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

[Future Boy Nemo]

Spoiler: Revision vs Succ talk

I agree with you totally ILikeFeet, but we have a term for this, it's called a new console. A new video game hardware device that plays a significant number of exclusive games with more powerful hardware is called a new console. We have many of these releases, like the PS5 for example.

Trouble here is the GBC. Aside from the caveat that it's only 2x more powerful than the Game Boy whereas Drake will punch around 10x over Switch, it's still "more powerful hardware" and "plays a significant number of exclusive games". But Nintendo considers it a revision and bundles its sales figures and place in history with the Game Boy.

If they're afraid of cutting Switch's momentum by releasing Switch 2 too early, but it's too late to stop the train from pulling into the station, then it makes sense to market it as 'We will continue to support other Nintendo Switch models, this is just a new way to experience our great library of games with a few exclusives', then later ramp up those exclusives after a few years when they think the market will be more accepting of a full successor. Start as New 3DS, end up as GBA: that's essentially what happened with the Game Boy Color, though I don't know if it was intentional then.

Revision and successor have definitions, yes, but it doesn't matter if the rest of the world agrees on it if Nintendo never uses either term and just says 'This is the Nintendo Switch Plus, a new model in the Nintendo Switch family of systems,' and follows the life cycle I outlined above. It's all just marketing, and the marketing people don't care about calling a spade a spade, they'll call a spade a 'NEW ULTRA HD GARDENING TROWEL!', then change the color of the rubber handle a year later and sell it as 'Spade 2.0'. Anyone who cares will know, and anyone who doesn't know won't care. The important thing will always be 'Does this new system have enough value in my eyes to be worth the price?' Better performance on old games alone might be enough for some people, while others wait for more exclusives, or a sale below MSRP. I don't think anybody is going to refuse to buy it because they didn't formally call it a successor system and kill the OG Switch's legs for no reason.

Sorry to keep this conversation going, I started writing this post over an hour ago and now it seems a waste to delete it. My personal stance is that TL;DR: Nintendo will position it as whatever they want, but the hardware is too powerful not to end up acting as a succ eventually.

---

Getting back to technical topics, I'm out of my depth here a bit, but what does everyone think of a hypothetical second TX1 dieshrink, with Nintendo using the power savings to overclock it this time instead of saving on battery life, benefiting games with dynamic res and moving frame-rates?

I remember a couple of pages ago someone mentioned the idea of a Switch TV stick and the idea stuck with me, and a die shrink is the only way I can think of within Nintendo's control to lower the price of the Lite below $200 without losing money. At first I thought the R&D would be too much for a late life revision, but since Drake is likely to be $400 at minimum, a $169 or cheaper Switch Lite or $150 Switch TV USB stick (damn Joycons keeping the price over the magic $100 price point) that can play Switch Sports, Mario, Pokemon, Ring Fit, et cetera could sell several dozen million more Switch systems over several years and become the new PS2 in terms of legs after successor release. Can you imagine that combined with a new line of Nintendo Selects, releasing just in time for the holidays with a big marketing campaign? The increase in software sales would be worth the low hardware margins. But maybe I'm too optimistic about its potential.

small edit: Just realized this is still barely a technical topic, oh well.

 

[Hermii]

Spoiler: Revision vs Succ talk

Trouble here is the GBC. Aside from the caveat that it's only 2x more powerful than the Game Boy whereas Drake will punch around 10x over Switch, it's still "more powerful hardware" and "plays a significant number of exclusive games". But Nintendo considers it a revision and bundles its sales figures and place in history with the Game Boy.

If they're afraid of cutting Switch's momentum by releasing Switch 2 too early, but it's too late to stop the train from pulling into the station, then it makes sense to market it as 'We will continue to support other Nintendo Switch models, this is just a new way to experience our great library of games with a few exclusives', then later ramp up those exclusives after a few years when they think the market will be more accepting of a full successor. Start as New 3DS, end up as GBA: that's essentially what happened with the Game Boy Color, though I don't know if it was intentional then.

Revision and successor have definitions, yes, but it doesn't matter if the rest of the world agrees on it if Nintendo never uses either term and just says 'This is the Nintendo Switch Plus, a new model in the Nintendo Switch family of systems,' and follows the life cycle I outlined above. It's all just marketing, and the marketing people don't care about calling a spade a spade, they'll call a spade a 'NEW ULTRA HD GARDENING TROWEL!', then change the color of the rubber handle a year later and sell it as 'Spade 2.0'. Anyone who cares will know, and anyone who doesn't know won't care. The important thing will always be 'Does this new system have enough value in my eyes to be worth the price?' Better performance on old games alone might be enough for some people, while others wait for more exclusives, or a sale below MSRP. I don't think anybody is going to refuse to buy it because they didn't formally call it a successor system and kill the OG Switch's legs for no reason.

Sorry to keep this conversation going, I started writing this post over an hour ago and now it seems a waste to delete it. My personal stance is that TL;DR: Nintendo will position it as whatever they want, but the hardware is too powerful not to end up acting as a succ eventually.

---

Getting back to technical topics, I'm out of my depth here a bit, but what does everyone think of a hypothetical second TX1 dieshrink, with Nintendo using the power savings to overclock it this time instead of saving on battery life, benefiting games with dynamic res and moving frame-rates?

The idea in theory is just fine. Would provide a nice boost in power, and perfect BC for a minimal investment.

 

[ReddDreadtheLead]

I’m not sure if the A57 CPU cores are compatible with the 10nm process.

