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

[olobolger]

I think the debate about whether it's a successor or a revision is going to continue even after the console is launched. I think the name and marketing, being continuist, is going to be ambiguous enough that people won't see it as one or the other (or both). From my point of view it's going to be a hybrid , and it's somehow the market and the circumstances what will determine how it will be defined in the future.

And this is because Nintendo, and many publishers, will continue to release games that are going to work on all models, albeit with improvements on the new SoC, and there will be publishers that will release certain games only for the new version. And I also think Nintendo will keep models with the current chip on the market (not as many, but maybe OLED, Lite and 2022/2023 model).

Maybe as time goes by, if the adoption of the new model is massive (as much as chip manufacturing allows it), the number of games released exclusively will increase, and then we could see it as if it had been a new generation. And then Nintendo could consider repeating the strategy (in 5/6 years time), perhaps by de-phasing the current models by then, or in a progressive way.

And if that doesn't happen and the percentage of users don't get so inclined towards this new version, then the number of exclusive games will be much lower, and we could see it as a Pro version. Whereas Nintendo would then fit their plans to have a new generation, and sold as such, in a shorter time range (3/4 years).

 

[ILikeFeet]

On this topic, does anyone have the results of the same benchmark for the A57?

NINTENDO Switch - Geekbench

Benchmark results for a NINTENDO Switch with an ARM ARMv8 processor.

browser.geekbench.com

 

[kvetcha]

NINTENDO Switch - Geekbench

Benchmark results for a NINTENDO Switch with an ARM ARMv8 processor.

browser.geekbench.com

So we’re looking at a 3x single core improvement and hoping that we get at least 6-8 cores, if not more?

 

[ILikeFeet]

So we’re looking at a 3x single core improvement and hoping that we get at least 6-8 cores, if not more?

8 cores at most

 

[Kenka]

NINTENDO Switch - Geekbench

Benchmark results for a NINTENDO Switch with an ARM ARMv8 processor.

browser.geekbench.com

An actual Switch benchmark? Unexpected but all the more beautiful.
So the increase in performance per core is line with what we envisioned.

 

[Thraktor]

I'm not talking about rumors, I mean actual data, information, release schedule, and the nvidia hack. All of it points to 2023. The device we saw is obviously not 8nm, and Nintendo does not have the ability to launch a 5nm system this year when they would have to compete with bigger dogs like Apple. BotW 2 is next year, which means a 2023 release would get it on the next console for its first holiday sales period, and the Mario Kart 8 DLC runs out in 2023, setting up nicely for a new Mario Kart game. I don't see what information at all points to a 2022 release beyond some guy on twitter told me it is.

Again, Nintendo isn't directly sourcing manufacturing capacity from TSMC or Samsung. Nvidia is, and they are, by any possible definition, one of the "big dogs". TSMC have been massively expanding their 5nm capacity this year and last, to the point where they will be manufacturing almost as many 5nm wafers as they make 7nm/6nm wafers by the end of this year. And given the density advantage of their 5nm processes, it will make TSMC 5nm by far the highest capacity process on a transistor basis (in fact it likely already is). Nvidia have confirmed that they've made multi-billion dollar pre-payments to secure capacity, and by the end of the year they'll be shipping the 814mm2 Hopper, and ~600mm2 and ~400mm2 Ada chips to customers manufactured on TSMC's N5 process (or Nvidia's 4N version of it), with the full lineup of Ada to follow. Drake, if manufactured on TSMC 5nm/4nm, would likely be around 80mm2, by comparison, and would take up a comparatively tiny number of wafers next to everything else Nvidia will be manufacturing on the same process.

That's not to say I believe it's definitely being manufactured on TSMC 5nm/4nm, and there are plenty of reasons they could choose to stick with Samsung 8nm. But the idea that "there isn't enough capacity" or "Nintendo can't secure the capacity" doesn't make sense. Taking die size into account, it's the highest capacity manufacturing process around by a large margin, and Nvidia, the company who would actually have to secure the capacity, have been able to secure enough capacity to move their entire HPC and gaming GPU lineup to TSMC 5nm in the same timeframe we're expecting the new Switch to launch.

 

[JoshuaJSlone]

I am mostly talking out of my ass but since Nintendo plan to sell 23 million consoles this fiscal year, we can assume they want the same output from their factory. So very crudely, they want to push 2 million devices per month or stockpile 6-8 million consoles for launch, if we follow your argument.

Switches are currently being produced at way higher rates than any major launch hardware has been, though, so pretty different from what a few months of release difference would make for 4K. After Switch's first 13 months on the market, 15 million had been shipped.

 

[BobNintendofan]

I think the debate about whether it's a successor or a revision is going to continue even after the console is launched.

Personally while I believe Nintendo will market it as both to try to have their cake and eat it too, with the massive power difference most of its game will likely have to be exclusive to the new console, so I wouldn't consider it a revision. It's already hard to port PS4 games to the switch and this thing looks a lot stronger than the PS4.

 

[Dakhil]

If this SOC is on 5NM it will outperform Series S after DLSS, that is crazy! With this performance VR form factor is possible.

Depends on which semiconductor foundry company Nintendo and Nvidia choose to work with for the fabrication of Drake, as well as what are the highest frequencies Nintendo chooses for the CPU and GPU on Drake, as well as the RAM, I think.

