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

[Angel Whispers]

And after T239 details from the illegal Nvidia leaks were discussed, kopite7kimi admitted to being wrong about T239 being Ada Lovelace based.


(kopite7kimi wasn't completely wrong since Drake's known to inherit at least one feature from Ada Lovelace (same with Orin).)

We're largely on the same page here. I'm just saying that if this is coming in 2023/2024, more Lovelace elements are possible, and pointed to what they might be. Also, I mentioned whispers from 2021. If this had released last year or around now, I wouldn't have brought up Lovelace, but with the new window, it's worth considering what else could be on the table, all the more so, given that we don't know if any tape-outs. Personally, I feel it will be the same lithographic process.

 

[TheSentry]

I have been slowly coming around to the view that if GPU performance specs are talked about on this thing, it will be presented as “Up To 3 or 4TF”. I don’t think they’ll be as shy about it primarily because the Switch’s success has restored much of the confidence lost during the Wii U/3DS era. That, and Nvidia will surely speak on it, too. The inclusion of DLSS also tells me that tech disruption will be a theme. The innovation in SoC design has alluded to this for some years, and I feel it will lead to some pleasant surprises.

Has Nintendo talked about specs and performance any time recently?

I haven’t heard them talk about chips or performance related stuff since the GCN days.

I don’t think Nvidia shared much about the Switch’s specs or technical details either once it was released (probably at the request of Nintendo).

I know some times they share how much RAM and what chip family is inside, but I don’t think they typically release many specific numbers.

So, other than tear downs and hacks once the new console gets released, I really doubt we’ll ever get much in the way of specifics like hertz and TFs.

Maybe I am wrong. Have I missed something in regards to Nintendo sharing this type of info?

 

[ILikeFeet]

If GPU at end really has 1536 Ampere cuda cores, 48 Tensor cores and 12 RT cores,
we dont know what will be customized and what exactly product (chip) will be at end in Switch 2.

this is the chip that will be used without further customization. it was made for Nintendo in the first place

 

[Angel Whispers]

If GPU at end really has 1536 Ampere cuda cores, 48 Tensor cores and 12 RT cores,
we dont know what will be customized and what exactly product (chip) will be at end in Switch 2.

1536 CUDA Core GPU (12SM) IS on the customised SoC. That was a big take from the Nvidia leaks. It's derived from the 16SM (2048 CUDA Core) Orin one. Unless this was the cancelled SoC, It'll be in the retail product.

 

[Angel Whispers]

Has Nintendo talked about specs and performance any time recently?

I haven’t heard them talk about chips or performance related stuff since the GCN days.

I don’t think Nvidia shared much about the Switch’s specs or technical details either once it was released (probably at the request of Nintendo).

I know some times they share how much RAM and what chip family is inside, but I don’t think they typically release many specific numbers.

So, other than tear downs and hacks once the new console gets released, I really doubt we’ll ever get much in the way of specifics like hertz and TFs.

Maybe I am wrong. Have I missed something in regards to Nintendo sharing this type of info?

Not lately, but that was under their predecessors, and before we knew the Switch's fate. I believe that, with confidence, they might begin to speak a little more on it, and we have seen in other examples that the words "Up To" can do a lot of heavy lifting. At least, that was my thought process.

 

[Hermii]

I was basing that assumption on a combination of the skepticism in this thread that Nintendo would go lower than the current Switch's clocks, combined with some chat here regarding Ampere's power curve, which becomes highly power-inefficient below a certain speed. I'm open to being corrected if I've misinterpreted anything.

edit: I should clarify that I’m mostly speaking to the lower end of the scale that @Simba1 had outlined.

Let's do some simple math. Assuming Switch clocks, we can infere an exact 6x multi right?

157x6 = 942, just under a tflop in portable.

393x6 = 2358. Just over 2,3 tflops docked.

Is my math way off?

 

[Thraktor]

Has Nintendo talked about specs and performance any time recently?

I haven’t heard them talk about chips or performance related stuff since the GCN days.

I don’t think Nvidia shared much about the Switch’s specs or technical details either once it was released (probably at the request of Nintendo).

