So it always costs the same to upscale to, say, 200% of the original rendered resolution? Doesn't matter if that's 540->1080 or 1080->2160? Is that right?it depends on the output resolution. it is a fixed cost based on your input and output ratio
Not without additional work (minimum recompiling the code or some translation layer)I have a bit of an offtopic question, if theoretically PlayStation or Xbox would switch to ARM, would backward compatibility still be?
So which is it, then? Does the time cost depend on the output resolution or the input resolution? Because I've now received two conflicting answers. Or is DLSS just not a fixed-cost at all? But I could have sworn that NVIDIA said that it is.
In any case, it should still be faster than native rendering, so my point still stands about it generally being preferable to save compute time by lowering the input resolution versus lowering the DLSS output resolution.
This must have been a rough estimation based solely on numbers. I have confidence that it'll be pushing beyond PS4-level visuals in reality.Since we are entertaining the 4Chan "leak", I find it hard to gloss over the whole "PS4 Performance" claim. If this leak were to end up being true and Drake is closer in performance to the PS4 than the Xbox Series S, then this would put 8nm back on the table, would it not? Seems like clock speeds of of the original Switch would be line up here. It would still be better than PS4, but just like Switch is a lot closer to PS3 than PS4, Switch Redacted would sit in a similar spot.
i've followed silently this thread for the last 300-350 pages, maybe more. First of all i want to thanks all of you, the thread sometimes goes nuts but it is fun and informative most of the times.
i have found this post on 4chan that went mostly ignored:
i know that 4chan leaks are unreliable and most of these infos can be derived by what we already knew and some patents but maybe there is something interesting?
p.s. My english is bad. Someone can explain to me what those ccd in the leaks are meant to be?
The issue is surmountable (Microsoft in particular has probably already done much of the necessary work for Windows), but it would be much more challenging.I have a bit of an offtopic question, if theoretically PlayStation or Xbox would switch to ARM, would backward compatibility still be?
no, it's fixed in that 540>1080 is always the same regardless of what you're rendering. if it's just a rectangle of solid color or a noise where no adjacent pixel has the same color.So it always costs the same to upscale to, say, 200% of the original rendered resolution? Doesn't matter if that's 540->1080 or 1080->2160? Is that right?
Nope, it’s BT4.xOh, well that's good then. How come the Switch currently uses SBC codec then? Does it not have BT5.0+?
No, different ISA. Same as how if Sony moved away from AMD to say, Intel for their GPUs, it would break compatibility due to different GPU ISA.I have a bit of an offtopic question, if theoretically PlayStation or Xbox would switch to ARM, would backward compatibility still be?
I am really considering an OLED but would hate to buy one only to have REDRAKED announced...
Isn't the Xbox OS just a collection of hypervisors and modified flavors of Windows? (IIRC the Xbox One OS is a modified windows 8)The issue is surmountable (Microsoft in particular has probably already done much of the necessary work for Windows), but it would be much more challenging.
It's definitely pretty, Windows-based, though the specific VM that games run under might be a bit more customized.Isn't the Xbox OS just a collection of hypervisors and modified flavors of Windows? (IIRC the Xbox One OS is a modified windows 8)
In that case they could probably do it pretty easily, at least compared to Sony. Windows on ARM already has an x86 compatibility layer that apparently works really well
Ahh, that's probably what it was. Yeah, that all makes sense now.If Nvidia ever said that DLSS is fixed-cost, my guess is they were referring to it being fixed-cost regardless of what's actually in the image being scaled. If you are, say rendering internally at 1080p and using DLSS to scale to 4K, then it should take exactly he same amount of time every single frame, which is quite important.
no, it's fixed in that 540>1080 is always the same regardless of what you're rendering. if it's just a rectangle of solid color or a noise where no adjacent pixel has the same color.
1080>2160 will always take longer than 540>1080 because there are just more pixels to work
$399 is the number. I do not see Nintendo going beyond that price. People can talk about inflation all they want, but consumers are already starting to change spending habits because of increased cost on everyday consumer goods. Plus, Nintendo is selling a lot more game digitally than in the past, so profits per game sold have gone up thanks to eliminating the retailer cut.This must have been a rough estimation based solely on numbers. I have confidence that it'll be pushing beyond PS4-level visuals in reality.
