• 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.
  • Furukawa Speaks! We discuss the announcement of the Nintendo Switch Successor and our June Direct Predictions on the new episode of the Famiboards Discussion Club! Check it out here!

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)

As if it wasn’t bad enough to wake up to this as an Xbox exec: https://www.bloomberg.com/news/arti...ly-posts-secret-game-plans-to-government-site

This just gets better and better lol

Ehh, it's super embarrassing, but there's nothing surprising there other than that Bethesda's management seemed really stupid (three year dev cycle projections even after Fallout 76, lmao)

The Xbox Next releasing 2028 with a huge focus on cloud integration and hardware that accelerates ML algorithms? Very expected.

The Series X dropping the disc drive is the only actual huge news.
 
I don't think the company that gave their 2D Mario team no deadline in early production, their massive Zelda project an extra year for polishing, and Nintendo Switch Sports five fucking years of off and on production would benefit by being owned by a content-starved wannabe juggernaut
Best case scenario, everyone resigns and Nintendo is left with a shell of a company. The new Nintendo becomes so westernized that Zelda literally becomes an AC clone and Mario Kart becomes a forza clone focused on realism.
 
Best case scenario, everyone resigns and Nintendo is left with a shell of a company. The new Nintendo becomes so westernized that Zelda literally becomes an AC clone and Mario Kart becomes a forza clone focused on realism.
yeah, it'd effectively be an IP sale
 
0
Hello everyone! Mike Odyssey here! Want to know more about the Switch 2? I've been covering a load of facts, not leaks or rumors but files by
Nintendo themselves on what the Switch 2 will be like! Here you go! and there is more where this comes from! Let me know if guys are interested in me sharing info I find here.

Is this a real post that I'm actually reading
 


Alex from DF and Bryan (from Nvidia) discuss briefly about the added cost of Ray Reconstruction (introduced on DLSS 3.5) on the timestamp - haven't finished watching the video yet, so there may be some other relevant snippets to [ ✂️ ].

Notably - Bryan states that Ray Reconstruction's cost currently sits at around 2ms on a 4090@2160p, which is "about 4x more expensive than Super Resolution". He then continues, explaining that is to be expected, since it's doing a lot more stuff. The focus of RR is on Image Quality, rather than performance.

So bringing the discussion back to Drake, I'm curious to see if RR really is worth the added overhead...? 🤔 I have a feeling that yes (especially since it's available on all RTX cards, even 2XXX), but it may only make sense on certain types of games or scenarios (i.e. Ray Tracing-heavy scenes or games).

Actually, side question: would it make sense to enable or disable DLSS features such as Ray Reconstruction on the fly, as an added tool for optimizations (for example, disable RR on scenes that don't benefit much from it)?
 
None of these things are competition, people aren't buying Nintendo products for ports, any other handheld out there would be lucky to sell in months what a Switch does in days
That is also true. Most people I know buy Nintendo hardware almost exclusively for Nintendo software. Although the ability to play a larger library of games certainly wouldn't hurt sales.
 
It's not the specs or processor, which were very good. It's the other side of the hardware.

From the other thread:



Microsoft didn't want to take responsibility for a long time, and by the time they did I got rid of it altogether.

The problem was also a topic of discussion in a Dutch consumer-rights television program, Kassa.

also the red ring bullshit. That was a mess
 
Is this a real post that I'm actually reading
Yeah the wording and video title... :)
That said, it's an interesting fun video with all these recents patents in one place. I hope the Wii U+ like streaming tech patent will make it to the final product.
 
0
Think he turned a corner somewhere, realised he was lost and thought "fuck it".
this thread is just getting filled with so much random shit, arguments and now self promo what is actually happening


I should be able to mass minimise all rattata tagged individuals just for the sake of readability
 
I know I’ve said it before, but September is just too late in my mind. If I had to guess, it’s gonna be launched somewhere in the March-June timeframe. Q1 of Nintendo’s current fiscal year (April-June) was massive for them, so unless they have some secret huge-hit Switch1 software hidden away, I feel they are going to need to do something to show shareholders they have a plan to achieve the infinite growth that they all seem to demand. (I’m very much looking forward to Princess Peach and the Paper Mario remaster, but I don’t think either of those titles will do all that much to move the needle.) By launching outside of the holidays, they could essentially add a second holiday-sized hardware sales period into their calendar, as well as allow for additional software to be released and sold (especially if they go the 1 big first party game per month cadence again.)