I know some times they share how much RAM and what chip family is inside, but I don’t think they typically release many specific numbers.

So, other than tear downs and hacks once the new console gets released, I really doubt we’ll ever get much in the way of specifics like hertz and TFs.

Maybe I am wrong. Have I missed something in regards to Nintendo sharing this type of info?

They'll release specs if it's beneficial to them to do so (ie if it makes them look good in comparison to the competition). With the Wii U, they were happy to talk about the fact that it had 2GB of RAM, which was 4x as much as the PS3 or XB360, but weren't disclosing any specs that would have put them in a less favourable light, like GPU Gflops.

With [REDACTED] almost certainly being a Switch form-factor device, and being on the market alongside power-hungry home consoles like the PS5 and XBSX, it's not likely that the new device will have any specs which outstrip the competition, so the chances of detailed specs being revealed seems very low. They may still release specifications on areas which could be considered on-par with the competition, such as stating that it's got an 8 core CPU, without divulging the cores or the clock speed. In the event that it has 16GB of RAM (which I don't think is likely), I could see them revealing that, although without specifying bandwidth or anything like that.

Actually, one area where they could have an advantage over Sony and MS's hardware is manufacturing process. If T239 does end up being manufactured on TSMC 4N, then I could see that being referenced in a press release or spec sheet.

 

[kvetcha]

Let's do some simple math. Assuming Switch clocks, we can infere an exact 6x multi right?

157x6 = 942, just under a tflop in portable.

393x6 = 2358. Just over 2,3 tflops docked.

Is my math way off?

iirc @oldpuck had estimated an absolute minimum of 1tflop portable. I was primarily raising an eyebrow at the expectation of 1.5tflops docked.

 

[Instro]

I actually wouldn't be surprised if they announce it to investors either immediately before or during the fiscal year earnings release. They have to make projections for the next FY, and absent of any new hardware, they're going to be projecting large drops in pretty much every metric. That's a tough sell to investors if your only commentary is that you're in "uncharted territory". I could see them releasing a press release along the lines of the original 3DS announcement, simply confirming they're going to release a new hardware platform this fiscal year, and it will be backwards compatible with Switch software, possibly not even including the name of the new device. Enough to head off concerns about declining sales of the original Switch while leaving the actual reveal for a later date.

I honestly don't think there's a big deal with announcing it before Zelda releases. Last of Us Part II released literally a week after Sony's not-E3 PS5 showcase in June 2020 and still broke sales records. I don't think they'll show, or talk about, higher res version of ToTK running on the new console until after the game launches, but simply announcing that a new console exists shouldn't meaningfully impact software sales.

Yeah if the expectation is that it's launching in the upcoming FY, then the device should be mentioned in some form during the investor meeting and in their financial data. If we look back at some of their recent hardware launches, that's been their practice, typically with them noting that they are not setting out separate forecast numbers for the new device.

I would actually be skeptical of it launching by the end of March 2024 if they don't mention it during the FY briefing.

 

[ILikeFeet]

Tower of Fantasy already has RT on PCs, and now it's getting it on mobile with what looks like RT refractions. shadows, and AO

OnePlus Community

Introducing our new OnePlus Community experience, with a completely revamped structure, built from the ground-up.

community.oneplus.com

 

[Dakhil]

If this had released last year or around now, I wouldn't have brought up Lovelace, but with the new window, it's worth considering what else could be on the table, all the more so, given that we don't know if any tape-outs. Personally, I feel it will be the same lithographic process.

Although there's no explicit proof, there was a LinkedIn profile of a Nvidia employee that mentioned electrically characterising and validating I/O interfaces (IC2, SPI/QSPI) for Ada Lovelace GPUs and T239 from April 2022 to present, which implies that Nvidia has been working with actual T239 silicon, I think there's a likely chance T239 was taped out at around early 2022.

