• 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!)

Reader mode
I believe the solution the PlayStation 5 is using is custom - the Geometry Engine. Mark Cerny talks about it here around 28:27 in the Road to PS5 video. It utilizes primitive shaders and its focus is mostly geometry optimization. So there’s no mesh shaders on PS5.
 
The issue regarding smaller BVHs for Switch 2 games is that

-Third-parties are not going to do this (they will almost exclusively do software RT as they don't even bother to allow the option of Hardware Lumen in most PC games and because the PS5 and Xbox Series X are horrible at hardware RT)

-Very small BVH structures for shitty raytracing is not something NVIDIA has any interest in and all of their research is about how to do path tracing better.

So EPD would be basically doing this themselves with very little support to lean on or reference.
 
7hatDeadCat said:
are there performance differences between amd and nvidia when it comes to mesh shading?
Click to expand...

There isn't good data on mesh shading out there, because it's so rare in games. The best we have are 3DMark's mesh shader benchmark, which I happen to think is especially bad by even 3DMark standards.

Looking at what data there is, and making some informed guesses, I would bet on Nvidia having better mesh shader performance per dollar on their PC cards, but that doesn't necessarily translate to a big "punching above its weight" on the console side.

Where it might matter is Unreal Engine. UE5 has a mesh shader backend of Nanite, and as @ReddDreadtheLead has pointed out, it seems like the mesh shader backend isn't actually usable at all on the other consoles. This might be worth it, but really, it takes someone doing some very specialized benchmarking to find out.

7hatDeadCat said:
is it anything like raytracing where the t239 may actually be competitive with the ps5/xsx?
Click to expand...
I want to single this out since it's become a bit of a meme. T239 won't be competitive with the PS5/XSX in RT performance. Where it probably will be competitive with the Series S in RT. Ports that have to turn off RT when scaling down to Series S size may be able to keep them on with Switch NG. And Nintendo, whose games are all bespoke to the hardware, will likely make use of those cores, but not necessarily the same level of RT effects.
 
The Switch 2 will almost surely be worse than the Series S at raytracing for third-party games as third-parties mostly do software RT.

Hardware Lumen is more expensive than Lumen Epic which is WAY more expensive than Lumen High which is what most Series S games will use.
 
ItWasMeantToBe19 said:
The issue regarding smaller BVHs for Switch 2 games is that

-Third-parties are not going to do this (they will almost exclusively do software RT as they don't even bother to allow the option of Hardware Lumen in most PC games and because the PS5 and Xbox Series X are horrible at hardware RT)

-Very small BVH structures for shitty raytracing is not something NVIDIA has any interest in and all of their research is about how to do path tracing better.

So EPD would be basically doing this themselves with very little support to lean on or reference.
Click to expand...
1. The Swtich 2 is not a pc. How third parties treat the pc version, is not necessarily indicative of the Switch versions treatment. A lot of third parties more or less treat the pc as an Xbox.

2. That's a bit of a generalization. Nvidia made NVN2s featureset with the idea that devs is going to utilize it. They are interested in scaling their technology from the low end to the high end.

3. Do you think the teams that gave us Witcher 3, Doom Eternal, Dying Light 2 etcetera are not going to get impress on Drake?
 
Hermii said:
1. The Swtich 2 is not a pc. How third parties treat the pc version, is not necessarily indicative of the Switch versions treatment. A lot of third parties more or less treat the pc as an Xbox.

2. That's a bit of a generalization. Nvidia made NVN2s featureset with the idea that devs is going to utilize it. They are interested in scaling their technology from the low end to the high end.

3. Do you think the teams that gave us Witcher 3, Doom Eternal, Dying Light 2 etcetera are not going to get impress on Drake?
Click to expand...

The major issue is that this doesn't just require mastery of the Switch 2 hardware, but mastery of bleeding edge algorithms that were just created.

NVIDIA created ReSTIR (their hardware RT algorithm) three years ago and currently refers to ReSTIR using this language.