I imagine that the performance after enabling DLSS could be in the range of around Xbox Series S performance to being relatively close to Xbox One X performance if Nintendo and Nvidia decide to use Samsung's 5LPE process node for the fabrication of Drake. At lower frequencies, Samsung's 5LPE process node is practically on par with TSMC's N7P process node. But at higher frequencies, Samsung's 5LPE process node can consume 24% more power than TSMC's N7P process node.

And in the scenario where Nintendo and Nvidia decide to use TSMC's N5 process node for the fabrication of Drake, I imagine the performance after enabling DLSS could be at the very least be around Xbox One X performance. Of course, that's dependent on what are the highest frequencies Nintendo chooses for the CPU and GPU on Drake, as well as the RAM. And I don't think the probability of Nintendo using TSMC's N5 process node for the fabrication of Drake is particularly very high, although not close to zero, especially if Nintendo's still targeting a launch window of holiday 2022 to early 2023.

Assuming Nintendo plans to continue having the console attached to a VR viewer headset, similar to Nintendo Labo VR, I think Nintendo needs to customise a 720p OLED display to include support for VRR and a refresh rate higher than 60 Hz (e.g. 120 Hz), or use a 1080p OLED display with support for VRR and a refresh rate higher than 60 Hz already included. And I personally think Nintendo needs to figure how to achieve better ergonomics if Nintendo plans to continue having the console attached to a VR viewer headset, with the VR viewer headset having a head strap. When I added a head strap to Nintendo Labo VR, I found that that the weight was a bit too top heavy for me. (I understand that Nintendo intends for the consumer to attach the Joy-Cons to the console and hold the Joy-Cons when using Nintendo Labo VR.)

 

[Hosermess]

I'd have trouble squaring that with Nate's sources (presumably devs?) being told to plan for a late 2022 release (which holds/held true as of... last summer, I think?).
But a one or two quarter delay has always been possible.

Nate also added that while that period is the target for games to be completed by, it doesn't necessarily mean that they would be released around that time. I think he mentioned they could possibly need to be finalized or submitted for testing/verification on the new console or something. I can see with DLSS their could need to be more tinkering than normal.

The moment the next Switch is announced the current Switch sales will collapse just like the PS4.

Not if it's more expensive. I'm guessing it'll be premium priced for a while when it's released.

What do they have to lose with a 2022 reveal, be it summer or Q4? They don't have to announce the release date or price right away but can say Spring 2023.

if Botw 2 ends up being a launch game, then it's likely there will be two versions.

I think they would have to announce the release date and price when they officially reveal it. Any kind of unknown would just be adding confusing and chaos for consumers and investors alike. Not ideal.

I can't see them having two versions. For people who buy physical copies, I think they need to have them work the same way in any model. Plug and play. Maybe you'd need to download some stuff on the Drake, if anything. I know Nintendo wants households to have multiple Switches, but forcing them to buy multiple copies of the same game for them is a bit extreme.

 

[Trevelyan]

Not if it's more expensive. I'm guessing it'll be premium priced for a while when it's released.

PS4 fell off hard once PS5 was released, even being supply constrained. It really depends how Nintendo markets this. For the record, I don’t think switch sales will drop off that much, mainly because they’ll market it similar to the N3DS, where it’s just an upgraded model and not a real successor.

 

[BobNintendofan]

PS4 fell off hard once PS5 was released, even being supply constrained. It really depends how Nintendo markets this. For the record, I don’t think switch sales will drop off that much, mainly because they’ll market it similar to the N3DS, where it’s just an upgraded model and not a real successor.

The issue is, sales right now are being propped up by enthusiasts rebuying into the OLED and those OLED sales will be the ones to absolutely tank if a new console is even hinted at. While a casual player may not care if a more expensive iteration is releasing later, the enthusiast crowd is not going to be satisfied with the cheaper inferior option.

 

[fwd-bwd]

Were you expecting at latest March 2023?

Nintendo's FY23 is April 2022 to End of March 2023

So if you were expecting Holiday 2023, that would be FY24

Many Japanese corporations, such as Nintendo, describe their FY as “fiscal year ending March XXXX”, their current FY being “fiscal year ending March 2023”. When translating that to English, however, it is typically translated to “FY2022” or “FY22” as customary in the west. This is not unlike how we translate the Japanese name from “family-name given-name” to “given-name family-name” in English. Translating it literally would lead to confusion.

I've had a wild theory that the reason Zynga devs went on background with Bloomberg was because they were racing to hit a 2022 deadline and were starting to suspect that Nintendo might pull the rug out from them by letting the hardware slip. Zynga doesn't have a console background and having project schedules that they've got financials riding on (and possibly built around a tie in to a franchise which might be time limited) dictated by platform holders is likely new to them.

This probably isn’t just a wild theory. Multiple devs, which may or may not include Zynga, voiced the exact concern of a rug-pull in the Japanese version of that Bloomberg report:

“Several game company executives said that if the launch is withdrawn, it will undermine the trust in Nintendo.” (machine translated)

I don’t know why Mochizuki removed that from the English version of the same story. This wasn’t the only instance either. His English article of PS5 (Standard Edition) reaching break-even diverged from the Bloomberg Japan reporting, in which Sony stated that PS5 was “proceeding” toward break-even. Not sure if these were honest mistakes or due to behind-the-scene politics.

 

[JoshuaJSlone]

PS4 fell off hard once PS5 was released, even being supply constrained.

Wasn't a lot of this that Sony just plain stopped producing many PS4s? I'm not sure how far back this started.

Looking back at shipments, PS4 peaked in 2017 and was on a slow downward trend for the next few years, but later 2019 and on when PS5 was being seriously talked about did see the start of a faster decline.