And I don't think Samsung's 5LPE process node or TSMC's N6 process node can be dismissed as possibilities.
One of the files in the illegal Nvidia leaks defined Samsung as the semiconductor foundry company being used for T239 (here, here, here, and here).
And during GTC 2021 (November 2021), Nvidia did mention datacentre products (BlueField-3, Quantum-2, ConnectX-7) that were fabricated using TSMC's 7N process node and were being sampled roughly around the same time Ada Lovelace GPUs and probably T239 were.

 

[Deleted member 887]

Are you suggesting the Switch [REDACTED] will eat a Series S for breakfast, or that it will be basically a Series S?

I think that oldpuck's referring to how if you take something the size of the One X but update its stuff, it's actually rather a beast.

For example, keep the 384-bit memory bus width, but update from GDDR5 to GDDR6. Keep the 40 CU's, but update from GCN to RDNA2, bump the clock up thanks to node + RDNA2 changes (to compare, the Series S is 20 CU's and Series X is 52 CU's). Update the CPU.
...actually, the 384-bit bus would be rather ludicrous with GDDR6; that's wider than the Series X (320-bit), for example. (Series S is 160-bit wide)

Yeah, thats what I meant. Microsoft overshot with the One X, and it's design is creaky (GCN, Jaguar, HDD), but it's performance is not - it's 1.5x the raw GPU performance of the Series S. Update its architecture, and you've got a machine that could keep pace with the PS5.

To answer @Concernt - it's going to underperform the Series S. It's not a GPU question, it's everything else. Those ARM CPUs are remarkably efficient, but there is no way you can get up to matching the sustained performance of Zen 2, with 16 threads, clocked out to nearly 4GHz. And while RDNA isn't as memory efficient as Ampere, Series S still has 2x the memory bandwidth of Drake.

Even if Nintendo decides to match Series S FLOP for FLOP, it will underperform Series S on many games because of these bottlenecks. And if you push docked mode that far, it starts affecting handheld mode too. If the gap between the two is too large, either docked mode has wasted power (because devs target handheld) or handheld has severely sacrificed games (because devs target docked)

Try to close that gap on the handheld side, you kill battery life and run hot. Try to fix that with a big battery and fan, and you've got SteamDeck with it's huge formfactor and screaming fan.

I think the smarter move is to keep the GPU better balanced against the rest of the machine, and keep handheld's performance gap about where the Switch already is. If I had to guess, I'd say that 3TF is about the peak there.

All that said: Series S doesn't have any exclusives! There are zero games on the market built from the ground up for its level of power. Microsoft is under increasing pressure to drop the rules which require that Series S only differ from Series X in visuals. Meanwhile Nintendo is going to be crafting exclusives for that 2.5-3TF box which are built for DLSS and Nvidia's superior RT solution. [REDACTED] is set to age better when compared to the Series S across the generation. I don't think that's in question

 

[Serif]

Is the difference between 8nm, 5nm, etc. meaningful in regards to what games can be ported to the system, or is it more of a question of 'max potential'?

 

[Deleted member 887]

Let's do some simple math. Assuming Switch clocks, we can infere an exact 6x multi right?

157x6 = 942, just under a tflop in portable.

393x6 = 2358. Just over 2,3 tflops docked.

Is my math way off?

iirc @oldpuck had estimated an absolute minimum of 1tflop portable. I was primarily raising an eyebrow at the expectation of 1.5tflops docked.

The math is pretty much on. The only thing I'd correct is that Nintendo originally set out for a 300MHz clock - ~150 GFLOPS - in handheld mode, but actually pushed it up to 420MHz right before launch, and most games use that profile instead. So portable comes out to about 1.2TF if you use that clock instead.

The 1TF minimum is a side effect of Ampere's power curve. Below 300-350MHz, Ampere stops reducing power draw or heat generated. The exact line is kinda fuzzy.

 

[Concernt]

Yeah, thats what I meant. Microsoft overshot with the One X, and it's design is creaky (GCN, Jaguar, HDD), but it's performance is not - it's 1.5x the raw GPU performance of the Series S. Update its architecture, and you've got a machine that could keep pace with the PS5.