(Not to mention, that same launch template worked really well for Switch1 (and allowed for some big splash marketing like NoA’s Super Bowl commercial.))

That all said, the realities of hardware and software production schedules and timelines could have made a H1 release impractical or impossible, so it’s not like a H2 release would be shocking to me. If I were Head Uncle at Nintendo, though, I would be aiming to have it out before Summer.
if Nintendo hope to launch Switch sucessor in H1 2024, when we could expect Nintendo to anounce the console? Will they sacrifice the first 2024 Direct to anounce it next console?
 


Alex from DF and Bryan (from Nvidia) discuss briefly about the added cost of Ray Reconstruction (introduced on DLSS 3.5) on the timestamp - haven't finished watching the video yet, so there may be some other relevant snippets to [ ✂️ ].

Notably - Bryan states that Ray Reconstruction's cost currently sits at around 2ms on a 4090@2160p, which is "about 4x more expensive than Super Resolution". He then continues, explaining that is to be expected, since it's doing a lot more stuff. The focus of RR is on Image Quality, rather than performance.

So bringing the discussion back to Drake, I'm curious to see if RR really is worth the added overhead...? 🤔 I have a feeling that yes (especially since it's available on all RTX cards, even 2XXX), but it may only make sense on certain types of games or scenarios (i.e. Ray Tracing-heavy scenes or games).

Actually, side question: would it make sense to enable or disable DLSS features such as Ray Reconstruction on the fly, as an added tool for optimizations (for example, disable RR on scenes that don't benefit much from it)?

Couldn't you offset the cost of Ray Reconstruction by using less rays? If RR does reconstruction better, than in theory you could get away with using less rays to resolved the same image. The cost may not be as heavy then.

As far as the bolded, I doubt it would make much sense. RR would effectively replace the denoising pipeline in the renderer. Doesn't make much sense to have both a "normal" denoiser and RR. The denoiser would eat into GPU headroom and not look as nice as RR. Can't see a scenario where both would be worth it, but my knowledge is also incredibly limited.
 
I was talking about Microsoft. Xbox refresh will be on 6nm. People were questioning why the chose 6nm instead of 4.
It’s cheaper at their size. Remember they have a chip that is 360mm^2 (360? Hm, interesting number). Shrinking to 6nm would give them a cheaper result than shrinking to 4nm, because at 4nm it wouldn’t be small enough to make it cheap enough to bite the bullet.

If it was a small SoC, like say 120mm^2, shrinking to 4nm would have probably been a more worthy endeavor.
 
if Nintendo hope to launch Switch sucessor in H1 2024, when we could expect Nintendo to anounce the console? Will they sacrifice the first 2024 Direct to anounce it next console?
I think they'd still do a February or March Direct tbh. If this Direct is before or after the reveal then the contents would change. They could still do a February Direct and drop Prime 2, show new footage of Peach, LM2, TTYD, etc. If they do one after the reveal in March they could just give a few more insights into Switch 2 games and cover what I suggested before. The Direct would most likely still happen either way.
 
0
The question is not about its profitability but about its opportunity cost. Subscription services only appeal to very hardcore gamers so it's reasonable that Microsoft could just release indie titles and old backlog titles and have the same amount of subscriptions (as that would still be worth it to hardcore gamers). If that is the case, then Microsoft just lost hundreds of millions of dollars releasing Starfield directly on GamePass because now hardcore gamers don't have to pay them more whereas casual gamers won't maintain a subscription after Starfield (because it isn't worth it to them).

This is basically what Sony and Nintendo have reasoned out and is likely correct. NSO+ doesn't receive day one Nintendo games, but it's still worth it to me. Nintendo would just be throwing away money if they put Pikmin 4 on there day 1.
also Microsoft have to pay fortunes, to have games day one on Gamespass, Gamespass is not a substainable market strategy
 
0
Couldn't you offset the cost of Ray Reconstruction by using less rays? If RR does reconstruction better, than in theory you could get away with using less rays to resolved the same image. The cost may not be as heavy then.

As far as the bolded, I doubt it would make much sense. RR would effectively replace the denoising pipeline in the renderer. Doesn't make much sense to have both a "normal" denoiser and RR. The denoiser would eat into GPU headroom and not look as nice as RR. Can't see a scenario where both would be worth it, but my knowledge is also incredibly limited.
Yeah, that's the thing, it's 4x the cost of SR on its own, but that is replacing the old Denoiser, so that cost is sort of obfuscated a bit as good hand-tuned denoisers are not usually cheap.
 