To be clear, right now, ReSTIR is a box of razor blades without handles (or a box of unlabeled knobs). It’s extremely powerful, but you have to know what you’re doing. It is not intuitive, if your existing perspective is traditional Monte Carlo (or real-time) sampling techniques.
People sometimes think SIGGRAPH paper = solved. Nope. We’ve learned a lot since the first paper, and our direct lighting is a lot more stable with that knowledge. We’re still learning how to do it well on full-length paths.
And there’s a bunch of edge cases, even in direct lighting, that we know how to solve but haven’t had time to write them up, polish, and demo.
We haven’t actually tried to solve the extra noise at disocclusions in (what I think of as) a very principled way. Right now a world-space structure is probably the best way. I’m pretty sure it can be done without a (formal) world-space structure, just “more ReSTIR.”
Click to expand...

Reading list on ReSTIR – Light is beautiful

Recently a short video from dark magic programmer Tomasz Stachowiak made the rounds in the graphics programming community, at the sound of jaws hitting the floor in its wake. It shows his recent progress on in his renderer pet project: … Continue reading →
lousodrome.net
 
ItWasMeantToBe19 said:
The Switch 2 will almost surely be worse than the Series S at raytracing for third-party games as third-parties mostly do software RT.

Hardware Lumen is more expensive than Lumen Epic which is WAY more expensive than Lumen High which is what most Series S games will use.
Click to expand...
But Nvidias solution with DLSS raytracing tech can work with pretty much any RT solution. It's just an AI basically taking a reference image, and prettying it up. The AI just needs a few references, and a bit of information. Like, let's say, you make a software raytracing only do 2 bounces of light, which is pretty lightweight, and then 10% of rays actually bounce. The AI of DLSS does the rest of the work for you. Which is why there are reports that Lumen's lighting system works better on Switch 2 than PS5.

It really doesn't matter the strength of the hardware, when you can fudge things so much with clever software. Again Switch 2 games that run a 720p, low settings, and low RT settings will look comparable to a 4K PS5 game. DLSS is just that good, and is only going to get better as the AI improves.
 
I've been dismissive of the possibility of a dockless system (and I indeed don't pay much mind to the "Super Play" Tweets), but I've never actually asked the siliconheads here if wireless casting is actually feasible. I assume there'd be proprietary technology involved, but is there any reason to believe that latency wouldn't be untenably poor regardless?
 
There are literally no reports of Lumen working better on the Switch 2 than PS5. No one would ever run hardware Lumen on the Switch 2 as it just eats performance currently and the PS5 is just obviously better at software Lumen because it's more powerful. If Epic made a much more limited version of hardware Lumen that was less resource intensive, that would be very helpful for the Switch 2, but that hasn't happened.
 
0
My takeaways thus far (since T239 was known/rumored to the public) is that if Nintendo EPD or other teams were to make proper full use of the SoC to give us the best visuals they can, the following features should ideally be implemented in a best-case scenario:

  • Mesh Shaders to optimize geometry and enable significantly more geometric detail and complexity at similar power draw vs. not using mesh shaders
  • Hardware-accelerated ray-tracing to greatly enhance lighting and enable more physically accurate reflections, shadows, ambient occlusion, and/or global illumination
  • Deep Learning Super Sampling (DLSS) to maximize image quality on a per-pixel level to output at higher resolutions and keep render resolution targets conservative and to remove flickering artifacts that would be there at native-res with or without TAA
  • DLSS 3.5 Ray Reconstruction which uses DLSS tech to improve the image quality of ray-traced geometry, such as those in reflections, while slightly increasing performance in those scenarios
  • Increased memory bandwidth and quantity compared to the original Switch to allow more texture and geometry assets to stream in and out of video memory, improving texture quality, geometry quality, and even ray-tracing performance while minimizing memory-related stuttering and bandwidth-related performance dips
  • Increased instructions-per-clock on the CPU to better utilize the improved single-core performance on the CPU due to higher clock speeds to potentially increase frame-rate targets in some games or to improve other aspects at a software level in games such as AI behavior or amount of NPCs on-screen or level of detail at a distance in game worlds and to increase scene complexity in ways that would tax the CPU more at the same FPS target
  • Increased storage bandwidth and read and write speeds to improve load times and potentially eliminate loading screens in some scenarios as well as speeding up installing downloaded games or content (this would also utilize the improved CPU and memory)
Note that I did not refer to software-level tweaks and potential custom optimizations at a hardware-level that are unknown at this point. I think if teams use the correct art style (precisely like EPD is known for doing) we could be seeing some mighty impressive results on this thing. Might not be the full-fledged CGI-like look that some are expecting but you never know until we see it in action of course.
 