To answer @Concernt - it's going to underperform the Series S. It's not a GPU question, it's everything else. Those ARM CPUs are remarkably efficient, but there is no way you can get up to matching the sustained performance of Zen 2, with 16 threads, clocked out to nearly 4GHz. And while RDNA isn't as memory efficient as Ampere, Series S still has 2x the memory bandwidth of Drake.

Even if Nintendo decides to match Series S FLOP for FLOP, it will underperform Series S on many games because of these bottlenecks. And if you push docked mode that far, it starts affecting handheld mode too. If the gap between the two is too large, either docked mode has wasted power (because devs target handheld) or handheld has severely sacrificed games (because devs target docked)

Try to close that gap on the handheld side, you kill battery life and run hot. Try to fix that with a big battery and fan, and you've got SteamDeck with it's huge formfactor and screaming fan.

I think the smarter move is to keep the GPU better balanced against the rest of the machine, and keep handheld's performance gap about where the Switch already is. If I had to guess, I'd say that 3TF is about the peak there.

All that said: Series S doesn't have any exclusives! There are zero games on the market built from the ground up for its level of power. Microsoft is under increasing pressure to drop the rules which require that Series S only differ from Series X in visuals. Meanwhile Nintendo is going to be crafting exclusives for that 2.5-3TF box which are built for DLSS and Nvidia's superior RT solution. [REDACTED] is set to age better when compared to the Series S across the generation. I don't think that's in question

I would say I'm under no illusions that it can outperform Series S in raw computational power, but I would say it could probably achieve better IQ at the same fidelity. Not quite eating for breakfast. But maybe a little nibble. As a treat.

Also while Series S has no "exclusives" to call its own, we can't pretend like Forza Horizon 5 doesn't look incredible on it. Even with Drake, Nintendo first parties will struggle to reach that sort of fidelity. However, I do think first parties will, with few exceptions, have much better IQ than Series S' rendition of Forza.

 

[Skittzo]

Is the difference between 8nm, 5nm, etc. meaningful in regards to what games can be ported to the system, or is it more of a question of 'max potential'?

The node size determines how high everything can be clocked while still using the target level of power. So max potential basically.

Basically it amounts to watts per clock, lower node equals lower watt per clock. Or, perhaps put in a more useful way, higher clock per watt.

 

[Deleted member 887]

I would say I'm under no illusions that it can outperform Series S in raw computational power, but I would say it could probably achieve better IQ at the same fidelity. Not quite eating for breakfast. But maybe a little nibble. As a treat.

Also while Series S has no "exclusives" to call its own, we can't pretend like Forza Horizon 5 doesn't look incredible on it. Even with Drake, Nintendo first parties will struggle to reach that sort of fidelity. However, I do think first parties will, with few exceptions, have much better IQ than Series S' rendition of Forza.

Oh I think Series S games look great! Series S is a neat machine, and game dev is more scalable than ever. I'm just saying that if REDACTED gets within spitting distance of Series S, then Nintendo's ability to take full advantage of their unique features - DLSS, RT - probably matters more than the extra performance that Series S has.

 

[Armoured_Bear]

Not lately, but that was under their predecessors, and before we knew the Switch's fate. I believe that, with confidence, they might begin to speak a little more on it, and we have seen in other examples that the words "Up To" can do a lot of heavy lifting. At least, that was my thought process.

22 years is "Lately"?

For Nintendo to start showing off about specs, they'd have to reverse their tech culture of the last 20 years, I don't expect that to significantly change, even if Switch 2 is anywhere near as powerful as some people believe (and no, they don't know, they believe).

 

[kvetcha]

The math is pretty much on. The only thing I'd correct is that Nintendo originally set out for a 300MHz clock - ~150 GFLOPS - in handheld mode, but actually pushed it up to 420MHz right before launch, and most games use that profile instead. So portable comes out to about 1.2TF if you use that clock instead.

The 1TF minimum is a side effect of Ampere's power curve. Below 300-350MHz, Ampere stops reducing power draw or heat generated. The exact line is kinda fuzzy.

Thanks, @oldpuck. I had a vague recollection of these numbers from a few months back, but didn’t have it in me to comb back through the thread to find them. I appreciate you.