Yep!

They don't seem to have reliable sources. If someone on the street tells you the tide will rise at midday, and the lifeguard says that high tide is 11am, and midday the tide is going out, which will you trust?
Wouldn't it be 10am in this analogy? The state of tide (i.e., the meaning of March) is still unknown.
 
Isn't there a Digital Foundry video in the same article going over Tegra Maxwell and mentioning 256 CUDA cores? wait nvm
Yes, someone else pointed that out to me too, which I also responded to saying DF went over Tegra X1 specs and nothing else (ie: storage, etc). Tegra X1 is not a custom chip, so the specs was known. The only thing DF was unsure about was which Tegra X1 variant would be used.
 
I predict we have no longer than 24 hours remaining before this forum splits into Famiboards March and Famiboard May, as the inevitable Team March and Team May side taking ensues, and polarisation spins, as it always does, out of control.

You incant with spells you cannot control, non-rodent.
Famiboards Summer (June to August) disagrees.
 


Alex from DF and Bryan (from Nvidia) discuss briefly about the added cost of Ray Reconstruction (introduced on DLSS 3.5) on the timestamp - haven't finished watching the video yet, so there may be some other relevant snippets to [ ✂️ ].

Notably - Bryan states that Ray Reconstruction's cost currently sits at around 2ms on a 4090@2160p, which is "about 4x more expensive than Super Resolution". He then continues, explaining that is to be expected, since it's doing a lot more stuff. The focus of RR is on Image Quality, rather than performance.

This is a great video. The missing piece of data we have about Drake is "what is the cost of current denoisers that might be used on Switch NG" and the only example I can think of is the UE5 denoiser. In short, if Fortnite gets DLSS 3 support, we can see if using RR is faster, with otherwise matched settings. If so, RR is probably a win for any title that would otherwise use RT on the hardware, even if it is expensive (because other denoisers are also expensive)

One thing to keep in mind is that the cost of DLSS across the board can be "hidden" via async rendering/compute, so it's not like a hard line.

Couldn't you offset the cost of Ray Reconstruction by using less rays? If RR does reconstruction better, than in theory you could get away with using less rays to resolved the same image. The cost may not be as heavy then.
It remains to be seen how well RR at low ray count compares to other denoisers are higher ray count. But it seems like one of the goals of RR is to let ray tracing scale with resolution. Currently, ray tracing resolution, input resolution, and output resolution are often set separately. If you're not using Super Resolution, then you probably want to set RT resolution lower than your rendering resolution for performance reasons. But because DLSS Super Resolution without RR interacts poorly with denoisers (good explanation of why early in the linked video, in fact), you are encouraged to do the opposite - to RT at your output resolution instead of the input.

RR not only produces higher quality denoised RT output, because the denoiser and super resolution are the same model, then Super Resolution can upscale your RT effects correctly. That means RTing at your input resolution, which is slower than without DLSS but faster than with DLSS, before RR was added.

Which is a sentence that is almost impossible to read and I apologize.
 
The Switch 2 is likely to be so bad at ray-tracing that DLSS 3.5's denoising probably wouldn't be helpful because few games will use quality or noticeable ray-tracing.

More of a Switch 3 thing.
???

I was under the impression Switch 2 would likely see both DLSS-SR and DLSS-RR usage? Have to admit I'm not deeply knowledgeable with DLSS tech in general.

Edit: oh i might have misread that, you said ray tracking and not ray reconstruction. Still not sure what that means, as I said I'm admittedly not deeply knowledgeable with graphical technology in general.
 
This is a great video. The missing piece of data we have about Drake is "what is the cost of current denoisers that might be used on Switch NG" and the only example I can think of is the UE5 denoiser. In short, if Fortnite gets DLSS 3 support, we can see if using RR is faster, with otherwise matched settings. If so, RR is probably a win for any title that would otherwise use RT on the hardware, even if it is expensive (because other denoisers are also expensive)

One thing to keep in mind is that the cost of DLSS across the board can be "hidden" via async rendering/compute, so it's not like a hard line.