 

[LiC]

Many Japanese corporations, such as Nintendo, describe their FY as “fiscal year ending March XXXX”, their current FY being “fiscal year ending March 2023”. When translating that to English, however, it is typically translated to “FY2022” or “FY22” as customary in the west. This is not unlike how we translate the Japanese name from “family-name given-name” to “given-name family-name” in English. Translating it literally would lead to confusion.

This probably isn’t just a wild theory. Multiple devs, which may or may not include Zynga, voiced the exact concern of a rug-pull in the Japanese version of that Bloomberg report:

“Several game company executives said that if the launch is withdrawn, it will undermine the trust in Nintendo.” (machine translated)

I don’t know why Mochizuki removed that from the English version of the same story. This wasn’t the only instance either. His English article of PS5 (Standard Edition) reaching break-even diverged from the Bloomberg Japan reporting, in which Sony stated that PS5 was “proceeding” toward break-even. Not sure if these were honest mistakes or due to behind-the-scene politics.

The "game company executives" were almost certainly from organizations who didn't have dev kits and were just being asked to weigh in to support (let's be honest here) the narrative of the article.

 

[mjayer]

I have done it
I HAVE FOUND A WAY TO CONVERT TFLOPS ACROSS uARCHS

---GPU FLOP Comparison Method--- Ampere: 2(SM Count * (128 * Clock speed)) Turing/Vega/GCN 2(SM/CU Count * (64 * Clock Speed)) RDNA1/IC-Less RDNA2 converted to GCN TFLOPs (2(CU Count * (64 * Clock Speed)))+25% Calculate the FP32/Cycle of a GPU with the equation then look at the TFLOP value of that GPU as rated by the manufacturer, the "Efficiency" of the TFLOPs will be exposed by the difference in the FP32/Cycle result Factor in % Additions to extrapolate back to a weaker/older uArch if the % difference is properly known (EX: RDNA1/IC-Less RDNA2 is 25% better in IPC than GCN, so it would be the GCN Equation + 25%)

And with this and some extrapolation Z0m3le and I have more or less determined Desktop Ampere to be equivalent to Polaris and therefore not so far off from the rest of GCN in regards to FLOP Efficiency

And that is not considering all the features that even Ampere has over GCN like Tile-Based Rasterization, Mixed Precision FP, The Tensor and RT cores, Primitive and Mesh Shaders, Variable Rate Shading (which can boost effective GPU perf up to 20% in some reports)

Which, considering Drake is running pretty much "Ampere+" with the extra L2 Cache, and RDNA2 reporting massive IPC uplifts with infinity Cache over the 25% from GCN to RDNA1, can likely give us determination on what clocks Drake needs to hit to match the PS4 Pro or even the Series S assuming different IPC uplifts over Ampere via us being able to covert IC-Less RDNA2 to GCN and Ampere and GCN being similar FLOP-to-FLOP

(Note, this is actually lowballing Ampere as modern games that take advantage of Ampere's features will outperform GCN, but this gives us a more level comparison between Ampere, GCN (PS4 Pro/One X), and IC-less RDNA2 (Series S). So technically these numbers are a lowball for modern effectiveness)


So for example

-PS4 Pro (GCN): 2(36*(64*0.911) = 4197.888 FP32/Cycle rated at 4.2 GCN TFLOPs -Series S (IC-Less RDNA2): (2(20(64*1.565)) = 4006.4 FP32/Cycle rated at 4 RDNA"2" TFLOPs + 25% = 5008 FP32/Cycle or 5 GCN TFLOPS ---Drake based on Default Ampere for Reference--- -Drake (Pure Ampere, OG Switch Docked Clocks): (2(12(128*0.768)) = 2359.296 FP32/Cycle or 2.35 Ampere TFLOPs -Drake (Pure Ampere, 1Ghz): (2(12(128*1)) = 3072 FP32/Cycle or 3 Ampere TFLOPs -Drake (Pure Ampere, 1.5Ghz): (2(12(128*1.5)) = 4608 FP32/Cycle or 4.3 Ampere TFLOPs -Drake (Pure Ampere, 1.63Ghz aka Matching Series S): (2(12(128*1.63)) = 5007.36 FP32/Cycle or 5 Ampere TFLOps. ---------------------------------------------------------- Now, assuming even just a marginal 10% increase in IPC over Ampere those values effectively become -Drake (768Mhz): 2.6 Ampere TFLOPs -Drake (1Ghz): 3.3 Ampere TFLOps -Drake (1.5Ghz): 5 Ampere TFLOps -Drake (1.63Ghz): 5.5 Ampere TFLOps

Even a 10% increase over Ampere due to that Cache is enough to make 1.5Ghz Drake match the Series S's GCN Equivalent!

Now, Assuming Drake gets the 25% boost AMD did just from going GCN to RDNA1

--Drake: 25% better than Ampere calculation-- -Drake (768Mhz): 2.95 Ampere TFLOPs -Drake (1Ghz): 3.8 Ampere TFLOps -Drake (1.5Ghz): 5.7 Ampere TFLOps -Drake (1.63Ghz): 6.2 Ampere TFLOps

You'd only need to hit 1.1Ghz to match the PS4 Pro if they pull out a 25% IPC increase through the cache (Which, considering reports of the 4070 with the exact same core count or less than the 3090 and the only major difference in raster perf seemingly being the Cache reporting an up to 30% increase over the 3090, that may very well be the case)

And only 1.3Ghz to match the Series S!

This is the content I come here for.