Raccoon said:
I've been dismissive of the possibility of a dockless system (and I indeed don't pay much mind to the "Super Play" Tweets), but I've never actually asked the siliconheads here if wireless casting is actually feasible. I assume there'd be proprietary technology involved, but is there any reason to believe that latency wouldn't be untenably poor regardless?
Click to expand...
It's technically possible, even at 4K, but it's expensive, and you get one of two options. Either the device is more expensive, a LOT more expensive, or something else gives, like being far less capable. In Wii U, just to stream 480p, with the source of it having the benefit of wall power, it was both more expensive AND less powerful than it could have been. While I personally REALLY want casting to come back to allow for Wii U and DS games and their play concepts, I can't see it happening for a number of years.
 
Very shitty hardware ray tracing is probably worse than baked GI and carefully placed cube maps.

Shitty ray tracing would be a lot easier to do (IF you're good at ray tracing algorithms), but I don't know which Nintendo devs are both

1. So cutting edge that they can easily implement ray-tracing
2. Want to save dev time even if it degrades the visuals.

You could use shitty RT alongside baked GI and carefully placed cube maps, but I don't know how much it would improve the image relative to the performance cost.
 
0
ItWasMeantToBe19 said:
The issue regarding smaller BVHs for Switch 2 games is that

-Third-parties are not going to do this (they will almost exclusively do software RT as they don't even bother to allow the option of Hardware Lumen in most PC games and because the PS5 and Xbox Series X are horrible at hardware RT)

-Very small BVH structures for shitty raytracing is not something NVIDIA has any interest in and all of their research is about how to do path tracing better.

So EPD would be basically doing this themselves with very little support to lean on or reference.
Click to expand...
  1. smaller BVH is a matter of inserting low level lods. third parties have done worse to get games running on switch
  2. the lumen thing is because software lumen is inherently faster at the moment
  3. smaller BVH is something Nvidia has put research into and put out tips for. but ultimately, it's on the developer because BVH size is dependent on the dev's own assets
EPD isn't going to do this on their own, Nvidia is creating the damn thing alongside them

ItWasMeantToBe19 said:
The major issue is that this doesn't just require mastery of the Switch 2 hardware, but mastery of bleeding edge algorithms that were just created.

NVIDIA created ReSTIR (their hardware RT algorithm) three years ago and currently refers to ReSTIR using this language.



Reading list on ReSTIR – Light is beautiful

Recently a short video from dark magic programmer Tomasz Stachowiak made the rounds in the graphics programming community, at the sound of jaws hitting the floor in its wake. It shows his recent progress on in his renderer pet project: … Continue reading →
lousodrome.net
Click to expand...
RESTiR has shown to be very scalable. in this thread, I've shown RESTiR running on a multitude of devices. many different technical engineers had done their own variations as a hobby and as research for big companies, including AMD. again, Nvidia is working alongside Nintendo with research, no one is a in the dark as you think they are
 
ItWasMeantToBe19 said:
Why would Nintendo launch the Switch 2 in July or August.
Click to expand...

Those are probably the least likely months, outside of January and February of course. Every other month is fair game. Nintendo was willing to release Zelda TotK in May, so if SNG is going to release in H1, I believe May stands a very good chance of being the month to do so. If its going to be H2, I like September a lot. Get the early rabid early adopters out of the way in September and October, and then flow into the holiday with good momentum. If Nintendo has ample stock, I could see them moving upwards of 20 million units in their first six months. Consumers like to upgrade. We see it all the time with people upgrading their TV or phone even though there is nothing wrong with their current model. For those 60 million Switch users that purchased their Switch 2017-2019, many of them will be eager to get a new model.
 