And Nintendo would have to pay to get the Maxwell 2.0 IP on the 10nm process, whether from Samsung or TSMC (the IP exists at both foundries but 16/14/12nm process I think).

i don’t think this is a cheap endeavor either.

 

[Look over there]

Setting aside whether it's financially worthwhile or not, A57 on a 10nm or newer generation node should be doable on the Samsung side at least.

Unless Samsung pulled that out of their rear end last year

 

[Pokemaniac]

Spoiler: Revision vs Succ talk

Trouble here is the GBC. Aside from the caveat that it's only 2x more powerful than the Game Boy whereas Drake will punch around 10x over Switch, it's still "more powerful hardware" and "plays a significant number of exclusive games". But Nintendo considers it a revision and bundles its sales figures and place in history with the Game Boy.

If they're afraid of cutting Switch's momentum by releasing Switch 2 too early, but it's too late to stop the train from pulling into the station, then it makes sense to market it as 'We will continue to support other Nintendo Switch models, this is just a new way to experience our great library of games with a few exclusives', then later ramp up those exclusives after a few years when they think the market will be more accepting of a full successor. Start as New 3DS, end up as GBA: that's essentially what happened with the Game Boy Color, though I don't know if it was intentional then.

Revision and successor have definitions, yes, but it doesn't matter if the rest of the world agrees on it if Nintendo never uses either term and just says 'This is the Nintendo Switch Plus, a new model in the Nintendo Switch family of systems,' and follows the life cycle I outlined above. It's all just marketing, and the marketing people don't care about calling a spade a spade, they'll call a spade a 'NEW ULTRA HD GARDENING TROWEL!', then change the color of the rubber handle a year later and sell it as 'Spade 2.0'. Anyone who cares will know, and anyone who doesn't know won't care. The important thing will always be 'Does this new system have enough value in my eyes to be worth the price?' Better performance on old games alone might be enough for some people, while others wait for more exclusives, or a sale below MSRP. I don't think anybody is going to refuse to buy it because they didn't formally call it a successor system and kill the OG Switch's legs for no reason.

Sorry to keep this conversation going, I started writing this post over an hour ago and now it seems a waste to delete it. My personal stance is that TL;DR: Nintendo will position it as whatever they want, but the hardware is too powerful not to end up acting as a succ eventually.

---

Getting back to technical topics, I'm out of my depth here a bit, but what does everyone think of a hypothetical second TX1 dieshrink, with Nintendo using the power savings to overclock it this time instead of saving on battery life, benefiting games with dynamic res and moving frame-rates?

I remember a couple of pages ago someone mentioned the idea of a Switch TV stick and the idea stuck with me, and a die shrink is the only way I can think of within Nintendo's control to lower the price of the Lite below $200 without losing money. At first I thought the R&D would be too much for a late life revision, but since Drake is likely to be $400 at minimum, a $169 or cheaper Switch Lite or $150 Switch TV USB stick (damn Joycons keeping the price over the magic $100 price point) that can play Switch Sports, Mario, Pokemon, Ring Fit, et cetera could sell several dozen million more Switch systems over several years and become the new PS2 in terms of legs after successor release. Can you imagine that combined with a new line of Nintendo Selects, releasing just in time for the holidays with a big marketing campaign? The increase in software sales would be worth the low hardware margins. But maybe I'm too optimistic about its potential.

small edit: Just realized this is still barely a technical topic, oh well.

We're past the point where I'd expect boosted clocks for any TX1 derived Switch. If they die shrink it again, it will probably be either because of availability or they want to go full passive cooling.

 

[ILikeFeet]

shrinking the A57 is the definition of squeezing blood from a stone. for the work required, literally using any other ARM is better in every way

 

[Deleted member 887]

Getting back to technical topics, I'm out of my depth here a bit, but what does everyone think of a hypothetical second TX1 dieshrink, with Nintendo using the power savings to overclock it this time instead of saving on battery life, benefiting games with dynamic res and moving frame-rates?

Not off-topic at all - a lot of folks (including me!) expected this when early rumblings of a revision started to come out, before it became clear that wasn't Nintendo's direction

There are some reasons that it wouldn't be as great an idea as it seems. The power boost would be modest. Nintendo would be the only real customer for the chip, making it expensive. It would lock Nintendo into their current architecture for a few more years, meaning no DLSS. Plus, the Maxwell architecture is old and games that use it and need faster clocks just means that Nintendo will need an even bigger jump the next time for backwards compat alone.

 

[Dakhil]

Has kopite reacted to the Drake leaks yet?

Outside of saying the illegal Nvidia leaks contain some files on T239, no. There are probably too much legal ramifications for kopite7kimi's source to divulge about the contents of the illegal Nvidia leaks without the risk of being fired.

I have no insight into fabs economics but I assume that those switches must have happened in the past for big volumes, yes? If it is the case, do we have any report of how these translated in terms of R&D expensives for the fabless customer(s)? I know that the jump from DUV to EUV makes past comparisons moot but it would still give us a point of reference.

Probably. But there's an article from a couple of months ago talking about how manufacturing costs can increase for EUV photomasks due to chip designs becoming increasingly complex, as well as how complex using EUV lithography can be.

Still, I think Nvidia will make it an objective to increase the margins it will make out of Drake. I hope this translates into them and Nintendo going for the cutting edge in terms of node.

Keep in mind that Nvidia increasing profit margins can also mean increasing the price of the SoC. Also keep in mind Nvidia's paying a very high premium to secure enough capacity from TSMC. Of course, I don't know how little profit Nintendo's willing to make initially before expecting higher profit.