 

[Goodtwin]

What are these minimum clocks in your opinion?

Its hard to say because we don't know what node Drake is being manufactured on. If its 8nm like Orin we could be looking at 300Mhz portable and 700Mhz docked. This is assuming Nintendo needs is keeping the powerdraw in line with what the original Switch consumed. If its on a better process, and there are numerous options for a more power efficient process than 8nm, then we could be looking at clock speeds significantly higher. Which brings us to.....

Actually the GPU numbers I gave in my post were found in NVN from the hack, it's a curious DLSS test, where there are 3 clocks given that are named by power consumption. 4.2w is 660MHz, 9.3w is 1.125GHz and a 3rd which I believe is a stress test, called 12w, and is 1.38GHz.

Its hard to believe these numbers are accurate for Drake on 8nm. Is it 4N? TSMC 7nm? We dont know, but when you look at the mobile based RTX3050M graphics card with its TDP of 45watts at 1.3Ghz, its hard to see how they could come up with enough energy savings to get to 12w at the same clock speed on 8nm. Its pretty clear to me that it has to be a better process than 8nm and is likely 4N, seeing as how all RTX40 cards are being manufactured on that process this year.

Is the difference between 8nm, 5nm, etc. meaningful in regards to what games can be ported to the system, or is it more of a question of 'max potential'?

We are talking about nearly double the clocks so double the performance. So yea, Drake on 5nm is significantly more powerful than Drake on 8nm where it would be clocked very low. 8nm Drake would be closer to PS4 than Xbox Series S, although it would still have the added perk of DLSS.

 

[Skittzo]

We are talking about nearly double the clocks so double the performance. So yea, Drake on 5nm is significantly more powerful than Drake on 8nm where it would be clocked very low. 8nm Drake would be closer to PS4 than Xbox Series S, although it would still have the added perk of DLSS.

No, we're talking nearly double the performance per watt, not necessarily clocks. We don't know how they'll use that extra efficiency, maybe they'll try for a better battery life than the OG Switch.

 

[RennanNT]

Nintendo originally set out for a 300MHz clock - ~150 GFLOPS - in handheld mode, but actually pushed it up to 420MHz right before launch, and most games use that profile instead.

Actually, Switch uses 3 different clocks: 307.2 MHz, 384 MHz, and 460 MHz. So. I'd say the original plan was 50% of docked mode, but they added the minimum clocks to maximize battery on non-demanding games and near launch BotW performance issues forced them to add the last one. You're probably mixing the last one with the minimum clock Orin supports.

 

[Goodtwin]

No, we're talking nearly double the performance per watt, not necessarily clocks. We don't know how they'll use that extra efficiency, maybe they'll try for a better battery life than the OG Switch.

At 8nm Drake would have to be clocked extremely low to fit within the thermal/power consumption of the Switch form factor, which Nintendo sees as 7-9watts total power draw for portable play and 12-15watts for docked. Im expecting very similar power draw for the next model, and Drake on 4N compared to 8nm deliveries significantly more performance. 4N gives Nintendo some flexibility with balancing battery life and performance, 8nm guarantees super low clock speeds just to get Erista levels of battery life (3 hours). On 4N, Nintendo could target 4 hours of battery life with much higher performance. It doesnt look like Drake on 8nm could be clocked low enough to get Mariko levels of power efficiency.

 

[Deleted member 887]

you're probably mixing the last one with the minimum clock Orin supports.

You're exactly right! Not the first time I've done that

 

[Hermii]

Actually, Switch uses 3 different clocks: 307.2 MHz, 384 MHz, and 460 MHz. So. I'd say the original plan was 50% of docked mode, but they added the minimum clocks to maximize battery on non-demanding games and near launch BotW performance issues forced them to add the last one. You're probably mixing the last one with the minimum clock Orin supports.

Actually I would say 307 MHz was probably the original plan. That leaves a 2,5 ratio between portable and docked, to match the 2,25 pixel ratio between 720p and 1080p, with some to spare.