ILikeFeet said:
  1. smaller BVH is a matter of inserting low level lods. third parties have done worse to get games running on switch
  2. the lumen thing is because software lumen is inherently faster at the moment
  3. smaller BVH is something Nvidia has put research into and put out tips for. but ultimately, it's on the developer because BVH size is dependent on the dev's own assets
EPD isn't going to do this on their own, Nvidia is creating the damn thing alongside them


RESTiR has shown to be very scalable. in this thread, I've shown RESTiR running on a multitude of devices. many different technical engineers had done their own variations as a hobby and as research for big companies, including AMD. again, Nvidia is working alongside Nintendo with research, no one is a in the dark as you think they are
Click to expand...

How shitty is hardware ray-tracing with super low level of detail assets would be my question here.

Would it even look as good as baked GI or use as few resources as baked GI?
 
hannibunny2411 said:
Pokemon leak debunked I guess.
Click to expand...
Bnois said:
It debunks Pokémon rumour for me. GF can’t do stuff properly on normal days, so I don’t see them releasing some NG patch next year for a game that is finished
Click to expand...
eye0h said:
Pokemon Leak is dead! 🦀

:(
Click to expand...
Why? It literally says "Hidden Treasure of Area Zero" in the leak, which was back in January. Same as the tweet. Isn't this confirmation it's legit?

Early '24 and Dec isn't that different. Maybe they're freeing up January for the Switch 2 event.
 
TLZ said:
Why? It literally says "Hidden Treasure of Area Zero" in the leak, which was back in January. Same as the tweet. Isn't this confirmation it's legit?

Early '24 and Dec isn't that different. Maybe they're freeing up January for the Switch 2 event.
Click to expand...

The names were announced in the trailer.

The “leak” implied that DLC2 was going to release slightly after the Switch 2 and that clearly did not happen.

(Game Freak is never going to update the performance of their games)
 
TLZ said:
Why? It literally says "Hidden Treasure of Area Zero" in the leak, which was back in January. Same as the tweet. Isn't this confirmation it's legit?

Early '24 and Dec isn't that different. Maybe they're freeing up January for the Switch 2 event.
Click to expand...
Frankly the details have gotten insanely muddled, and the leak isn't even that far off. "Early 2024" and it got moved to December 14th. I know people like doom-posting about leaks, but do some critical thinking people.

I understand not really believing the leak at this stage, but I don't think it's 100% debunked after one reveal... especially since that leak was around 6 or more months ago.
ItWasMeantToBe19 said:
The “leak” implied that DLC2 was going to release slightly after the Switch 2 and that clearly did not happen.
Click to expand...
"implied" doesn't mean much for leaks, especially when people can accidentally make implications that they didn't intend to make.
Concernt said:
* Hidden text: cannot be quoted. *
Click to expand...
Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.
 
ItWasMeantToBe19 said:
How shitty is hardware ray-tracing with super low level of detail assets would be my question here.

Would it even look as good as baked GI or use as few resources as baked GI?
Click to expand...
you keep calling the hardware shitty, but do you know what the RT Cores do? and what they do with the assets?

and in addition to that, how common methods of GI work? how do some of these methods utilize ray tracing?
 
The leak was always super suspicious as it had the traditional form of

1. List details anyone who saw a trailer early could provide

2. Then promise things people care about way more using your “credibility” from getting trailer details right.

The idea that GF would bother to update SV was also super suspicious as GF has cut off communication about SV’s performance and has refused to update the horrible engine. GF has issued one “performance patch” in 12 months for SV, where they put fewer objects on screen.

So when the “leak” starts getting major details super wrong…
 
ItWasMeantToBe19 said:
The names were announced in the trailer.

The “leak” implied that DLC2 was going to release slightly after the Switch 2 and that clearly did not happen.

(Game Freak is never going to update the performance of their games)
Click to expand...
I don't think I'm willing to throw out the leak entirely considering how much else it's gotten right thus far. A leak happening a year out from when it could be proven definitely has the possibility of some things getting mucked up as internal timelines change. I'll be surprised now if there's no next gen patch for SV and the DLCs now.
 
SoldierDelta said:
Frankly the details have gotten insanely muddled, and the leak isn't even that far off. "Early 2024" and it got moved to December 14th. I know people like doom-posting about leaks, but do some critical thinking people.
Click to expand...
So that's what it was? I don't follow pokemon as a franchise closely (and thus I don't care much about leaks/rumors for that franchise).