Getting back to technical topics, I'm out of my depth here a bit, but what does everyone think of a hypothetical second TX1 dieshrink, with Nintendo using the power savings to overclock it this time instead of saving on battery life, benefiting games with dynamic res and moving frame-rates?

There was a questionable rumour talking about Nintendo apparently asking Nvidia and TSMC to provide a Tegra X1 sample that was die shrunk using TSMC's N7 process node, and to increase the Cortex-A57's frequency from 1.9 GHz to 2.52 GHz, although Nvidia and TSMC apparently warned Nintendo about the difficulty of increasing the Cortex-A57's frequency past 1.9 GHz. And apparently, the power consumption of the die shrunk Tegra X1 sample was higher than anticipated when the Cortex-A57 was running at a frequency of 2.52 GHz.

 

[ReddDreadtheLead]

Setting aside whether it's financially worthwhile or not, A57 on a 10nm or newer generation node should be doable on the Samsung side at least.

Unless Samsung pulled that out of their rear end last year

This would be delving into the financial aspect of it, but wouldn’t their engineers need to familiarize themsleves with the A57 ARM IP to replicate it on the newer node process?

Either way, rebuilding a chip is more costly than just making a new one from scratch.

I don’t see why they wouldn’t just go with an A73 quad core cluster and just have 2 Maxwell 2.0 SMS from the tegra X1 and consider that the “pro” edition. Sure it has a newer generation of a CPU but it’s more affordable. Maybe have two memory channels for a 50GB/s bandwidth and an increase of RAM to like 6GB.

And no company intentionally tries to take a more expensive route to reach a result if they can get something better for cheaper, even Sony and Microsoft.

 

[ILikeFeet]

And no company intentionally tries to take a more expensive route to reach a result if they can get something better for cheaper, even Sony and Microsoft.

bUt nIntEnDo!

granted, the Wii U was literally the only option for what they wanted. just turned out what they wanted wasn't very good

 

[Look over there]

For such a hypothetical device (a cheaper revision+ entry option for the long haul), it might be worthwhile to update the memory controller too. I'm not sure how long plain LPDDR4 will remain in production; probably better to go with a single 64-bit LPDDR5/5X module.
...also, quick check with catalogs show that Samsung and Micron's 'LPDDR4' pages all show 4X speeds. Similar enough with SK Hynix. When was the last time plain LPDDR4 was displayed in a catalog?

 

[Deleted member 1324]

I was under the impression that Nate had said there was a sizeable update regarding the new console coming on that podcast and that was why people were so hyped for it and it took longer to arrive. He basically said he expects it to be between an Xbox One S and a PS4 Pro.

Very disappointing if it’s closer to the lower end of his expectations as XBOS was like ~5% more powerful than the OG XBO. We’ve went from predicting a 6x leap for the next Switch over the current console after that Nvidia info coming out to potentially having a system hitting it’s highest bound dynamic resolution limit and more stable framerates

 

[Look over there]

Oh, Nate more recently said that such an update would be coming... sometime later. Relevant post.

 

[LoneRanger]

I was under the impression that Nate had said there was a sizeable update regarding the new console coming on that podcast and that was why people were so hyped for it and it took longer to arrive. He basically said he expects it to be between an Xbox One S and a PS4 Pro.

Very disappointing if it’s closer to the lower end of his expectations as XBOS was like ~5% more powerful than the OG XBO. We’ve went from predicting a 6x leap for the next Switch over the current console after that Nvidia info coming out to potentially having a system hitting it’s highest bound dynamic resolution limit and more stable framerates

Having a raw GPU number similar to XOne S (~1.4 TFlops) that with DLSS it can produce an image equivalent to a PS4 Pro/XOne X is still an impressive jump (note: this all about docked mode). Add new architectural features + RT core + better CPU/RAM and it would be a very competitive device.

It also lineup what our french poster said a long time ago (iirc late 2020).

 

[Deleted member 1324]

Just got a 3060 upgrade for my PC. My life is now Deep Learned Super Sampled (or Deep Learned Anti Aliased, depending on which method).

Playing Control at max settings + max ray-tracing with DLSS is definitely an eye opener. Yes, there is the "DLSS fuzziness" that graphics snobs often highlight, but given the game already has tons of post processing like film grain and seeing it run at ridiculous framerates, I really don't mind the trade off since, if anything, it's actually free anti-aliasing. And that's not even getting into DLAA, which is basically what we could see should "Drake" ever manage to hit 4k60 native for some titles. My guess is it would be very useful for games like Mario Kart which don't rely on too many fancy post-processing and lighting effects (Tensor Cores acting as the hardware accelerated AA).

In short, while I think it's prudent to temper expectations (DLSS doesn't come free, after all, and there's the bandwidth constraints too), I think I'm sufficiently excited to see this tech being utilized in a low powered handheld.

Nice. Hope you enjoy your new toy!!

 

[Sol]

I was under the impression that Nate had said there was a sizeable update regarding the new console coming on that podcast and that was why people were so hyped for it and it took longer to arrive. He basically said he expects it to be between an Xbox One S and a PS4 Pro.

Very disappointing if it’s closer to the lower end of his expectations as XBOS was like ~5% more powerful than the OG XBO. We’ve went from predicting a 6x leap for the next Switch over the current console after that Nvidia info coming out to potentially having a system hitting it’s highest bound dynamic resolution limit and more stable framerates

This is what confuses me.

Going from the Tegra X1 to a brand new, next-generation custom Nvidia SoC with A78 CPU cores, RT cores, DLSS, and a massive 12 SM Ampere based GPU, is not a "revision" in terms of the hardware. That's a full next-gen leap with mammoth increases in computational horsepower and new technologies. It's in no way comparable to either the 3DS --> New 3DS (which had the exact same GPU), or the Xbox One to Xbox One S (which is literally the exact same chip).