 

[Dakhil]

SK Hynix may head up consortium to buy Arm

Post Nvidia deal collapse, Intel expressed similar interest back in February

www.theregister.com

 

[fwd-bwd]

The issue is, sales right now are being propped up by enthusiasts rebuying into the OLED and those OLED sales will be the ones to absolutely tank if a new console is even hinted at.

This can very well happen, but may not be to the extent that you think.

• In the latest investor Q&A, Furukawa stated that the sales of evergreen titles (e.g., MK8D) may start to decline if there are less newSwitch buyers.
  • This makes sense, because buyers upgrading to OLED don’t have to repurchase their games.
  • Looking back, Furukawa gave us the raison d'être of MK8D Booster Pass, because Nintendo thinks that the MK8D sales may start to decline.
• However, MK8D remains among the best selling Switch titles. It stands to reason that there are still enough new Switch buyers vs. the OLED upgraders.
  • I mean at least for now. Not sure how this would shift over time, particularly when the new Switch model is finally revealed.
• Even if the would-be upgraders are to choose the new model over OLED, Nintendo still captures the sales—at a higher price point.
  • There will also be OLED owners upgrading again (triple dipping).
  • So even if the OLED sales figure drops, in totality it may be net plus for Nintendo.

 

[Mr Swine]

I honestly don’t see the next Switch needing to slip to March 2023 if it isn’t manufacturing related.

There are plenty of games coming out this year that Nintendo can advertise on making the games run better and so on.

If Zelda would get delayed again to summer I don’t think Nintendo would simply delay the Switch again

 

[Thraktor]

In that case, probably not, due to the increased likelihood of scaling artifacts. That isn't what I'm proposing though.

I'm proposing, for example, you have a 1080p portable device with an NVIDIA GPU (laptop, tablet, handheld, doesn't matter). You enable your device with the ability to use the equivalent to Deep Learning Dynamic Super Resolution. This will unlock the ability to use your native resolution as the base internal resolution for DLSS to upsample to up to 4 times its resolution (normally, you cannot do this if your device isn't capable of the equivalent to DSR or DLDSR). This is very different from rendering from a resolution lower than your native resolution and upsampling; effectively the method I'm suggesting uses AI-enhanced Ordered Grid Super Sampling Anti-Aliasing. It's very expensive when done traditionally but can be cost-effective when done intelligently.

The reason I'm asking about this is because rendering options on portable devices typically choose performance over quality. In this case, it will be more expensive than native rendering, but only by a little bit because of the overhead. I'm just wondering if you all will be interested in super high image quality on portable devices with a very slight hit to the performance as a DLSS alternative to pure performance improvements.

I'm basically asking if you all would be interested in a feature like this




on a portable device? Or would the slight hit to performance not be worth it?

Thanks for the clarification, in that case I can definitely see the benefit of having it. If we're talking about a PC scenario, then it's always useful to have options like that, even if only future hardware would really use them. On the console side, if you're not using the tensor cores for anything else, and you're able to squeeze it in while still hitting your target framerate, then all the better. Although I imagine in that case the memory bandwidth might be the limiting factor.

 

[Alovon11]

This is the content I come here for.

Hey, it's just math

But yeah, also in a similar vein, but looser, I did calculate the RTFLOP performance as we know the effective RTFLOPs of the Series X and NVIDIA states the RTFLOps of the Ampere cards.


And the math per RT core / per Ic-less Ray accelerator does work out to be that 12SM drake would be stronger than the PS5

Series X is 13RTFLOPs (They say over 25TFLOPs when ray tracing, so 12FP32 TFLOPs + 13RTFLOPs (at best) = 25TFLOPs total)
That works out to 13/52=0.25 RTFLOP/Ray Accelerator for IC-less RDNA2

0.25*20 = 5 RTFLOPs for Series S
0.25*36 = 9 RTFLOps for PS5

Ampere's RTFLOP calc is far easier as NVIDIA just gives us the pure-RT workload that Ampere cards can do in RTFLOPs (Aka path tracing)

EX:
RTX 3090: 69RTFLOPs/ 82RTCores = 0.84 RTFLOPs/A-RT Core

So Ampere is 0.84 RTFLOPs/RT core versus IC-less RDNA2 at 0.25 RTFLOPs/RT core.

So Drake at 12SMs would be 10RTFLOps.

So yes. it ray traces (in raw acceleration) better than the PS5 and that's before DLSS or any prediction of how performance can be improved if an RT method really likes Ampere's BVH Traversal.

 

[ReddDreadtheLead]

The 4700S multi core score strikes me as usually in the 8000's, with a decent amount of low 9000's. Peak of 9430 so far.
Its single core is typically 900's/1000's, with a peak of 1085.
Base frequency's 3.6 ghz.

Do note that, those scores are at the 4GHz rather than the 3.6GHz base that GB5 does it in.

 

[Foltzie]

I have done it
I HAVE FOUND A WAY TO CONVERT TFLOPS ACROSS uARCHS

You'd only need to hit 1.1Ghz to match the PS4 Pro if they pull out a 25% IPC increase through the cache (Which, considering reports of the 4070 with the exact same core count or less than the 3090 and the only major difference in raster perf seemingly being the Cache reporting an up to 30% increase over the 3090, that may very well be the case)

And only 1.3Ghz to match the Series S!

Is it feasible for the Drake Switch to run the GPU at much higher than 768MHz while docked?

 

[Alovon11]

Is it feasible for the Drake Switch to run the GPU at much higher than 768MHz while docked?