 

[Bosintang]

I have a hard time seeing Nintendo going for such an excellent chip and being cheap on the node. I don't think they'll fumble that next console. They will cheap out on some stuff like storage or even construction probably though.

 

[Angel Whispers]

Not sure who needs to hear it, but Ubisoft, Activision, T2, EA, etc. haven’t expressed dissent about the XSS, and all would’ve been consulted before it was greenlit. It isn’t going anywhere, it isn’t “holding anything back” for “third parties”, and Microsoft isn’t about to start mandating a policy which would allow their partners to skip it, or turn it into an “impossible port” machine. This idea needn’t be entertained, as any developer crying about it is probably a bad carpenter blaming tools. They’ll get over it soon enough, just like they did when making games for the PS3 and its Cell architecture. They’ll have to, because if they’re already not supporting Nintendo platforms in the same way, one suspects that they can’t afford to have two platforms treated as such.

 

[Intoxicate]

I‘m still happy with my OLED Switch which fixed the flaws of the OG for me (Bezelsize, kickstand). I hope Nintendo takes their time to release a real upgrade and not just an new 3DS kinda version. Since we didn’t get anything now, I hope they‘re aiming for 3nm or lower I can skip the „OG Switch 2“ with bad battery life.

 

[pelusilla6]

We got the Steam Deck and the Series S plus a lot of low budget PCs, we shouldnt be worried about good ports i guess!

 

[Shish]

Yeah, thats what I meant. Microsoft overshot with the One X, and it's design is creaky (GCN, Jaguar, HDD), but it's performance is not - it's 1.5x the raw GPU performance of the Series S. Update its architecture, and you've got a machine that could keep pace with the PS5.

To answer @Concernt - it's going to underperform the Series S. It's not a GPU question, it's everything else. Those ARM CPUs are remarkably efficient, but there is no way you can get up to matching the sustained performance of Zen 2, with 16 threads, clocked out to nearly 4GHz. And while RDNA isn't as memory efficient as Ampere, Series S still has 2x the memory bandwidth of Drake.

Even if Nintendo decides to match Series S FLOP for FLOP, it will underperform Series S on many games because of these bottlenecks. And if you push docked mode that far, it starts affecting handheld mode too. If the gap between the two is too large, either docked mode has wasted power (because devs target handheld) or handheld has severely sacrificed games (because devs target docked)

Try to close that gap on the handheld side, you kill battery life and run hot. Try to fix that with a big battery and fan, and you've got SteamDeck with it's huge formfactor and screaming fan.

I think the smarter move is to keep the GPU better balanced against the rest of the machine, and keep handheld's performance gap about where the Switch already is. If I had to guess, I'd say that 3TF is about the peak there.

All that said: Series S doesn't have any exclusives! There are zero games on the market built from the ground up for its level of power. Microsoft is under increasing pressure to drop the rules which require that Series S only differ from Series X in visuals. Meanwhile Nintendo is going to be crafting exclusives for that 2.5-3TF box which are built for DLSS and Nvidia's superior RT solution. [REDACTED] is set to age better when compared to the Series S across the generation. I don't think that's in question

it seems interesting, at least i think, the GPU of Nintendo redacted Is very powerful for a portable console.

No Questions Charlie.

 

[Goblin_knight]

Speaking of ports I wonder if Nintendo is willing to touch up Wind wakerHD/Twilight princessHD? Maybe some RT enhancements or just 4k with DLSS?

 

[Concernt]

Speaking of ports I wonder if Nintendo is willing to touch up Wind wakerHD/Twilight princessHD? Maybe some RT enhancements or just 4k with DLSS?

Those games absolutely would not need DLSS to hit 4K on Drake. I expect no enhancements other than raw 4K.

 

[LukasManak22]

Speaking of ports I wonder if Nintendo is willing to touch up Wind wakerHD/Twilight princessHD? Maybe some RT enhancements or just 4k with DLSS?

At this point TP HD should have the same upgrade that WW HD had graphics wise.

 

[Instro]

Twilight Princess getting a big upgrade to look like that WiiU tech demo would be the real dream.

