NintenDuvo
Bob-omb
- Pronouns
- He him
Man according to some of you in this thread, there will never be a right time to acknowledge the next Switch since it could harm the sales of a 8 year old system , make it make sense
-
Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)
- Thread starter Dakhil
- Start date
you have your low poly in the bvh. the model you see can be as high poly as you want
NineTailSage
Bob-omb
What I found interesting were some The Phawx's views on mobile tech and where these type of devices need to go.
He praises Nintendo and Nvidia for what they did with og Switch because they focused heavily on min/maxing performance and efficiency.
Maybe in relation to other handheld gaming devices a 600Mhz Switch 2 would look very low clocked.
MLID also mentioned the possibility that Nintendo could be trying to low clock portable mode to get by with a passively cooled handheld and then in docked mode, receive forced air induction from the actual docking station (to achieve 4Tflops).
The Phawx actually had some interesting takes.
I especially liked the mention of how beneficial a 256-bit bus could be in a handheld gaming device, (while achieving a similar efficiency) by clocking the RAM low.
He has a lot of great takes on how manufacturers usually tend to throw Higher clocks at a problem, that doesn't always give proportional performance with the higher usage of power.
other handhelds throw out any notion of efficiency. when you're pulling 25W, Drake is never gonna win on paper specs
Shoe
Cappy
- Pronouns
- He/Him
They made a comment in this video about the possibility of not having a fan at all on Switch 2 (unlike Switch 1) and only relay on passive cooling, then having a much larger fan in the dock for significantly improved performance. Something like that could be indicative of the comments of the new dock having a "bulge" or whatever it was.
oldpuck
Like Like
- Pronouns
- he/they
If we knew the exact power consumption, we'd know the node, because we could reverse engineer from there. Best I can do is tell you what Orin (8nm) consumes in that configuration, then guess that 4N would be ~50% of that.
Orin at 800MHz consumes 2.2W per TPC, at least when the GPU is active. Drake has 6 TPCs, that's 12.4W. Guessing 4N would get you down to 6.2W. That's still way too high. 800MHz seems way out of the question.
- Pronouns
- he / him
Depends what angle you're coming at. If you think of it as some fractional version of the big boy consoles, then yeah, having one spec actually match them while other things are far behind would probably be overkill. But PlayStation and Xbox just had the smallest (by percentage) RAM increase ever by far, so if anything I think they got stuck with underkill and if Switch 2 can reasonably avoid being stuck with the same that'd be good. But 12GB would also fulfill that.
According to Sony's official data
PS1 962m software / 102.4m hardware = 9.4
PS2 1537m software / 155.0m hardware = 9.9
So 60% more games overall, or 5.5% more games per system. Really even more in favor of PS2 game buyers, since some number of those PS1 games would've been sold to people using them exclusively on PS2.
Sure. But taking this analogy further the Witcher 3 Switch port gets to have higher resolution, and also we're comparing it to the game on a half-power version of Xbox One.
Kevin
Chain Chomp
It's like those people who can't stop blaming pandemic for game delays. Some people just can't move on. Switch could be 10 years old and completely abandoned by consumers and folks would still try and justify why Nintendo is right to not announce a successor yet.
Yea that was just them speculating, but I don't agree with this.
Pretty much every device with passive (fanless) cooling have thermal throttling. That is when the system automatically down clocks if it gets too hot during sustained workloads (like gaming). Great for a smartphone, but a no go for a dedicated game system, You don't want your frames to start dropping after half an hour.
NineTailSage
Bob-omb
I think that was the conversation they were having that made it interesting.
The Phawx points out that these handheld devices that are running at 25w aren't doing so at peak efficiency, they are throwing more clock speeds to increase performance but they aren't addressing the initial bottleneck issue.
As gaming aficionados here we all understand that paper specs don't give a real world performance scenario of how games will actually run.
The best thing Switch 2 has going for it will be that games are tailor made for the actually hardware and not just running some PC port.
SoldierDelta
Designated Xenoblade Loremaster
- Pronouns
- he/him
We've all pretty much seen this at this stage though. You compare a game like Doom 2016 on Switch and you can see the cutbacks already. It looks rough in places. Then you look at Tears of the Kingdom and Xenoblade 3 and you wonder what the fuck Bethesda was doing making Doom for Switch until you remember that Doom was never originally made for Switch and instead consoles stronger than it. This has happened again and again for many systems, and I don't really see it changing any time soon.
NineTailSage
Bob-omb
We haven't seen what I'm expecting we will see with Switch 2 though...
Switch got a bunch of late ports that most really didn't expect to do much as far as sells go.