If it's 5nm of either or TSMC 6N, yes

Heck, it likely can hit 1Ghz at 8N at least

 

[Simba1]

NINTENDO Switch - Geekbench

Benchmark results for a NINTENDO Switch with an ARM ARMv8 processor.

browser.geekbench.com

I guess that switch was overclocked, because it says 1.78 GHz?
In that case results would be much worse because Switch runs at 1GHz.

 

[Kenka]

I guess that switch was overclocked, because it says 1.78 GHz?
In that case results would be much worse because Switch runs at 1GHz.

We would need power curves to give a definite answer with confidence

 

[Skittzo]

I guess that switch was overclocked, because it says 1.78 GHz?
In that case results would be much worse because Switch runs at 1GHz.

1.78GHz is actually the max frequency the Switch runs at, but only in short bursts during a handful of games to assist in loading.

They probably used the maximum available clock in the firmware.

 

[Zedark]

Hey, it's just math

But yeah, also in a similar vein, but looser, I did calculate the RTFLOP performance as we know the effective RTFLOPs of the Series X and NVIDIA states the RTFLOps of the Ampere cards.


And the math per RT core / per Ic-less Ray accelerator does work out to be that 12SM drake would be stronger than the PS5

Series X is 13RTFLOPs (They say over 25TFLOPs when ray tracing, so 12FP32 TFLOPs + 13RTFLOPs (at best) = 25TFLOPs total)
That works out to 13/52=0.25 RTFLOP/Ray Accelerator for IC-less RDNA2

0.25*20 = 5 RTFLOPs for Series S
0.25*36 = 9 RTFLOps for PS5

Ampere's RTFLOP calc is far easier as NVIDIA just gives us the pure-RT workload that Ampere cards can do in RTFLOPs (Aka path tracing)

EX:
RTX 3090: 69RTFLOPs/ 82RTCores = 0.84 RTFLOPs/A-RT Core

So Ampere is 0.84 RTFLOPs/RT core versus IC-less RDNA2 at 0.25 RTFLOPs/RT core.

So Drake at 12SMs would be 10RTFLOps.

So yes. it ray traces (in raw acceleration) better than the PS5 and that's before DLSS or any prediction of how performance can be improved if an RT method really likes Ampere's BVH Traversal.

Shouldn't we add an adjustment for clock speeds in here?

 

[Alovon11]

Shouldn't we add an adjustment for clock speeds in here?

Well problem is we don't really know how the RT cores or Ray Accelerators scale per-clock at this moment.

But even then Drake will easily be on the PS5-side of the performance spectrum versus Series S when regarding RT acceleration.

 

[EddyZacianLand]

It's not being released alongside BOTW2, bc when in the hell would they announce it???
They wouldn't want it to announce before the holiday season because that would mean a terrible holiday season

 

[Look over there]

Hmm, interesting.
Compare https://browser.geekbench.com/search?q=nintendo+switch to https://browser.geekbench.com/search?q=jetson+nano
The Jetson Nano should be TX1 rejects, so still A57s. But there is a discrepancy in score normalized to per ghz. Which leads to scrolling through some more of the Switch results... my suspicion is that it's down to the OS? The Linux scores seem to be better than Android on average.

 

[BobNintendofan]

I think Nintendo will market this as a sort of half revision/half new console to continue the switch hype but I don't really understand the notion that this will actually be a revision some people seem to have. As in most/all first party games will be cross gen, it wont have unique features, etc.

This thing appears to be at minimum 12x stronger than the Switch when you account for DLSS, RT, etc. How do you take a Switch 2 game and run it on a regular Switch when it can barely function with PS4 ports? Especially considering Nintendo is a stickler for quality and consistency. They're not releasing the next 3D Mario to the regular switch in 540p with drops to make it work, that's for sure. The only way to make cross gen releases work would be to first make them for the Switch and then bump resolution and framerate for the next console. And if they were going to do that, why even put so much power in this device in the first place? You wouldn't need RT cores, you wouldn't need 12 SM if this is just gonna be a 4k switch with third party support. They would be driving up the price for no reason. The purpose of all that increased power can only be one thing - to run exclusive first party games built from the ground up for the hardware.

If this is going to be a revision, I also heavily question the release timing. A revision 6 years into the life of a console makes no sense. If you look at Japanese sales charts where they release data of console sales by SKU, the OLED is already about half of the switch sales. What this suggests is that Nintendo is already approaching saturation of the market in terms of new customers. OLED sales are much more likely to be rebuys than other Switch units, and they are already starting to take over the majority of sales. I would not be shocked at all if switch sales were 40% rebuys right now. By 2023, when this thing launches I would expect rebuys to make up 60% or more of continued Switch sales. So if the substantial majority of your customers are existing customers looking for a new version, why not just launch a new console? If only 5-6 million new customers are buying in a year at that point, I don't see the need for continuing the switch at all.

 

[Heruderu]

I’ve been out of the loop. Last time I read about this, it was still called Switch Dane.

If it’s potentially using 5nm process, how much more powerful are people expecting it to be in relation to the previous expectations of Samsung’s 8nm? Also, doesn’t Switch V2/Lite/OLED have 4 SMs? Drake has 12??? That’s crazy! Was anyone expecting this before the Nvidia stuff leaked?

Thank you all for any answer. The thread moved very fast since Dane days and I wasn’t able to follow any of it

 

[mjayer]

I think Nintendo will market this as a sort of half revision/half new console to continue the switch hype but I don't really understand the notion that this will actually be a revision some people seem to have. As in most/all first party games will be cross gen, it wont have unique features, etc.