It's not even comparable to PS4 --> PS4 Pro. The PS4 Pro is using the same Jaguar CPU cores (just upclocked), and has the same GCN based GPU (just larger).

So if all the leaked information we have is correct, I really don't see how Drake is in any way comparable to the New 3DS, Xbox One S, or PS4 Pro. It's far more substantial.

And in terms of marketing or branding of the console, selling it as a "revision" to consumers wouldn't make much sense to me. Again, given the internal specs and capabilities of the system, this is basically the latest and greatest that Nvidia and Nintendo have come up with.

So after this releases, that's it for a while. No custom SoC that Nvidia works on after the release of Drake will be able to significantly go beyond it at the same price point. Sure, they could build a $2,000 SoC that could blow Drake away, but in the near term, there's nothing Nvidia and Nintendo could release after Drake for 4-5 years that would offer substantial, across the board increases in capabilities.

So the idea that this is just going to be a Pro or New Switch, and then Nintendo's real next-gen platform is coming in 2024 or 2025 just doesn't add up. If that were to happen, then Drake would end up being much more of a generational leap coming from the Switch than this next platform would be succeeding Drake. We would be looking at ~2028 as the earliest that Nintendo would release a true next-gen successor for the Switch platform. Which means that eventually this revision unit would have to become the standard, default, baseline Switch - unless Nintendo wants to keep making Switch games for another 6 years.

So....yeah. After seeing the 12 SM leaks, along with everything else about this being a brand new custom chip with DLSS and RT, I'm going with the notion that this is a full blown next-generation system from Nintendo, and will be positioned and marketed as such.

 

[Deleted member 1324]

Oh, Nate more recently said that such an update would be coming... sometime later. Relevant post.

Ah my bad. Thanks.

 

[Hosermess]

Yeah the 2024-25 thing for the next system for Nintendo was a bit odd, I guess he could just be going by the average lifespan for past revisions, but who knows. Considering he hasn't heard anything recently to confirm his previous Drake timeline. I would guess it was nothing more than just a date he threw out there and not based on some neat info he had gathered.

 

[ReddDreadtheLead]

Did he mean One S or One X

Because these are two very different pieces of kit

 

[mariodk18]

Nintendo has made exclusives for every version of upgraded hardware. And not just small games either. The Famicom Disk System had Zelda and Metroid. Th Satellaview had its own software including remixes of previous games. Both the NES and SNES had more advanced chips that could be added to cartridges to boost performance or add new effects. The N64 DD was meant to have sequels to the biggest N64 games (SM64 2, Zelda, Animal Crossing).

With regards to handhelds, GBC had enhanced ports and exclusives from Nintendo and third parties. DSi had its own online store for downloadable games. Even the 3DS had a few exclusives from Nintendo (although they were down ports).

Looking at Nintendo's history, they always want to have some kind of software to sell their hardware. Now with the Switch and being in the HD era, dev times take much longer and Iwata stated one of the goals for Switch and beyond was to be sure that software carries forward to ensure their is no drastic split in userbase. What this means is that when Drake launches, Nintendo will want to show off its features with titles that take advantage of its capabilities. GBC was able to add color to GB games (although purely aesthetic) with some games even taking special advantage, such as having multiple colors. Similarly, even though the n3DS didn't launch with exclusive hardware, many titles took advantage of faster loading or improved 3D with head tracking. There were also several titles that ran smoother or had higher framerates, but Nintendo themselves would actually wait until the following year to an exclusive title: Xenoblade 3D.

So it will be interesting to see what Nintendo does this time for Drake. They no longer have a more powerful system's library to take advantage of for downports. And it's because of that that we won't see Nintendo release any major exclusive for a while. It will be difficult to abandon a large userbase (also similar to Iwata's desire) when their previous console was so successful and when dev times for big HD games take several years. The reason Microsoft is moving to Series exclusives more quickly is because their previous console didn't do as well as they wanted. It's the same reason Nintendo abandoned the Wii U. It's easier to leave behind a userbase that doesn't exist.

Considering the above, Nintendo will do as Sony has done and release Switch titles with extra bells and whistles to entice power users in its initial days. Third parties will fill in the role of bringing lower risk games from older consoles the Switch couldn't handle. That way, Nintendo obtains the best of both worlds.

So when will Nintendo release their own exclusives to take full advantage of this new hardware? I find it difficult to believe that Nintendo would only give out dev kits for third parties to utilize- especially with what seems like a massive leap in power and feature set.

The trick is to follow the software. They are a software and hardware company and they ultimately will want you to buy their hardware with their games. What console do you think Botw 3 will release on? The next Animal Crossing? Splatoon 4? Smash? Mario Kart 10, Odyssey 3, Metroid Prime 5? The earliest many of these titles would likely release (especially with the pandemic) is 2024/2025 (Animal Crossing, Mario Kart) giving an average 3 year dev time + 1 for COVID. Speaking of 2025, that's when we would be due for a new generation for Pokemon (gen 10). After 7+ mainline Pokemon games, it's likely GF would be ready to move on to new hardware. All of this makes the most likely launch for hardware late 2022/ early 2023 as many have been saying.

Nintendo doesn't want to launch too late and lose momentum. They also would prefer to go for more of a soft launch given that they wouldn't be able to fulfill demand in within the first 1-2 years. Let the hardcore audience and scalpers fight it out at the start and then start rolling in with the big market stuff. No Nintendo console has ever launched with Mario Kart or a new Pokemon gen; those titles tend to launch a year or two after enough of an install base exists to capitalize on.