 

[Serif]

At this point TP HD should have the same upgrade that WW HD had graphics wise.

I remember them showing their lighting experiments.

 

[LukasManak22]

I remember them showing their lighting experiments.

Takizawa showing the real meat

 

[D36]

even if Switch 2 is anywhere near as powerful as some people believe (and no, they don't know, they believe).

I mean, that belief is substantiated with very well educated guesses. While we can't know for sure how powerful the succ will be until we get our hands on it, we've mostly gotten a good grasp of those specs.

 

[Hermii]

I mean, that belief is substantiated with very well educated guesses. While we can't know for sure how powerful the succ will be until we get our hands on it, we've mostly gotten a good grasp of those specs.

A lot of well educated guesses in 2016, assumed 16nm and much higher clocks. Nintendo could still disappoint us.

 

[D36]

A lot of well educated guesses in 2016, assumed 16nm and much higher clocks. Nintendo could still disappoint us.

True, but I don't think they can drop the ball that hard. I strongly believe that this device will be sufficient regardless of who's right or wrong. IMHO, "Because Nintendo" is as weak of a counterargument as it's ever been.

 

[Serif]

Didn't DF straight up drop the NX clocks before the reveal

 

[D36]

While we're discussing the past, what exactly were people expecting the NX to be that was in the realm of possibility at the time, especially with what was shown before it was called the Switch?

 

[Skittzo]

At 8nm Drake would have to be clocked extremely low to fit within the thermal/power consumption of the Switch form factor, which Nintendo sees as 7-9watts total power draw for portable play and 12-15watts for docked. Im expecting very similar power draw for the next model, and Drake on 4N compared to 8nm deliveries significantly more performance. 4N gives Nintendo some flexibility with balancing battery life and performance, 8nm guarantees super low clock speeds just to get Erista levels of battery life (3 hours). On 4N, Nintendo could target 4 hours of battery life with much higher performance. It doesnt look like Drake on 8nm could be clocked low enough to get Mariko levels of power efficiency.

I'm trying to say that 5nm doesn't necessarily mean double the clocks, it will depend on how they want to utilize the increased efficiency over 8nm.

 

[Goodtwin]

A lot of well educated guesses in 2016, assumed 16nm and much higher clocks. Nintendo could still disappoint us.

True, but I think the results have proven to far exceed what people were expecting from an underclocked X1. The X1 also had the rotten luck of being designed on a DOA node. If the X1 had been 16nm from the get go, Nintendo could have raised the clock speeds higher and still gotten better battery life than Erista Switch units. Right now we have a pretty large performance window for Drake. Basically it could be close to PS4 or PS4 Pro, its somewhere in between.

I'm trying to say that 5nm doesn't necessarily mean double the clocks, it will depend on how they want to utilize the increased efficiency over 8nm.

I understand that, I'm just pointing out that 4N allows them to nearly double the clocks at the same power draw if they want to. If its 4N I expect better than 3 hours of battery and if its 8nm, 3 hours is probably best case scenario.

 

[Instro]

Didn't DF straight up drop the NX clocks before the reveal

Yes, they published full hardware info for a retail unit back in November/December 2016 before the January event. Unless you're talking about the October event, in which case no, all we had prior to that was what chip it was using. That said I believe the correct clock speeds were leaked on Twitter or Reddit prior to the October event, not sure when exactly though.

 

[Gerald]

Reminder that the onus is on individuals to get a grip of their own emotions, rather than demand everybody else be negative under the facade of "keeping their expectations in check". If it doesn't happen, it doesn't happen. Nothing discussed here has been outside the realm of possibility, and he didn't pull that info from a magic hat. Even a 2GHz CPU would be clocked at around 60% of its capacity (3.3GHz), so, it's already clocked lower than its potential. But just remember that whatever you post here, it's going to have RT, DLSS, and it will need to be able to perform them competently, so, wherever you believe the final specs will land would do well to be reconcilable with that. A 1.5TF docked performance would doom it to failure, while needing to clock closer to the Switch Lite to get there, being a waste of the Earth's resources, and entirely unfit for a generational purpose in 2024 and beyond. It would impress no-one. Lowball approach is still lowballing, especially in the face of what is known, and no matter how much it wants to masquerade as "realistic".