This thing appears to be at minimum 12x stronger than the Switch when you account for DLSS, RT, etc. How do you take a Switch 2 game and run it on a regular Switch when it can barely function with PS4 ports? Especially considering Nintendo is a stickler for quality and consistency. They're not releasing the next 3D Mario to the regular switch in 540p with drops to make it work, that's for sure. The only way to make cross gen releases work would be to first make them for the Switch and then bump resolution and framerate for the next console. And if they were going to do that, why even put so much power in this device in the first place? You wouldn't need RT cores, you wouldn't need 12 SM if this is just gonna be a 4k switch with third party support. They would be driving up the price for no reason. The purpose of all that increased power can only be one thing - to run exclusive first party games built from the ground up for the hardware.

If this is going to be a revision, I also heavily question the release timing. A revision 6 years into the life of a console makes no sense. If you look at Japanese sales charts where they release data of console sales by SKU, the OLED is already about half of the switch sales. What this suggests is that Nintendo is already approaching saturation of the market in terms of new customers. OLED sales are much more likely to be rebuys than other Switch units, and they are already starting to take over the majority of sales. I would not be shocked at all if switch sales were 40% rebuys right now. By 2023, when this thing launches I would expect rebuys to make up 60% or more of continued Switch sales. So if the substantial majority of your customers are existing customers looking for a new version, why not just launch a new console? If only 5-6 million new customers are buying in a year at that point, I don't see the need for continuing the switch at all.

So I think it might be somewhere in-between. When it first comes out, for a least a year, all first party software will be basically Switch 1 developed but running at 4k + whatever simple improvements they can fit in there on Drake. Then slowly they'll start putting out software that's Pro only native, but perhaps run on the cloud for Switch 1, or we could even see docked only?

So it's a new gen device with a sort of soft launch? They'll be getting lots of third party ports they are missing out on and also introduce a new pricing teir to their platform in the short term.

Dunno but it appears they're doing what we've wanted for so long and are actually putting out a powerful piece of tech that is competitive with the competition. At least until the PS5 pro and Series+ comes out.

 

[mjayer]

I’ve been out of the loop. Last time I read about this, it was still called Switch Dane.

If it’s potentially using 5nm process, how much more powerful are people expecting it to be in relation to the previous expectations of Samsung’s 8nm? Also, doesn’t Switch V2/Lite/OLED have 4 SMs? Drake has 12??? That’s crazy! Was anyone expecting this before the Nvidia stuff leaked?

Thank you all for any answer. The thread moved very fast since Dane days and I wasn’t able to follow any of it

Switch 1 was only 2 SM.

General expectation for Drake is something like:

GPU
8nm = between PS4 and PS4 Pro
5nm = around PS4 Pro maybe a bit above
and then keep in mind that's without factoring DLSS on top

CPU
I think this is more of a mystery until we know the cores but it should be something like 10x the current switch CPU so great boost irregardless of the node.

 

[ReddDreadtheLead]

Is it feasible for the Drake Switch to run the GPU at much higher than 768MHz while docked?

Or course, but that’s up to Nintendo if they feel comfortable with that.

It doesn’t have to be limited to 768MHz.

I guess that switch was overclocked, because it says 1.78 GHz?
In that case results would be much worse because Switch runs at 1GHz.

Those are overclocked units, yes. Geekbench runs at the highest available frequency that is allowed.

Switch would be 50-60% of those numbers.

Hmm, interesting.
Compare https://browser.geekbench.com/search?q=nintendo+switch to https://browser.geekbench.com/search?q=jetson+nano
The Jetson Nano should be TX1 rejects, so still A57s. But there is a discrepancy in score normalized to per ghz. Which leads to scrolling through some more of the Switch results... my suspicion is that it's down to the OS? The Linux scores seem to be better than Android on average.

Yeah it’s why I mentioned that Linux should be considered here, it doesn’t produce the same values as the android equivalent for some reason or rather I should say, it doesn’t produce equivalent enough numbers.

I’m not sure if the android one is more accurate or the Linux is more accurate.


I mean this was posted on the NVidia forum and it is clocked to 2.2GHz!

N/A - Geekbench

Benchmark results for a N/A with an ARM ARMv8 processor.

browser.geekbench.com

Jetson Nano

 

[ShaunSwitch]

I think Nintendo will market this as a sort of half revision/half new console to continue the switch hype but I don't really understand the notion that this will actually be a revision some people seem to have. As in most/all first party games will be cross gen, it wont have unique features, etc.

This thing appears to be at minimum 12x stronger than the Switch when you account for DLSS, RT, etc. How do you take a Switch 2 game and run it on a regular Switch when it can barely function with PS4 ports? Especially considering Nintendo is a stickler for quality and consistency. They're not releasing the next 3D Mario to the regular switch in 540p with drops to make it work, that's for sure. The only way to make cross gen releases work would be to first make them for the Switch and then bump resolution and framerate for the next console. And if they were going to do that, why even put so much power in this device in the first place? You wouldn't need RT cores, you wouldn't need 12 SM if this is just gonna be a 4k switch with third party support. They would be driving up the price for no reason. The purpose of all that increased power can only be one thing - to run exclusive first party games built from the ground up for the hardware.