If the Switch is slightly above seventh generation but with an eighth-gen featureset, then it follows that Switch 2 would be slightly above eigth generation but with a ninth-gen featureset. If I had to say anything, then I reckon it should sit just a bit above the Steam Deck. And I think this'll be the first Nintendo console to be sold at a slight loss. If it goes above the $400 mark, then they'll be pricing out kids and families that just want to play Mario Kart.
I am really considering an OLED but would hate to buy one only to have REDRAKED announced...
Although, I wonder if they might do a repeat of the Wii U, where they have a $350-$400 'basic bundle' that they sell at a loss, and then a $425-$450 'premium bundle' that includes a game and maybe some other little trinkets so they can break even. Then they can advertise it at $350, when in reality most people will get FOMO once they're at the checkout and drop the full amount for the premium bundle. That's exactly what happened with the Steam Deck - the most expensive model was the best selling one, despite the lack of value for money.$399 is the number. I do not see Nintendo going beyond that price.
Do you have a citation about the most expensive one being the best selling one?Although, I wonder if they might do a repeat of the Wii U, where they have a $350-$400 'basic bundle' that they sell at a loss, and then a $425-$450 'premium bundle' that includes a game and maybe some other little trinkets so they can break even. Then they can advertise it at $350, when in reality most people will get FOMO once they're at the checkout and drop the full amount for the premium bundle. That's exactly what happened with the Steam Deck - the most expensive model was the best selling one, despite the lack of value for money.
hell, MK8 and Smash can probably do 2160p native.There is good reason to believe that even on the lower end expectations for Drake, it would be capable of rendering many of Nintendo's first party games at 1440p before applying DLSS. Mario Kart 8 and Smash Bros already render at 1080p with less than 400Gflops of performance. At minimum, we are looking at more than 5x the performance of the current Switch. The frame time to take 1440p to 2160p should be manageable. Nintendo likes to fake higher end rendering techniques rather than trying to implement real time resource heavy equivalents.
Do you have a citation about the most expensive one being the best selling one?
hell, MK8 and Smash can probably do 2160p native.
and I wouldn't say Nintendo likes to fake high end techniques. they simply don't have room to do a lot of them given the switch. I wouldn't be surprised if Nintendo moved to a ray traced GI solution for the Drake-oriented Zelda simply because they already have a probe-based GI system for BotW/TotK and they now have the horse power for higher fidelity
The Steam Deck is aimed at a specific audience that has the ability and desire to pay more than the average household that wants to play the new Mario.Although, I wonder if they might do a repeat of the Wii U, where they have a $350-$400 'basic bundle' that they sell at a loss, and then a $425-$450 'premium bundle' that includes a game and maybe some other little trinkets so they can break even. Then they can advertise it at $350, when in reality most people will get FOMO once they're at the checkout and drop the full amount for the premium bundle. That's exactly what happened with the Steam Deck - the most expensive model was the best selling one, despite the lack of value for money.
I'd say there's absolutely no need to use DLSS for Switch backwards-compatibility. [READACTED] will be powerful enough to run Switch games at its native resolution, no sweat. Unless they achieve BC through software emulation (which is not outside the realm of possibility - the PS3 and Xbox 360 did it, and the Switch is super easy to emulate), in which case, they would probably need to employ DLSS to get beyond 1080p.hell, MK8 and Smash can probably do 2160p native.
DLSS slots into the middle of the rendering pipeline, so it cannot be employed for post-rendering upscaling.I'd say there's absolutely no need to use DLSS for Switch backwards-compatibility. [READACTED] will be powerful enough to run Switch games at its native resolution, no sweat. Unless they achieve BC through software emulation (which is not outside the realm of possibility - the PS3 and Xbox 360 did it, and the Switch is super easy to emulate), in which case, they would probably need to employ DLSS to get beyond 1080p.
FortniteHow many PS5 games use RTGI.
You seem to have seriously misunderstood my point. I'm saying that, for those households that don't want to pay out the nose for the new system, they could offer a basic set that costs $400 max, to lure them into the idea of buying one. Then, a lot of people who only initially planned to get the cheap set will inevitably end up getting the more expensive set anyway, because that's just how most people think.The Steam Deck is aimed at a specific audience that has the ability and desire to pay more than the average household that wants to play the new Mario.