The brand is a known quantity though and how expensive games are costing to be made.
I can totally see it becoming the lead platform for many future games or having more efforts placed into maximizing full usage of T239.
OP
OP
Keep in mind that a higher RAM bus width does not come free since a higher RAM bus width requires more wires.
- Pronouns
- he/him
This is a neat idea in theory. Relying on passive cooling has a few benefits:
-No moving parts (more durable).
-More attractive; cooling fans generally aren't something people like in their electronics so much as they put up with them
-More space to work with; no fan and vents means that space can be allocated towards other things, or removed entirely
Thing is, I'm not really sure what Nintendo would do with that space. All signs point to a physically larger device, so it wouldn't be smaller despite being passively cooled. Just a larger battery? That would result in pretty sweet battery life, especially since it should have a lower power draw in order to be passively cooled. I'm not sure that's the tradeoff the majority of consumers want, though.
I somewhat selfishly want a larger power gap than 2x between docked and handheld modes, in part out of hope that they support 120 Hz displays through the dock* (I'm assuming the handheld display remains at 60 Hz). Lowering the handheld power is one way to get closer to that, but ideally they just bump up the clocks of docked mode more.
*The more I think about this, the less likely this seems - or even if it is hardware supported, the less likely it seems many games will take advantage of it.
It's not going to throttle. If Nintendo were to go this route, it'd be clocked low enough for the device to run indefinitely at a set clock with no fan. Though I suppose you could look at that as being pre-throttled (though it is different, as it'd be providing a stable clock).
7hatDeadCat
Tektite
- Pronouns
- he/him
also, as far as i know lpddr5 only comes with a 64bit bus width maximum, thatd force a 4 chip ram setup
Lancelot
Like Like
Wow, I must say the last two pages have been very interesting regarding the actual chip again...
Maybe a range of 600-650mhz is actually possible to achieve? If all these claims from MLID are actually legit, that could truly get us right into 1.3 Ghz by a factor of 2x, hence "near 4 TFLOPS". The scaling fits like a glove this way, and some efficiency enhancements as a byproduct of being a custom chip could really get us there.
- Pronouns
- He/Him
???
One of the biggest games to release just last year was a game that was absolutely hit by the pandemic, having been in development since 2017. And if you're talking about the other day when I brought up that Switch 2 development was hit by the pandemic, then like.. yeah, it woulda been, unless you believe development didn't start until just a couple years ago, which we know would be wrong because of how far back evidence of the chip stretches.
Unless you just don't think the pandemic was a massive thing that walloped multiple industries over the course of an entire year and then had far-reaching effects on schedules for the following years?
I know especially in America there's been an attitude of "pandemic's over, get over it" since like 2021 but uh.. it wasn't, and we're still feeling the effects of it in many ways.
Goodtwin
Like Like
Anyone claiming 8nm, especially those who claim to have sources, but then do not address the obvious problems of power draw for a chip that big are either very ignorant or are intentionally avoiding the elephant in the room. I'm not saying 8nm is impossible, but when talking about T239 on 8nm, if power draw concerns aren't a topic of conversation, then I have to question your knowledge of the topic at hand. I want to slam my head on the table when the DF guys talk about T239 being on 8nm, but nobody ever brings up power consumption concerns. It's a big ass concern for a portable console. Erista Switch units were already on the low end of battery life at 3 hours, hard to imagine T239 getting that on 8nm.
Giancarlo
Nintendo connoiseur
it seens that everyone forget, Nintendo has stated Nintendo Switch will be they main focus this year, how can you expect Nintendo to knowledge it next console, if Switch is still selling well and it main focus?
Nintendo president says current Switch will be Nintendo’s “main business” in 2024
Nintendo president Shuntaro Furukawa has said in a conference call after the earnings report that the current Nintendo Switch will be the company’s main business for 2024. This comes after speculation…
mynintendonews.com
OP
OP
Yes for the LPDDR5 modules since smartphone SoCs only have support for a 64-bit RAM bus width. And I don't really expect that to change anytime soon.
Steve
Paratroopa
- Pronouns
- he/him
We see these affects especially for Sony and Xbox, since all the exclusives that were meant to arrive, most likely got delayed indifferently.
That doesn't account chip shortage and scalping.
Like in a world where covid didn't happen we would have gotten the Switch 2 earlier, same goes for Totk, pikmin 4 and much more. Like, damn, those Switch sales number when covid was happening is insane, especially Animal crossing and how that sky rocketed animal crossing into a huge Ip.