People are really saying the leak was debunked because it's released in December 14 instead of early 2024? That's kinda crazy. That's like saying a hypothetical MP4 leak from last summer saying it'd release early 2024 is debunked because we got surprised with a Dec 14 release date for MP4?

When I saw mentions of "leak is debunked", I initially there were was some other conditionals attached to the release of the Pokemon DLC and now that it was announced for December, then something else in the leak couldn't possibly be real, but I don't know what that would be because I don't follow Pokemon franchise/news/leaks/rumors.
 
darthdiablo said:
So that's what it was? I don't follow pokemon as a franchise closely (and thus I don't care much about leaks/rumors for that franchise).

People are really saying the leak was debunked because it's released in December 14 instead of early 2024? That's kinda crazy. That's like saying a hypothetical MP4 leak from last summer saying it'd release early 2024 is debunked because we got surprised with a Dec 14 release date for MP4?

When I saw mentions of "leak is debunked", I initially there were was some other conditionals attached to the release of the Pokemon DLC and now that it was announced for December, then something else in the leak couldn't possibly be real, but I don't know what that would be because I don't follow Pokemon franchise/news/leaks/rumors.
Click to expand...
It could be something more that I'm missing, Pokemon is one of the series that I don't care much about, but the leak didn't even get anything wrong with the exception of "Early 2024" as far as I can tell. The initial "Pokemon leak is dead" posts were definitely about that, if nothing else.
 
The leak is debunked from the beginning because there’s no chance GF would ever bother to miss the Christmas period for the DLC no matter how unpolished it was.

(GF is not really countering those claims by refusing to show anything related to DLC2, lol)

The leak’s clear implications were the leaker was implying that the Switch 2 would launch holiday 2023, and DLC2 would launch in January or February 2024 with a Switch 2 performance patch.

But the Switch 2 was clearly never intended to release holiday 2023.
 
Concernt said:
* Hidden text: cannot be quoted. *
Click to expand...
No no, you're in the clear, didn't think you were coming off as snippy at all haha

Edit: whoops, I did use trigger when I meant shoulder buttons 😅 My bad for causing confusion in the first place
 
Raccoon said:
I've been dismissive of the possibility of a dockless system (and I indeed don't pay much mind to the "Super Play" Tweets), but I've never actually asked the siliconheads here if wireless casting is actually feasible. I assume there'd be proprietary technology involved, but is there any reason to believe that latency wouldn't be untenably poor regardless?
Click to expand...
At the point at which there is proprietary technology, there is at least a dongle. The theoretical latency could get down to the same as a wireless controller, effectively - you'd just be sending rendered frames to the screen instead of controller inputs to the console, and then onto the screen.

Practically speaking, at the point where you have a dongle - which will likely need external power - and you have a device that needs charging... just dock the thing. It's a simpler solution, and you're not stuck with a Wii U GamePad sized controller.
 
0
Raccoon said:
I've been dismissive of the possibility of a dockless system (and I indeed don't pay much mind to the "Super Play" Tweets), but I've never actually asked the siliconheads here if wireless casting is actually feasible. I assume there'd be proprietary technology involved, but is there any reason to believe that latency wouldn't be untenably poor regardless?
Click to expand...
You can get some pretty high quality wireless video streams, but I'm not sure how much of that fits into a console price range. There's also the practical issue of not having wall power being supplied to the actual console in dockless scenarios, which is definitely a limiting factor.
ItWasMeantToBe19 said:
The leak is debunked from the beginning because there’s no chance GF would ever bother to miss the Christmas period for the DLC no matter how unpolished it was.

(GF is not really countering those claims by refusing to show anything related to DLC2, lol)

The leak’s clear implications were the leaker was implying that the Switch 2 would launch holiday 2023, and DLC2 would launch in January or February 2024 with a Switch 2 performance patch.

But the Switch 2 was clearly never intended to release holiday 2023.
Click to expand...
It is very obvious from how the release window of this DLC was being talked about that they were fully prepared for it to slip into next year. That was never really a point of contention with the rumor. Based on some recent oddities in the ranked ruleset schedule, the Indigo Disk release date was most likely in flux up until like the past month or so.
 
Reader mode
You must log in or register to reply here.
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