Taking everything into account, a launch for Drake within the next few years would:

• Keep Switch momentum and mindshare high
• Allow older games to have increased sales with patches
• Allow Nintendo platforms to continue receiving third party support
• Build an install base for the next 5-7 years

tldr: Follow the software breadcrumb trail and you will find the most likely release date for the next Nintendo hardware (in this case, Drake)

 

[Pokemaniac]

Yeah the 2024-25 thing for the next system for Nintendo was a bit odd, I guess he could just be going by the average lifespan for past revisions, but who knows. Considering he hasn't heard anything recently to confirm his previous Drake timeline. I would guess it was nothing more than just a date he threw out there and not based on some neat info he had gathered.

If Nintendo wants to fully embrace the rolling generations approach, another new system around 2025-2026 isn't unreasonable, but is a little on the earlier end. That's about when Nvidia's Atlan SoC is scheduled to be ready. It wouldn't be a "true successor" to the Switch or anything like that, though, it would be the same thing to Drake as Drake is to Switch 1.

 

[Zeroth]

This is what confuses me.

Going from the Tegra X1 to a brand new, next-generation custom Nvidia SoC with A78 CPU cores, RT cores, DLSS, and a massive 12 SM Ampere based GPU, is not a "revision" in terms of the hardware. That's a full next-gen leap with mammoth increases in computational horsepower and new technologies. It's in no way comparable to either the 3DS --> New 3DS (which had the exact same GPU), or the Xbox One to Xbox One S (which is literally the exact same chip).

It's not even comparable to PS4 --> PS4 Pro. The PS4 Pro is using the same Jaguar CPU cores (just upclocked), and has the same GCN based GPU (just larger).

So if all the leaked information we have is correct, I really don't see how Drake is in any way comparable to the New 3DS, Xbox One S, or PS4 Pro. It's far more substantial.

And in terms of marketing or branding of the console, selling it as a "revision" to consumers wouldn't make much sense to me. Again, given the internal specs and capabilities of the system, this is basically the latest and greatest that Nvidia and Nintendo have come up with.

So after this releases, that's it for a while. No custom SoC that Nvidia works on after the release of Drake will be able to significantly go beyond it at the same price point. Sure, they could build a $2,000 SoC that could blow Drake away, but in the near term, there's nothing Nvidia and Nintendo could release after Drake for 4-5 years that would offer substantial, across the board increases in capabilities.

So the idea that this is just going to be a Pro or New Switch, and then Nintendo's real next-gen platform is coming in 2024 or 2025 just doesn't add up. If that were to happen, then Drake would end up being much more of a generational leap coming from the Switch than this next platform would be succeeding Drake. We would be looking at ~2028 as the earliest that Nintendo would release a true next-gen successor for the Switch platform. Which means that eventually this revision unit would have to become the standard, default, baseline Switch - unless Nintendo wants to keep making Switch games for another 6 years.

So....yeah. After seeing the 12 SM leaks, along with everything else about this being a brand new custom chip with DLSS and RT, I'm going with the notion that this is a full blown next-generation system from Nintendo, and will be positioned and marketed as such.

I've stated this months ago. It makes 0 sense to create an all new console hardware and make it a revision, leaving room for the true successor later on.

What will probably happen is that Nintendo will not brand this as a Switch 2, but basically slowly phase out the current Switch and make Drake the successor in all but name. In that sense it would be a "Switch Pro" but for all purposes it would be a Switch 2.

 

[mariodk18]

I mean Nintendo is seemingly going to wait 6+ years to release its first upgrade for Switch, but then we’re also supposed to believe they’ll release the “true” next-gen successor 3-4 years after that? Either Drake is getting a massive delay to 2025 or the 2025 thing is speculation.

If he’s here, I would love to get a clarification on the “real” next-gen system in 2025 that Nate was talking about. That doesn’t make any sense because what software would Nintendo even have prepared for it? Is Nintendo not going to develop any software targeting Drake and only let 3Ps take advantage? Would the 2025 system have NVN3?

 

[Pokemaniac]

I mean Nintendo is seemingly going to wait 6+ years to release its first upgrade for Switch, but then we’re also supposed to believe they’ll release the “true” next-gen successor 3-4 years after that? Either Drake is getting a massive delay to 2025 or the 2025 thing is speculation.

If he’s here, I would love to get a clarification on the “real” next-gen system in 2025 that Nate was talking about. That doesn’t make any sense because what software would Nintendo even have prepared for it? Is Nintendo not going to develop any software targeting Drake and only let 3Ps take advantage? Would the 2025 system have NVN3?

What I think people are trying to do is extrapolate from previous Nintendo revision systems, when Drake really doesn't resemble them technically. There is no perfect comparison for Drake in Nintendo's history, and we're really best off not trying to put it in a box based on what they've done in the past.

That said, I don't think the idea of another system in 2025 is inherently unreasonable. It just probably isn't going to be positioned substantially differently from Drake if it happens unless Nintendo pulls a "third pillar".

 

[ReddDreadtheLead]

Would the 2025 system have NVN3?

presumably, yes it would use NVN3. It would be a different GPU after all. Having its own featureset.

 

[SiG]

Nice. Hope you enjoy your new toy!!

I am. Just got Metro Exodus since it was on sale. Had ray-tracing and settings at full-blast...DLSS on quality, of course, and maybe for a one or two instances where there were frames drop (but not of the stuttery type), the game ran very smooth.