LiC
Member
MLID having an accurate video about PS5 Pro specs really shouldn't be taken as a sign of credibility as a source. Considering the same specs were reported on from other outlets very shortly afterward, and confirmed to come from third-party developer portal-type info, it's pretty clear that some third-party dev was leaking those documents to multiple places, and MLID was just the fastest (read: did the least due diligence) to share them.
If the same thing happened with Switch 2 info, then we'd also expect to see it from much more reputable outlets than this. But that's not what happened, since MLID is claiming that his information comes from talking to people who work at Nvidia, and even worked on T239 itself. Which is kind of MLID's gimmick as I understand it. He just fabricates quotes from what would be incredibly rare firsthand sources, usually fictitious AMD employees, making whatever point he or his audience want to hear. It's absolutely ridiculous to imagine somebody with that level of access and all these carefree inside sources authorizing a YouTuber to share their exact words, and he then goes on to get things majorly wrong all the time anyway. But that's somehow what MLID is trying to sell you.
If the same thing happened with Switch 2 info, then we'd also expect to see it from much more reputable outlets than this. But that's not what happened, since MLID is claiming that his information comes from talking to people who work at Nvidia, and even worked on T239 itself. Which is kind of MLID's gimmick as I understand it. He just fabricates quotes from what would be incredibly rare firsthand sources, usually fictitious AMD employees, making whatever point he or his audience want to hear. It's absolutely ridiculous to imagine somebody with that level of access and all these carefree inside sources authorizing a YouTuber to share their exact words, and he then goes on to get things majorly wrong all the time anyway. But that's somehow what MLID is trying to sell you.
Aufhebung RPG
History is tragedy becoming comedy
I'm more excited to see what EPD's reliance on drake's RT performance will do for ray tracing and other rendering techniques.
ERezaSwitch
Piranha Plant
But they also said this "He then went on to say that they will discuss any future plans at its next earnings briefing." which is on the 7th.it seens that everyone forget, Nintendo has stated Nintendo Switch will be they main focus this year, how can you expect Nintendo to knowledge it next console, if Switch is still selling well and it main focus?
Nintendo president says current Switch will be Nintendo’s “main business” in 2024
Nintendo president Shuntaro Furukawa has said in a conference call after the earnings report that the current Nintendo Switch will be the company’s main business for 2024. This comes after speculation…
Steve
Paratroopa
- Pronouns
- he/him
Aufhebung RPG
History is tragedy becoming comedy
Portable consoles and home consoles can be described as completely different technical difficulties, and portable consoles are limited by battery technology and radiating technology to do much more.
I mean, 4t is perfectly feasible for drake, but when the theoretical peak is already set at 3.4tflops, it doesn't make much sense to further increase the clock frequency significantly and it affects the power consumption and endurance, I personally think that 1.1ghz and 1.2ghz would be the more probable clock frequencies (550mhz and 600mhz in portable mode).
Goodtwin
Like Like
I just watched the MLID podcast segment surrounding Switch 2. Everything he is claiming to have a source on, could simply be taking info from the Nvidia hack and also from skimming through a forum such as this one. Even the quote from his "source" states that 8nm would be a good fit because it's cheap, but doesn't actually say it will be 8nm. This is the source that is not close to the project. The other "source" never talked about 8nm. At this point, I think his sources are either fictitious or he has been duped.
8nm would actually be very interesting to disect and learn how Nvidia got it to work. It would be one hell of a science project. The question would remain, if Nvidia can offer this level of performance with this level of power draw on the inexpensive readily available 8nm process, wouldn't they be doing it for other manufactures? Even for mobile graphics cards, they could dominate the market with inexpensive high performance low powered graphics cards if they can get 8nm to operate significantly more efficient than any product on 8nm currently on the market.
8nm would actually be very interesting to disect and learn how Nvidia got it to work. It would be one hell of a science project. The question would remain, if Nvidia can offer this level of performance with this level of power draw on the inexpensive readily available 8nm process, wouldn't they be doing it for other manufactures? Even for mobile graphics cards, they could dominate the market with inexpensive high performance low powered graphics cards if they can get 8nm to operate significantly more efficient than any product on 8nm currently on the market.
I still wouldn't be surprised if it hits 4tf or near it in docked mode. Portable is a different story for battery life of course.
Aufhebung RPG
History is tragedy becoming comedy
Lancelot
Like Like
Well, it's more power. What else does it need to make sense?
Aufhebung RPG
History is tragedy becoming comedy
Considering the inches of power consumed by portable consoles, I don't see how such a breezy horsepower boost is going to help a console that is primarily a portable feature
Lancelot
Like Like
The 4 tflops figure is for docked mode, this is an hybrid console not just a handheld... There are no power concerns during that mode other than heat, but it's up to Nvidia to figure that out.