It remains to be seen how well RR at low ray count compares to other denoisers are higher ray count. But it seems like one of the goals of RR is to let ray tracing scale with resolution. Currently, ray tracing resolution, input resolution, and output resolution are often set separately. If you're not using Super Resolution, then you probably want to set RT resolution lower than your rendering resolution for performance reasons. But because DLSS Super Resolution without RR interacts poorly with denoisers (good explanation of why early in the linked video, in fact), you are encouraged to do the opposite - to RT at your output resolution instead of the input.

RR not only produces higher quality denoised RT output, because the denoiser and super resolution are the same model, then Super Resolution can upscale your RT effects correctly. That means RTing at your input resolution, which is slower than without DLSS but faster than with DLSS, before RR was added.

Which is a sentence that is almost impossible to read and I apologize.
Worth mentioning that NVN2 had an early ray tracing sample with a toggle to cut the resolution in half and then scale it back up with DLSS. This was before RR was implemented (if indeed it is going to be implemented in NVN2, at launch anyway), but it shows a desired/desirable pattern of getting cheaper ray tracing by rendering it at input resolution before upscaling.
 
???

I was under the impression Switch 2 would likely see both DLSS-SR and DLSS-RR usage? Have to admit I'm not deeply knowledgeable with DLSS tech in general.

DLSS-RR is really good denoising for ray-tracing.

But I don't think it makes much sense if your ray-tracing itself is... bad.

Ray-tracing will also be a huge CPU hog due to BVH stuff as no custom hardware for BVHs has been created yet (it probably will soonish).
 
0
I don't think there is a correlation. If T239 is designed for 8nm, it will be 8nm. 5nm likewise.

Moving from 8nm to 5 is more than a die shrink, it's a complete redesign. Moving a chip from Samsung to Tsmc it pretty much needs to go back to the drawing board.

That being being said, from what I consider expert analysis in this thread, I think it would take a small miracle for 8nm 12sm to be running efficiently.
I feel like it is a complete redesign to change foundries especially but just given how big of a product the Switch is, I'm willing to bet NVIDIA is willing to put in the work. The Switch is one of the only products apart from datacenter and AI where I can see NVIDIA actually jumping through hoops to ensure industry-leading performance. If this next device has to compete with a freaking iPhone that runs Resident Evil Village and RE4 Remake and Death Stranding among other AAA games, I think there's a good chance NVIDIA pulls out all the stops for Nintendo.
 
The Switch 2 is likely to be so bad at ray-tracing that DLSS 3.5's denoising probably wouldn't be helpful because few games will use quality or noticeable ray-tracing.

More of a Switch 3 thing.

More pessimistic takes, it has already been confirmed by VGC that Switch 2 Ray Tracing capabilities will be superior to current-gen.

You consider the Pokemon 4chan post to be extremely accurate.

A post which said the Switch 2 was launching before or on the same day as SV DLC2.

SV DLC 2 is probably releasing in December.

We consider the 4chan post to be accurate to this date, let's see if his DLC take is real too

DLC 2 is releasing Winter 2023, which could be any day from December to March included.
 
Typical oldpuck W
Is that why things like RT reflections are more stable and more accurately reflect color and light in Phantom Liberty, since the denoiser retains more detail and Super Resolution properly upscales the RT effects now as a result?

Couldn't that also mean that you could claw a tiny bit of performance back by rendering the RT stuff at lower resolutions, with SR + RR closing the gap in resolution? Or is RR purely for image quality purposes? I haven't looked too much into it, admittedly.
 
I'm almost certain people will still use the Pokemon leak as credible when SV DLC2 launches in December with the Switch 2 still not shown.

What makes you think they going to release it in December ? They spaced it out by 6 months with Pokemon Sw/Sh
If they follow the same marketing pattern, the DLC pt 2 would release in Feb/Mar
 
Does it? I'm not getting 4-5 hours on Tears of the Kingdom on a single charge. I'm not saying you can't shrink AAA experiences down to a handheld at a robust level or that "last gen AAA" experiences can't get great battery life.

But as long as "AAA" means "boundary pushing for current gen" and Nintendo handheld means "comfortable mass market form factor", then it seems like that high battery life for those experiences specifically will always be out of bounds. But maybe what I'm really showing here is my disdain for AAA as a both a designation and a genre, more than the capabilities of the hardware.
Honestly now that I think about it I wouldn't mind playing ToTK on the new Switch with the same FSR/dynamic 720p graphics configuration as the OG Switch in handheld at a closer 30FPS lock and getting 5+ hours on a single charge without external batteries.
 
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