If this is going to be a revision, I also heavily question the release timing. A revision 6 years into the life of a console makes no sense. If you look at Japanese sales charts where they release data of console sales by SKU, the OLED is already about half of the switch sales. What this suggests is that Nintendo is already approaching saturation of the market in terms of new customers. OLED sales are much more likely to be rebuys than other Switch units, and they are already starting to take over the majority of sales. I would not be shocked at all if switch sales were 40% rebuys right now. By 2023, when this thing launches I would expect rebuys to make up 60% or more of continued Switch sales. So if the substantial majority of your customers are existing customers looking for a new version, why not just launch a new console? If only 5-6 million new customers are buying in a year at that point, I don't see the need for continuing the switch at all.

I agree with you here.

The more I think about Drake, the less likely I feel that it is a pro style console.

Even if it's an iterative successor like many believe, it's still a successor, and whilst Nintendo may have an extended cross gen period of say two years, with this much of a power difference between the switch and Drake i can't see them supporting both devices longer than two years post Drake launch.

The biggest reason for me is the tensor cores and RTX cores. If Nintendo wants to make the most of this hardware, and they will, they will be developing with that hardware in mind, not the original switch.

Third parties I can see bringing Drake exclusive software day one. Theres a better chance of release date parity across PS5, series X and switch Drake compared to OG switch with the power gap being smaller.

 

[BobNintendofan]

So I think it might be somewhere in-between. When it first comes out, for a least a year, all first party software will be basically Switch 1 developed but running at 4k + whatever simple improvements they can fit in there on Drake.

I just cant see the strategic value for this approach for Nintendo. Maybe the regular switch would sell an extra 3-4 million units, which would net Nintendo probably around 300-400 million dollars in profit. However, they would be handicapping the launch of their new console in order to achieve that. I'm skeptical that even in the single financial year of its launch that they would benefit from this approach rather than just doing a typical move over.

 

[fwd-bwd]

If this is going to be a revision, I also heavily question the release timing. A revision 6 years into the life of a console makes no sense. If you look at Japanese sales charts where they release data of console sales by SKU, the OLED is already about half of the switch sales. What this suggests is that Nintendo is already approaching saturation of the market in terms of new customers. OLED sales are much more likely to be rebuys than other Switch units, and they are already starting to take over the majority of sales. I would not be shocked at all if switch sales were 40% rebuys right now. By 2023, when this thing launches I would expect rebuys to make up 60% or more of continued Switch sales. So if the substantial majority of your customers are existing customers looking for a new version, why not just launch a new console? If only 5-6 million new customers are buying in a year at that point, I don't see the need for continuing the switch at all.

There's no data substantiating the claim that OLED sales were "rebuys". Even if true that people are willingly rebuying in droves, it'd indicate that Nintendo should continue releasing revisions (instead of ceasing as you suggested) to give people what they want. This is the iterative update model that phone/tablet/PC consumers are familiar with, and it protects their software investments from becoming obsolete quickly. Looking at how much the PS5 games are still cross-gen, not to mention the Series X/S tiers, the console market may be moving in that direction too.

 

[BobNintendofan]

There's no data substantiating the claim that OLED sales were "rebuys". Even if true that people are willingly rebuying in droves, it'd indicate that Nintendo should continue releasing revisions (instead of ceasing as you suggested) to give people what they want. This is the iterative update model that phone/tablet/PC consumers are familiar with, and it protects their software investments from becoming obsolete quickly. Looking at how much the PS5 games are still cross-gen, not to mention the Series X/S tiers, the console market may be moving in that direction too.

I don't think people specifically want a revision, it's just all that's on the market. If there were a new console instead, I'm confident they would prefer that.
As for the comparison to PS5/XBX I don't think it's as clear cut as you believe it is.

The PS5 is only around 4-5x stronger than the PS4. Making PS5 games run on PS4 isn't a crazy ask. The console we are seeing looks more around 15x stronger than the Switch. Games would have to either be rebuilt from scratch, or built for the switch and then bumped up for the new console which would be an immense disappointment and a baffling waste of hardware power when they could have easily gone with a cheaper more budget option and still done 4k switch ports.

Also, I don't think it's fair to assume this extensive crossgen period was necessarily even the original plan of Sony and Microsoft. It is being driven in large part by the chip shortages. We have no idea how much support would have been cross gen normally.

 

[Golden Darkness]

So, given that it is based on what should be the same tech as the upcoming Switch, we have someone with an Orin Dev Kit:


However, at 15W power, it acts 4 core CPU machine running at max 1.1 GHz and the GPU is running at max 420 MHz.

 

[TurtleRock]

If history is anything to go by I think a few things are probably certain.

• This new device will likely have the word Switch in it. Because it's now the "family name" for this series of products.
• They could either market it as a successor (GB to GBA, GBA to DS, DS to 3DS) or an upgrade to an existing family (Gameboy Color, New 3DS).
• Given the assumed specs, I think we're more into successor territory but that's just my opinion.
• If old Switch sales drop off it likely won't happen for at least a year, so maybe 2024. The drop off of GBA and DS and of DS to 3DS wasn't too bad from what I recall.
• Console are a different beast since they have all usually died off long before their successor hit the market. I think Nintendo has seen the Switch as their flagship continuation of the portable consoles so they have no issue releasing a successor in a hot market.

I also think that it makes for sense to position it as a successor but with the benefit of backwards compatibility. From a marketing perspective you get way more buzz from people buying "the next thing" than if you would position it as a marginal upgrade for enthusiasts. Six years in I really don't see the benefit of treating hardware with this level of a jump as "marginal upgrade."

EDIT: Also at the end of the day I don't think it will matter too much. This is likely gonna sell no matter how they choose to market it.