Personally I think they won't sell their next hardware at a loss for exactly the same reason they won't go over $400: the bad economy. Households are going to be careful and so is Nintendo.
The whole question is what you can put in a $400 machine without losing money. And I love the contributions in this thread in part because they try to answer this question, among other things.
Fortnite
Metro Exodus Enhanced Edition
Avatar Frontiers of Pandora
Lords of the Fallen
i've followed silently this thread for the last 300-350 pages, maybe more. First of all i want to thanks all of you, the thread sometimes goes nuts but it is fun and informative most of the times.
i have found this post on 4chan that went mostly ignored:
i know that 4chan leaks are unreliable and most of these infos can be derived by what we already knew and some patents but maybe there is something interesting?
p.s. My english is bad. Someone can explain to me what those ccd in the leaks are meant to be?
Woman, have you ever played a fsper on Wii and Wii U like the call of dooty series? Nothing feels more satisfying than 720 no scoping OSOK someone with a Dragunov from the other side of the map.Stationary sensor bar or lighthouse? No. God no. Please, anything but that. Improve gyro, don't regress.
I didn't say it would be done in post. They could easily update the games to add a DLSS pass into the rendering pipeline only when running on the new system.DLSS slots into the middle of the rendering pipeline, so it cannot be employed for post-rendering upscaling.
Feeling pretty doubtful that Open Air Zelda 3 will (on a platform massively weaker than the PS5) utilize an extremely demanding feature used in fewer than 5 PS5 games released or announced.
extremely demanding
This is almost definitely what we're getting. The BCM4356 in the current Switch is retired. Broadcom kept it around for a long time only because it happened to be significantly lower power draw than their other chips, but they've finally moved over to this Ultra Lower Power device.
- How about Bluetooth version? Maybe something similar to aptX Low Latency for headphones?
- How about WiFi version? Are we expecting WiFi 6/6E?
Think of it this way, the Nintendo switch is six years old and is in its seventh year on the market. You can buy an OLED switch right now, and you can remain with that even after the successor is announced and released, but unless Nintendo is literally doing something that no one else is doing, which is having 8 to 9 year cycles on the market before successors announced, you would basically just be buying something that will be short-lived, depending on how are you utilize it.I am really considering an OLED but would hate to buy one only to have REDRAKED announced...
RTXGI runs on a potato. It’ll be fine.Feeling pretty doubtful that Open Air Zelda 3 will (on a platform massively weaker than the PS5) utilize an extremely demanding feature used in fewer than 5 PS5 games released or announced.
Could easily be that the retail system has 12GB of RAM, whereas the dev kit has 16GB. Dev kits usually have more ram than retail units.16GB of RAM with only 10 GB for games and 2GB for OS. Yeah, the math doesn't add up there lol.
Wow... why though? That seems insanely overkill.XBox Series devkits have 40GB of RAM.
here's RTGI on a device weaker than the PS4
because we're just now getting out of the cross-gen era of current gen consoles, and UE5 games are just now starting to come out. you can't just "bolt on" GI like you can shadows and reflectionsIf it's not demanding, why isn't it used much then
You really can't have too much ram for game development. And they don't really care what the form factor or power draw is for a devkit.Wow... why though? That seems insanely overkill.
I know, but... 40GB? Nearly three times that of the retail system? That's insane.You really can't have too much ram for game development. And they don't really care what the form factor or power draw is for a devkit.
If TotK was delayed from March to May (just 2 months!), is it possible it’s due to financial report reasons?The only part of the tablet part of the Switch OLED TotK Edition that is different than the standard edition would be the back panel. Nintendo would easily be sitting on stock pile of Switch OLED tablets that either haven't had the back plates on them yet, or are easily swapped out when they decide to do a special edition. With that said, I would be curious to know what firmware is one Switch units being sold today? It seems that the Zelda TotK delay was indeed needed for game development and not to coincide with new hardware like I originally expected, so it is very well possible that they started to manufacture these special editions far ahead of time and seeing as how this firmware dates back to June of 2022, maybe Nintendo didn't expect the delay to be quite so hefty. A March 2023 release date could have been the original intent when they first announced the delay.