1.5 teraflops in docked and 1teraflops in handheald sounds bad imo

 

[Mr.Gamerson]

I don't think Nintendo will reference the actual numbers as much when showing off the hardware capability differences, but I do think they should have demonstrations showing how much more they can do with their games because of the improved hardware when they do their reveal presentation for Switch 2. Maybe have Jensen be a part of the show to demonstrate the improvements in mobile Tegra hardware since the original Switch.

 

[Simba1]

Reminder that the onus is on individuals to get a grip of their own emotions, rather than demand everybody else be negative under the facade of "keeping their expectations in check". If it doesn't happen, it doesn't happen. Nothing discussed here has been outside the realm of possibility, and he didn't pull that info from a magic hat. Even a 2GHz CPU would be clocked at around 60% of its capacity (3.3GHz), so, it's already clocked lower than its potential. But just remember that whatever you post here, it's going to have RT, DLSS, and it will need to be able to perform them competently, so, wherever you believe the final specs will land would do well to be reconcilable with that. A 1.5TF docked performance would doom it to failure, while needing to clock closer to the Switch Lite to get there, being a waste of the Earth's resources, and entirely unfit for a generational purpose in 2024 and beyond. It would impress no-one. Lowball approach is still lowballing, especially in the face of what is known, and no matter how much it wants to masquerade as "realistic".

Lol, no one demanding anything and I certainly not demanding people to be negative. And I dont think I am negative same he dont think he is positive, I just saying what I thinking same like he or you are. Of Course it doesn't happen it doesnt happen.

Current Switch A57 CPU was also clocked way below its capacity (2GHz), reminder that runs only at 1GHz when regularly was used in phones running at 1.8GHz in claster of 4 and in Nvidia Shield was running at 1.9GHz. Nintendo could easily run A57 at 1.5GHz, but the went with 1GHz.
You could find A78 running at 3.3GHz usually at claster of one, even A78 at claster of 4 would be hard to find running above 2GHz, while I dont think there is mobile devices that runs 8 A78 cores running at 2GHz.
I disagree that 1.5TF not counting gains from new tech/arhitecture and DLSS would be doom or failure (it would basically be around 4x stronger than current Switch GPU only by raw numbers so realistically much more), or that Switch 2 needs to have specs that are comparable to Xbox Series S in order that some people say its strong enough.
Also I like how basically think that 1.5TF (my lowest estimate) is not possible but you don't commment Z0m3le high end estimate of 3.5TF.

Yes, Switch 2 will have impressive hardware and power in any case, but have on mind that we talking about mobile hardware and Nintendo needs to pay attention about small form factor (less space and more heating) and battery life, something that only home console manufactures doesn't.

T239 seems like a pretty sure bet?

Sure, but further customization could be done until manufacturer of Switch 2 parts starts.
I would love that Switch 2 has 3.5 TF, but I dont think its realistically to expect something like 3-3.5TF in basically handheld device, even Steam Deck thats much bigger than current Switch has 1.6TF.

 

[Hermii]

While we're discussing the past, what exactly were people expecting the NX to be that was in the realm of possibility at the time, especially with what was shown before it was called the Switch?

Nothing too crazy really. 16 nm and higher clocks was probably the most common assumption. There was also talk of A72 and TX2 memory bus width, but I don't remember how many people believed that. 100% off the shelf cought people off guard.

 

[Mr Swine]

Wouldn’t a Geforce MX 570 with 2048 Cuda Cores be closest to Drake as it’s a low power Ampere chip?

 

[D36]

Nothing too crazy really. 16 nm and higher clocks was probably the most common assumption. There was also talk of A72 and TX2 memory bus width, but I don't remember how many people believed that. 100% off the shelf cought people off guard.

I should have specified, but could you put it in layman's terms? I'm not all that tech literate, I'm barely able to parce what people are saying when they start getting into the nitty-gritty