 

[BobNintendofan]

I think it all boils down to this.
Nintendo is a business and they will do what makes them more money.
The purpose of a revision for a company is essentially to stall a generation out in order to fully saturate the market before moving on.
But when the market becomes saturated, there is no longer a reason to stall out a generation. When you've run out of new customers to buy your existing games, it's time to make new games and new hardware to repeat the process.

We are clearly approaching market saturation. The Switch will likely end this year around 125 million consoles. If they go as far as to wait until late 2023, they should end up close to 140 million. How many more potential new customers even exist?

 

[Alovon11]

So, given that it is based on what should be the same tech as the upcoming Switch, we have someone with an Orin Dev Kit:


However, at 15W power, it acts 4 core CPU machine running at max 1.1 GHz and the GPU is running at max 420 MHz.

Yeah, honestly that puts the nail in the coffin for 8nm Drake to me.

If they had to drop 2/3rds of the cores and drop the GPU down to 420MHz with an unknown number of GPU cores turned off running at 15W, how in the heck is 8nm Docked Drake supposed to run at least 4 Cores at 1Ghz with the 12SMs running at 786MHz to clock-match OG Switch for Backwards Compatibility.

Even saying the DLAs and the PVAs suck back power, that is still dropping 66% of the CPU Cores, and the remaining running at 1Ghz, and at absolute best having the 12SMs of Drake hit 786mhz

And even then it likely would need to suck back more than 15W for the SoC alone to hit it.

And don't even bring up "They can turn off SMs to make it work", there is still no concrete/reasonable indication that Drake will do that based on the information we have at the moment, so Occam's razor does dictate that they likely have to make all 12SMs hit 768Mhz for B/C when docked.

 

[ReddDreadtheLead]

So, given that it is based on what should be the same tech as the upcoming Switch, we have someone with an Orin Dev Kit:


However, at 15W power, it acts 4 core CPU machine running at max 1.1 GHz and the GPU is running at max 420 MHz.

It’s a bit tricky with this, as we don’t know all of the details for that TDP.

 

[Deleted member 887]

The purpose of a revision for a company is essentially to stall a generation out in order to fully saturate the market before moving on.
But when the market becomes saturated, there is no longer a reason to stall out a generation. When you've run out of new customers to buy your existing games, it's time to make new games and new hardware to repeat the process.

We are clearly approaching market saturation. The Switch will likely end this year around 125 million consoles. If they go as far as to wait until late 2023, they should end up close to 140 million. How many more potential new customers even exist?

You're totally right, but I think the dynamic of clear and successive generations has started to break down. Nintendo has long been a company that would see a market dynamic and be the first to try to bypass it rather than "win" on the existing terms. We look at the install base of The Switch vs PS5 but no one looks at the install base of the iPhone 13, just the install base of iPhones.

Nintendo clearly wants to move to One Nintendo Account with One Store, they've long done rolling backwards compatibility in their mobile lines with lots of refreshes. Sony and Microsoft are hitting the limits of the Power Wars. Nintendo has a form factor that works, they don't want to change it up, and they don't have a secondary console line to support the "new" generation in case it fails. This sliding revisions strategy is a risk, but it fits where they are, and where the market is going.

 

[Kenka]

I think it all boils down to this.
Nintendo is a business and they will do what makes them more money.
The purpose of a revision for a company is essentially to stall a generation out in order to fully saturate the market before moving on.
But when the market becomes saturated, there is no longer a reason to stall out a generation. When you've run out of new customers to buy your existing games, it's time to make new games and new hardware to repeat the process.

We are clearly approaching market saturation. The Switch will likely end this year around 125 million consoles. If they go as far as to wait until late 2023, they should end up close to 140 million. How many more potential new customers even exist?

There is no evidence that there is saturation. The Switch has no competition in its market and I don't think that the audience for a 300 USD portable machine with interchangeable software is only a crowd of 150 million people. Switch can go for 200 million.

Depending on its strategy in China and other BRIC members, it could go higher. Maybe not much higher but still.

 

[BobNintendofan]

There is no evidence that there is saturation. The Switch has no competition in its market and I don't think that the audience for a 300 USD portable machine with interchangeable software is only a crowd of 150 million people. Switch can go for 200 million.

Depending on its strategy in China and other BRIC members, it could go higher. Maybe not much higher but still.

I just posted the evidence. In Japan they give breakdowns of sales by switch model and the OLED is making up about half the sales. The OLED is much more likely to be a rebuy than the other two units are, so it's safe to say rebuys are up a lot. Of course we cant say how high until Nintendo tells us (if they tell us), but it's up substantially nonetheless. And it's just common sense. There's only so many people willing to buy a video game console, particularly for $300 and without substantial third party support. They could have the console on the market for 50 years and they're never making 200 million sales.

 

[Dakhil]

So, given that it is based on what should be the same tech as the upcoming Switch, we have someone with an Orin Dev Kit:


However, at 15W power, it acts 4 core CPU machine running at max 1.1 GHz and the GPU is running at max 420 MHz.

I think there needs to be benchmarks for the Orin NX devkits before any educated, speculative conclusions can be made.

 

[ReddDreadtheLead]

Sony and Microsoft are hitting the limits of the Power Wars.

Maybe.

I think it depends.

I think if things don’t work out, they’ll try to stretch the generation even longer and longer until it works out where they feel it’s ok enough for them to move on.

But I think we may be approaching the realm of diminishing returns. For the high end anyway.

TDP in the next few years will be wild. PS3 had a high one, but nothing has match that since.