Since we are entertaining the 4Chan "leak", I find it hard to gloss over the whole "PS4 Performance" claim. If this leak were to end up being true and Drake is closer in performance to the PS4 than the Xbox Series S, then this would put 8nm back on the table, would it not? Seems like clock speeds of of the original Switch would be line up here. It would still be better than PS4, but just like Switch is a lot closer to PS3 than PS4, Switch Redacted would sit in a similar spot.
That's up to them. It wouldn't be easy, but it would be possible. Emulating x86 on ARM is notoriously easier than emulating ARM on x86, and MS has at least heavily invested in things like AOT recompilation for backwards compatibility.I have a bit of an offtopic question, if theoretically PlayStation or Xbox would switch to ARM, would backward compatibility still be?
because we're just now getting out of the cross-gen era of current gen consoles, and UE5 games are just now starting to come out. you can't just "bolt on" GI like you can shadows and reflections
It's kinda small, honestly. A medium tier workstation at a AAA developer is going to have something like 128GB, and I've seen high end graphics workstations with a terrabyte.I know, but... 40GB? Nearly three times that of the retail system? That's insane.
It's an interesting proposition, though I don't know how easy it would be to get that to work in practice. One of the properties of emulation is that the software does not know it is actually being run on a different device, but of course that doesn't prelude the possibility that you could have nascent features in there that Activate based on an OS check of some kind. It does mean you have to have a complete and separate rendering pipeline bolted onto the game, so it would require per-game manipulation, which kind of hampers the use case of software emulation versus porting.I didn't say it would be done in post. They could easily update the games to add a DLSS pass into the rendering pipeline only when running on the new system.
The entry price we are talking about, $400, is already high. A Nintendo console that costs $400 would not be something that will be perceived by the very general audience (i.e. not a forum of enthusiasts like us) as a "cheap" product.You seem to have misunderstood my point. I'm saying that for households that don't want to pay the nose for the new system, they could offer a basic package that costs $400 maximum, to lure them into buying one. Then a lot of people who initially only planned on the cheap set will inevitably end up getting the more expensive set anyway, because that's exactly what most people are thinking.
Hell, Nintendo already uses this pricing structure for games: people spend over $100 on TotK to get all the (largely useless) bonus items that come with these more expensive editions. Once someone is interested in something, they are more likely to spend more money, whereas if the collector's edition was the only version available for purchase, I guarantee that most of those same people who bought the collector's edition would have been turned -off by the high price. It's a simple psychological marketing tactic.
Or better yet, remove Wi-Fi altogether to improve everyone else's experience online.The important thing is that they have improved the network card. Needs better Wi-Fi reception.
ILet's not react like Nintendo fanboys and accept the truth: when it was released, the Nintendo Switch was indeed already technologically outdated compared to other HD hybrid game consoles available on the market. The very relevant Forbes journalist is also right to mention the games from 10 years ago. Everyone knows that the HD hybrid consoles available in 2013 were already at least as technically good.
mmm noOr better yet, remove Wi-Fi altogether to improve everyone else's experience online.
Generally yes. There would be certain points of ridiculousness (like scaling 360->2160), but I'd usually pick 720->2160 over 1080->1440 if they were both similarly viable.In any case, it should still be faster than native rendering, so my point still stands about it generally being preferable to save compute time by lowering the input resolution versus lowering the DLSS output resolution.
Let's not react like Nintendo fanboys and accept the truth: when it was released, the Nintendo Switch was indeed already technologically outdated compared to other HD hybrid game consoles available on the market. The very relevant Forbes journalist is also right to mention the games from 10 years ago. Everyone knows that the HD hybrid consoles available in 2013 were already at least as technically good.
The problem is that the gyro+accelerometer method leads to too much pointer drift. And yeah having the sensor bar attached to the current dock form wouldn't work but having a small USB-C rechargable bar would work much better especially since emitting two IR lights requires very little power anyway.It really doesn't...
Sensor bar based tracking is a no-go for so many reasons. The dock can be placed sideways, behind TVs, in cabinets, etc., and a LOT of people, including some of Nintendo's devs, use it like that. Adding a sensor bar to it (or the console bezel) would be absurd. It isn't 2006 anymore; there are better ways to achieve decent tracking.