Also discovered DLSS Swapper. Presumably the algorithm can be updated on a driver level or per-game basis, with the general model being improved with newer versions. Allowing older titles to retain the newest version compatible with it "without breaking" seems very doable, although I think with the upscaling/reconstruction input standards being set in place, hopefully this means the Drake wouldn't need to worry about subsequent games maintaining forward compatibility for succeeding generations.

 

[Branduil]

An additional all-new Nintendo system 3 years after Drake makes no sense. It took them maybe 6 years to move on from TX1, working on an entirely new chip with all-new architecture and features, and then they're just gonna use that on a revision before moving on in a couple years? You can debate how Nintendo is going to position Drake, but for all intents and purposes this IS their next system. You don't spend this much time and money on a new architecture to immediately abandon it.

 

[Bosintang]

I told you it would be 2024 at the earliest if the rumored, very powerful, system indeed materializes. Anything within the next 12 months ought to be a more limited upgrade, and I for one believe that the ship has sailed for such upgrade so I don't even expect it. But in fairness, I also believed that the ship had sailed for a Mario Kart DLC.
End of 2024 for that almost Series S level console is my current bet.

 

[Alovon11]

I told you it would be 2024 at the earliest if the rumored, very powerful, system indeed materializes. Anything within the next 12 months ought to be a more limited upgrade, and I for one believe that the ship has sailed for such upgrade so I don't even expect it. But in fairness, I also believed that the ship had sailed for a Mario Kart DLC.
End of 2024 for that almost Series S level console is my current bet.

Again, the timings of the data in the leak and previous leaks just don't line up with a 2024 window.

Do you think Nintendo would let developers have near or true final-devkits for 2 years?

And devkits in general for more than 4?

Drake itself is that very powerful system, it can reach Series S level when docked for the GPU if the cache adds enough IPC over ampere and it's clocked high enough (I even did the math earlier in the thread), and that is before DLSS which can take a system even a bit behind the Series S and bring it within striking distance of the PS5's native presentation (Resolution/FPS) GPU-wise if devs are clever enough.


So no, right now the window is still FY2023 (Now to Mid-Late March 2023 iirc), with a lean towards Early March 2023 because of a (logically assumed) connection between BOTW2 and Drake.

Although Nintendo could very well have a Mario game to pair with Drake's launch or just pair it with Pokemon and have Monolith or another studio handle the enhanced versions of the games if they wanted it out in 2022 to have manufacturing and supply somewhat ironed out by BOTW2's launch.

(Note that is a possibility, not that I am inclined to think that is the case atm, I am leaning towards BOTW2 pairing atm much to my sadness I would have to wait a third of a year after Xenoblade 3 to get it in 4K)

 

[Bosintang]

Again, the timings of the data in the leak and previous leaks just don't line up with a 2024 window.

Do you think Nintendo would let developers have near or true final-devkits for 2 years?

And devkits in general for more than 4?

Drake itself is that very powerful system, it can reach Series S level when docked for the GPU if the cache adds enough IPC over ampere and it's clocked high enough (I even did the math earlier in the thread), and that is before DLSS which can take a system even a bit behind the Series S and bring it within striking distance of the PS5's native presentation (Resolution/FPS) GPU-wise if devs are clever enough.


So no, right now the window is still FY2023 (Now to Mid-Late March 2023 iirc), with a lean towards Early March 2023 because of a (logically assumed) connection between BOTW2 and Drake.

Although Nintendo could very well have a Mario game to pair with Drake's launch or just pair it with Pokemon and have Monolith or another studio handle the enhanced versions of the games if they wanted it out in 2022 to have manufacturing and supply somewhat ironed out by BOTW2's launch.

(Note that is a possibility, not that I am inclined to think that is the case atm, I am leaning towards BOTW2 pairing atm much to my sadness I would have to wait a third of a year after Xenoblade 3 to get it in 4K)

You'll see.

 

[Alovon11]

You'll see.

That's not a great rebuttal there

Anyway there isn't really anything to pair Drake with past BOTW, nothing as big as the 2022-March 2023 slate or BOTW2 on its own

And the same thing sort of applies for any 2024/2025 hardware, the next biggest thing to potentially launch with it would be Mario Kart 9/10 or a new Mario but the Odyssey team's next project likely is a 2022/2023 title in correspondence with the Mario Movie so that likely can't be it either

 

[Hermii]

This would be delving into the financial aspect of it, but wouldn’t their engineers need to familiarize themsleves with the A57 ARM IP to replicate it on the newer node process?

Either way, rebuilding a chip is more costly than just making a new one from scratch.

I don’t see why they wouldn’t just go with an A73 quad core cluster and just have 2 Maxwell 2.0 SMS from the tegra X1 and consider that the “pro” edition. Sure it has a newer generation of a CPU but it’s more affordable. Maybe have two memory channels for a 50GB/s bandwidth and an increase of RAM to like 6GB.

And no company intentionally tries to take a more expensive route to reach a result if they can get something better for cheaper, even Sony and Microsoft.

I remember this argument that a new chip is cheaper than a die shrink was used back in the day, for why Mariko wouldn’t be a die shrink. We know how that turned out.

Every console in history had die shrinks, so how can that be true?

 

[Bosintang]

That's not a great rebuttal there

Anyway there isn't really anything to pair Drake with past BOTW, nothing as big as the 2022-March 2023 slate or BOTW2 on its own

I'm not invested enough in that topic, sorry. Nevertheless, I'm sure it'll be 2024 if the device is indeed the one rumored, and this forum's favorite podcaster appears to be of a similar opinion.