NateDrake
Chain Chomp
Correct. MLID got the PS5 Pro specs because they were available on the Sony dev portal and some foolish partner shared the documents with MLID. It wasn't due to investigation or conversations with individuals claiming to be in the know. However, due to that occurrence, people will now take these Switch 2 claims as credible. Much of what is being said (clock speed aside) isn't exactly new, either. Ray-tracing and such has been discussed since last year.
Aufhebung RPG
History is tragedy becoming comedy
We have to follow the clock frequency x 2 relationship between portable and docked modes, and 1.3ghz means portable mode will be clocked at 650mhz, which is technically perfectly feasible, I'm just not sure if this level of horsepower increase will affect endurance
oldpuck
Like Like
- Pronouns
- he/they
I mean, if it were me, I'd certainly grill them. But if I had a known Nvidia source saying "it's 8nm, I only worked on the memory controller, so don't ask me about power draw" I wouldn't then keep the info secret just because I didn't have all the details.
Sources are not obligated to spill every detail, and reporters* are told by half their audience that they don't burn their sources by disclosing every detail, and the other half for not having complete documentation. It'll be what it'll be
MLID's credibility doesn't rest on this claim alone. He's dubious for other, repeated behavior.
Lancelot
Like Like
Most likely not, it's within the realm of possibility. Again, this is all from MLID's video claiming to know the clocks of the device which are "below 800mhz", if they really intend to achieve 4 tflops, that's what it takes, 1.3 ghz as you said. That's what the numbers say, there isn't much room to speculate here.
oldpuck
Like Like
- Pronouns
- he/they
550MHz seems to be peak power efficiency. If Drake at 550MHz is too much power, then Nintendo should probably disable cores instead of drop the clocks significantly. It would be both more power efficient and cheaper.
I'd say anything between 500Mhz and 600Mhz is in the "highly likely" range. Nvidia has some limited ability to tweak the power curves, but if they do that, I would expect the curve to push the clock higher, not lower.
It's funny to watch this evolve from "OldPuck's loud theory, but just theory" to "hard fact" over the years.
I think the 2x relationship is the most likely one, but it's not locked in stone.
Aufhebung RPG
History is tragedy becoming comedy
I'm more concerned about whether horsepower has any effect on RT performance, and I'd like to know the cost of switch2 in portable mode with RT turned on.
darthdiablo
+5 Death Stare
- Pronouns
- he/him
And I thought MLID was mostly dismissed dozens of pages ago too.
RT is going to be more of a design choice rather than something you hope to have enough power to use. Because if it's the latter, you're never going to have RT.
Portable mode does have enough power for RT, you're just going to have to design around that limit
oldpuck
Like Like
- Pronouns
- he/they
Of course RT is affected by horsepower. It's affected three ways.
1) RT cores run at the same clocks as the rest of the GPU. Faster GPU, more RT power, slower GPU, less RT power.
2) RT is costly. The faster that non-RT code runs, the more time to allocate to RT, and vice versa.
3) RT cores speed up the process of light bouncing through a scene. You still need to draw that light, which comes down to regular shaders.
RT is not a switch that is either on, or off. There are multiple affects that you might layer into a scene with RT. You can do ray traced shadows, or ray traced reflections, or ray traced ambient occlusion, or any combination. You could do ray traced global illumination. Or you could fully path-trace a game and do 100% of rendering via ray-tracing.
And you can do each of these effects at higher or lower resolution. You can use more rays or fewer rays. Rays can have more bounces or fewer bounces.You can include all of the scene in ray tracing, or just part of it. Or you can do the whole scene, but with lower poly geometry.
There are hundreds if not thousands of combinations. But, to give you two examples: Unreal Engine actually kinda does have an RT switch. On low-end Nvidia hardware, turning it on costs something like 4% of performance. Control has a "medium" RT setting which is basically just reflections. Turning it on eats ~50% of performance on the same hardware.
So you can see that there is wildly different results here.
necrolipe
Universo Nintendo Owner
- Pronouns
- he/him
Last time I heard anything related to clock speed was around ~600Mhz, but I prefer to avoid commenting on because this was not locked on my talks last year
Kise Ryota
Then I'll say "I told ya"
If I'm getting something like 650-1300 (~2TF-4TF), then I really don't care if it is SEC 8nm lol
But I can't see those clocks being achievable with that node. What MLID says means nothing to me.
But I can't see those clocks being achievable with that node. What MLID says means nothing to me.
Last edited by a moderator:
