from the listings until it can be made public.

It has happened before with practically almost all the games leaked via ratings.We shouldn't be surprised by this.
 
Ah yes, thank you for tagging me!

For those newer to the thread: a couple of years ago (god, has it been that long? yikes), one of the main debates about DLSS was whether computational cost depended on the ratio of output resolution to input resolution or just the number of pixels in the output resolution. The internal architecture of DLSS is a type of neural network called a convolutional autoencoder. I've written more about the details of how those work here. One of the main distinctions between a convolutional neural network and other kinds of neural networks, like the multilayer perceptron, is that you can use the same convolutional architecture on images with different resolutions. This would imply that the cost of the neural network portion of DLSS depends only on output resolution...

...as long as the assumption held true that each of the modes (Quality, Performance, etc.) used the same architecture, anyway. It would have been possible, for example, to use a neural network with more layers or more channels in each layer in Performance mode than in Quality mode to marginally increase the final image quality. It would hardly be noticeable on an RTX 4090, but on the much less performant Switch 2, it could have been an issue.

LiC's post shows that the different modes do indeed share the same architecture, which confirms that the cost of the neural network scales with output resolution. I think that's become the predominant opinion in the thread anyway, thanks to finally having some good primary sources measuring cost in the DLSS programming guide. But still, it's nice to have confirmation.
Do you have any thoughts on why the dynamic resolution floor at 1/2 scale might exist? If there's only one algorithm, and we know 1/3 scaling is possible, then it's curious that they wouldn't support dynamic res with inputs down to 1/3.

Assuming it's not just an artificial restriction, that means the algorithm can handle a constant 1/3 scale just fine, and it can handle a history of frames of varying size between 50% and 100% scale, but it can't handle the variable range and 1/3 scale at the same time, which I can't really wrap my head around.
 
What chances do you think we have of GTA VI being released on Switch 2 (Next Gen) on the same day as the rest of the consoles?
Or at least, on a very very close date.

I get the feeling that RockStar doesn't want to but I wouldn't understand why.
Maybe they see Nintendo as an inferior place to bring new great games on day 1?
Extremely unlikely. GTA VI is a game made for PS5/Series X first and foremost, even a Series S port will be out of the table until further confirmation. I love entertaining the possibility, but tricks aren't going to cut it for a game needing all the legit compute power it can possibly get.
 
Extremely unlikely. GTA VI is a game made for PS5/Series X first and foremost, even a Series S port will be out of the table until further confirmation. I love entertaining the possibility, but tricks aren't going to cut it for a game needing all the legit compute power it can possibly get.
You don't think GTA6 would be on Series S? Even though Microsoft says devs have to put games on both Series X and Series S or they don't get to release?
 
Extremely unlikely. GTA VI is a game made for PS5/Series X first and foremost, even a Series S port will be out of the table until further confirmation. I love entertaining the possibility, but tricks aren't going to cut it for a game needing all the legit compute power it can possibly get.
There is no way GTA VI doesn’t hit Series S, even if it looks like GTA III.
 
Extremely unlikely. GTA VI is a game made for PS5/Series X first and foremost, even a Series S port will be out of the table until further confirmation. I love entertaining the possibility, but tricks aren't going to cut it for a game needing all the legit compute power it can possibly get.
Where do you see it say Series S port will be off (out of? sic) the table?
 
What chances do you think we have of GTA VI being released on Switch 2 (Next Gen) on the same day as the rest of the consoles?
Or at least, on a very very close date.

I get the feeling that RockStar doesn't want to but I wouldn't understand why.
Maybe they see Nintendo as an inferior place to bring new great games on day 1?

We had RDR in 2023, we will have RDR2 on Switch 2 in 2024, and GTA VI in 2025
 
Do you have any thoughts on why the dynamic resolution floor at 1/2 scale might exist? If there's only one algorithm, and we know 1/3 scaling is possible, then it's curious that they wouldn't support dynamic res with inputs down to 1/3.

Assuming it's not just an artificial restriction, that means the algorithm can handle a constant 1/3 scale just fine, and it can handle a history of frames of varying size between 50% and 100% scale, but it can't handle the variable range and 1/3 scale at the same time, which I can't really wrap my head around.
Yeah, even NVIDIA saying (and RR for the most part showing) that Ray Reconstruction doesn't play nice with Ultra Perf could just mean they didn't feed the RR Algorithm enough data/the right data to properly handle 1/3rd Res Rays.

Rather than there being a unique algorithm
 
0
As was pointed out by necrolipe, RDR1 was already rated weeks ago. So the initial tweet was incorrect about the listing be about RDR1 (assuming Necrolipe is correct).


Nope, RDR2 wasn't rated this week. They updated the RDR2 rating and accidentally revealed the Switch rating the game secretly has. Hid it again afterwards.

Ratings by Brazilian board is separate, if this was rated just now, it would show up as a separate rating. This Switch rating of RDR2 was rated in the past, we don't know how long ago. And don't forget, Brazilian board do not require full playable build of the game, Astroneer Switch was rated there with just a small proof. For all we know, RDR2 Switch was in the works some time ago but got canned.
 
Can you imagine the console warrior reactions if the first peak at GTA6 was within the Switch 2 Announcement video? I doubt R* would go for that but stranger things have happened. If it was a launch title (because the stars just happen to align) I could actually see that as a possibility.

OR perhaps (more likely?) there is a rockstar event earlier that week that shows the game off, doesn't mention a Switch 2 port at all, and then it just shows up in the Switch 2 Announcement video. That would be cool too.

But I just crave the meltdowns of it being shown on Nintendo first... Gimme.

GTAVI isn't releasing until like 2025.
 
I still like "Super Nintendo Switch":
1. A call-back to one of the greatest Nintendo consoles of all time
2. The former of which has aged enough that pinning it to Switch would be like putting an adjective on a noun, like "Super Switch" ("Super Nintendo Switch" being its full name, just as how people shorten Nintendo Switch to just being called the Switch).
3. Not just "New", not just an "iterative successor" ("2"), but "Supercharged" with new "super features like DLSS/machine learning" and possible some enhanced form of gimmick.
4. If it has backwards compatibility support, would definitely convey this idea better than from GameBoy Advance to Nintendo DS did.

I wonder if they'd move the 'Super' to the end of the name to match the current naming scheme:
  • Nintendo Switch
  • Nintendo Switch Lite
  • Nintendo Switch OLED
  • Nintendo Switch Super
The problem with this scheme is that it doesn't have a lot of legs: where do you go after Super? I guess they could do Ultra as a callback to the N64, but that's a more obscure reference that the average person isn't going to get.
 
I wonder if they'd move the 'Super' to the end of the name to match the current naming scheme:
  • Nintendo Switch
  • Nintendo Switch Lite
  • Nintendo Switch OLED
  • Nintendo Switch Super
That it would look like just another model of original Switch is a big reason I think anything that comes after Switch and is a word would be a bad idea.
 
Please read this new, consolidated staff post before posting.

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


Back
Top Bottom