 

[Alovon11]

I'm not invested enough in that topic, sorry. Nevertheless, I'm sure it'll be 2024 if the device is indeed the one rumored, and this forum's favorite podcaster appears to be of a similar opinion.

Way to shutdown a legitimate point though.

That's not how debate works.

And also, are you talking about Grubb? Honestly I don't see any point in Grubb's ideas here.

The NVIDIA Hack exposed Drake and NVN2, and the state of the data implies they are not 2+ years out.

Heck NVN2 at this point has had all Maxwell/TX1 hook-ins purged in favor of Turing or Ampere-Derived hooks

 

[Bosintang]

Way to shutdown a legitimate point though.

That's not how debate works.

And also, are you talking about Grubb? Honestly I don't see any point in Grubb's ideas here.

The NVIDIA Hack exposed Drake and NVN2, and the state of the data implies they are not 2+ years out.

Heck NVN2 at this point has had all Maxwell/TX1 hook-ins purged in favor of Turing or Ampere-Derived hooks

Once again, I'm not invested enough to "debate". I expressed an opinion, you disagree with it, you wish to argue about it, I don't, let's leave it at that. It's really no big deal.

 

[Alovon11]

Once again, I'm not invested enough to "debate". I expressed an opinion, you disagree with it, you wish to argue about it, I don't, let's leave it at that. It's really no big deal.

Alright, but still I don't see how you can look at the evidence and logically think that it would be that far out still.

You can't even "Because Nintendo" it because NVIDIA is making the SoC and the API so they sort of are dragging Nintnedo kicking and screaming into the next hardware generation.

Also not to mention Mariko would be entering EoL at the production level in 2024/2025 because LPDDR4X will be winding down.

Kinda can't make your SoC if the RAM for it is ending production

That's my last point in this specific discussion, make of it what you will

 

[mariodk18]

Once again, I'm not invested enough to "debate". I expressed an opinion, you disagree with it, you wish to argue about it, I don't, let's leave it at that. It's really no big deal.

You don’t want to debate on a forum?

 

[ReddDreadtheLead]

I remember this argument that a new chip is cheaper than a die shrink was used back in the day, for why Mariko wouldn’t be a die shrink. We know how that turned out.

Every console in history had die shrinks, so how can that be true?

Shrinking a chip for a node it has no IP on and has to essentially be rebuilt is not comparable to shrinking a chip into an existing pathway.

No one said no console has had a die shrink, simply that if they are going to bother with having a TX1 on a newer node, or even swapping fabs, that isn’t going to be at all cheap.

Let alone that a Tegra X1 at SEC will not be the same as a TX1 at TSMC due to the intricacies of said chip.

They’re better off reserving that R&D into an actual new chip

and they didn’t even shrink the TX1 again for the OLED, they used the already shrunk TX1.

 

[Mr Swine]

Only way Drake is only a marginal upgrade is if the 12SM GPU and CPU are downclocked so is roughly only 2x as powerful as Switch. But then why not do that with Erista?

 

[Hermii]

Only way Drake is only a marginal upgrade is if the 12SM GPU and CPU are downclocked so is roughly only 2x as powerful as Switch. But then why not do that with Erista?

Or why go for 12sm? This is a custom chip, made for Nintendo. Nintendo could have gone for 2SM if they wanted to.

That would just have been a needlessly expensive and large chip, for the same power level they could have gotten with a much smaller chip with higher clocks-

 

[Hermii]

Shrinking a chip for a node it has no IP on and has to essentially be rebuilt is not comparable to shrinking a chip into an existing pathway.

No one said no console has had a die shrink, simply that if they are going to bother with having a TX1 on a newer node, or even swapping fabs, that isn’t going to be at all cheap.

Let alone that a Tegra X1 at SEC will not be the same as a TX1 at TSMC due to the intricacies of said chip.

They’re better off reserving that R&D into an actual new chip

and they didn’t even shrink the TX1 again for the OLED, they used the already shrunk TX1.

I was just saying, I heard the argument before and it was wrong back then.

But that doesnt mean its not right this time. Thanks for the detailed reply

 

[Deleted member 1141]

Once again, I'm not invested enough to "debate". I expressed an opinion, you disagree with it, you wish to argue about it, I don't, let's leave it at that. It's really no big deal.

I mean, the problem is you're spreading information without giving full reasoning why you're saying this, and most of the information you've been saying amounts to "you'll see" and "it'll be about series s level by end 2024, but not before then". Like, that's not how any of this works. Nvidia literally has the chips up for pre-order for consumers and developers for release later this year and Nintendo just had a job listing for testing of finalized hardware and controllers/accessories for "mass production". Like, to me that seems like next year is the target at the latest. Plus with all the leaks, we know it's Orin, especially due to the Nvidia hack.

Another thing is Nintendo and Nvidia already reserved chip production way ahead of time and it's not something they can just cancel on a whim without serious consequences.

I don't get where you're coming from at all, especially with you're "you'll see".

Like, if you're going to post here like this, people are going to call you out and also push you to address why you think this way.

 

[ReddDreadtheLead]

Nintendo would have to almost turn the device off if they want to get even close to “twice the power” of Mariko

And waste battery life in the process since you can only go down so much before you just start to have leakage for powering nothing

I don’t even think the CPU can be clocked low enough for that to be feasible considering the CPU, A78 at 1GHz, is the equivalent of 3 A57 CPUs at 1GHz, pulls ahead a bit too.

Now imagine 7 available for games? What are you going to clock it to? 21MHz like the SNES?

I suppose calling the the Super Nintendo’s Enervated Switch or SNES makes sense here

I was just saying, I heard the argument before and it was wrong back then.

But that doesnt mean its not right this time. Thanks for the detailed reply

I don’t really find the application of that comparable at all. It’s implying Nintendo shrunk the TX1 again after shrinking it already for the OLED model when that clearly isn’t what they did.

 

[ShaunSwitch]

I would say I find the reasoning behind Drake being a 2024 chip bizarre at this point, if there were any actual reasoning behind it.

Reasons Drake is likely to launch within a year:

1. Its based on Orin, which is available later this year.

2. Even if its on a more advanced node, we know Nvidia has advanced nodes reserved at TSMC. No need to wait for capacity to free up if its already reserved.

3. Killer game line up for late 2022- early 2023, similar to the original launch line up of the switch.

4. Dev kits have been rumoured to be in devs hands since late 2020 (correct me if I'm wrong.)

5. Job postings related to mass production by Nintendo. Arguably this could be for a new accessory or mini console.

6. The trusted insiders on this board have always pointed to late 2022 with a slip to early 2023 being possible.

The only way I see this time range being inaccurate is if Drake is cancelled somehow, but with the NVN2 leak information and the fact the next device would have to be Atlan based in 2026 I don't see this being true.

 

[Hermii]

Nintendo would have to almost turn the device off if they want to get even close to “twice the power” of Mariko

And waste battery life in the process since you can only go down so much before you just start to have leakage for powering nothing

I don’t even think the CPU can be clocked low enough for that to be feasible considering the CPU, A78 at 1GHz, is the equivalent of 3 A57 CPUs at 1GHz, pulls ahead a bit too.

Now imagine 7 available for games? What are you going to clock it to? 21MHz like the SNES?

I suppose calling the the Super Nintendo’s Enervated Switch or SNES makes sense here

I don’t really find the application of that comparable at all. It’s implying Nintendo shrunk the TX1 again after shrinking it already for the OLED model when that clearly isn’t what they did.

Off course we know Nintendo didn’t go this route. This is purely a hypothetical alt history scenario.

 

[Kenka]

12 SM is way much more than what is inside Erista/Mariko. If you are Nvidia and want to keep the succ's soc yields as high as the one for the current Switch, you have to logically choose a smaller node. How low? I have no idea. But a very dumb and rough assumption would be that 12 SM = a die 6 times bigger. So the node should be sqrt(6) smaller.

That's 14nm/sqrt(6) which is 6 nm (or 5.7 actually). So a hypothetical TSMC 6 nm node would do the job according to my assy fraction. How does that compare to Samsung 8 nm? No idea.

I repeat myself but I wish we had more insights inside the economics of this.

Keep in mind also that the way the nodes are named has nothing to do anymore with the wavelength of the beam of light used for lithography. So mostly, discussing nodes is pretty much futile.

 

[Ojoloco]

I would say I find the reasoning behind Drake being a 2024 chip bizarre at this point, if there were any actual reasoning behind it.

Reasons Drake is likely to launch within a year:

1. Its based on Orin, which is available later this year.

2. Even if its on a more advanced node, we know Nvidia has advanced nodes reserved at TSMC. No need to wait for capacity to free up if its already reserved.

3. Killer game line up for late 2022- early 2023, similar to the original launch line up of the switch.

4. Dev kits have been rumoured to be in devs hands since late 2020 (correct me if I'm wrong.)

5. Job postings related to mass production by Nintendo. Arguably this could be for a new accessory or mini console.

6. The trusted insiders on this board have always pointed to late 2022 with a slip to early 2023 being possible.

The only way I see this time range being inaccurate is if Drake is cancelled somehow, but with the NVN2 leak information and the fact the next device would have to be Atlan based in 2026 I don't see this being true.

Did you have sources for the point 2 and 5?

 

[Hermii]

12 SM is way much more than what is inside Erista/Mariko. If you are Nvidia and want to keep the succ's soc yields as high as the one for the current Switch, you have to logically choose a smaller node. How low? I have no idea. But a very dumb and rough assumption would be that 12 SM = a die 12 times bigger. So the node should be sqrt(12) smaller.

That's 20nm/sqrt(12) which is 14 nm. Since that is the node for Mariko, then we can divide by sqrt(12) again. That gives us a node of 10 nm, So a hypothetical TSMC 10nm node would do the job according to my assy fraction. How does that compare to Samsung 8 nm? No idea.

I repeat myself but I wish we had more insights inside the economics of this.

Keep in mind also that the way the nodes are named has nothing to do anymore with the wavelength of the beam of light used for lithography. So mostly, talking about nodes (on this board at least), is pretty much futile.

TX1 is 2sm, so its 6 times bigger.

And your napkin math is all wrong, it doesnt scale linearly.

 

[Deleted member 1141]

Did you have sources for the point 2 and 5?

Here on the job listing info:

キャリア採用：募集要項 - ハードウェアエンジニア（京都勤務）｜採用情報｜任天堂

任天堂株式会社のキャリア採用：募集要項 - ハードウェアエンジニア（京都勤務）のページです。

www.nintendo.co.jp

As for the other one, I assume is the recent deal Nvidia made with TSMC, which I don't have an immediate source on. It may or may not be Nintendo related though. Either way, still worth mentioning.

 

[Kenka]

TX1 is 2sm, so its 6 times bigger.

And your napkin math is all wrong, it doesnt scale linearly.

I knew that the assumptions were wrong but thank you for pointing out that I can't count. Meh. I edited my post to make it at least coherent.
Do you have a methodology? Just to satisfy my own curiosity.