-
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.
-
Do you have audio editing experience and want to help out with the Famiboards Discussion Club Podcast? If so, we're looking for help and would love to have you on the team! Just let us know in the Podcast Thread if you are interested!
StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)
- Thread starter Dakhil
- Start date
D36
#WOHOYHASAWAKENED
- Pronouns
- He/Him
Ahhh, gotcha. What exactly would they have done for Drake?
OP
OP
Theoretically, more CPU cores per cluster (up to 14 CPU cores per cluster) and larger L3 cache (up to 32 MB of L3 cache).
I don't know if this has been discussed yet, but while we see the benefit of DLSS in terms of GPU performance, what sort of benefit would using DLSS give in terms of RAM bandwidth? We all know that Switch hits bandwidth bottlenecks regularly, and with the ROG Ally hitting the market, it's pushing upwards of 4.3TFlops of GPU power in its Turbo mode with only 102.4GB/s bandwidth, which imo is going for massive bottlenecks there.
[REDACTED]'s RAM bandwidth is supposedly just as much, but based on the tests information from the leaks, the device will be pushing for less on GPU load. That could still be starved somewhat. That is, until DLSS comes into play with how I look at it. Say there was no DLSS, and the target res was 1080p. It would require a certain GPU load and RAM bandwidth for what is being rendered. But there's a chance that either one could be a bottleneck. Personally, RAM bandwidth would get hit first. Incorporating DLSS Performance mode, and that res gets reduced to 540p. It won't necessarily be 1/4 the GPU and RAM bandwidth load, but it will certainly be lower than native 1080p. DLSS then brings that back up to 1080p, using prior frames, motion vectors, etc. That will use RAM bandwidth, and while I don't know how much, the overall would certainly be less than native 1080p.
So now a native 540p render with lower GPU and bandwidth loads vs 1080p. Now, imagine a dev having that amount of GPU resources available. They decide they want to up the settings of their renders, like from Medium to High. That requires additional GPU load. But how much additional RAM bandwidth would be required? Just as much? More? Less? If less, then that would help greatly in terms of bottlenecks.
[REDACTED]'s RAM bandwidth is supposedly just as much, but based on the tests information from the leaks, the device will be pushing for less on GPU load. That could still be starved somewhat. That is, until DLSS comes into play with how I look at it. Say there was no DLSS, and the target res was 1080p. It would require a certain GPU load and RAM bandwidth for what is being rendered. But there's a chance that either one could be a bottleneck. Personally, RAM bandwidth would get hit first. Incorporating DLSS Performance mode, and that res gets reduced to 540p. It won't necessarily be 1/4 the GPU and RAM bandwidth load, but it will certainly be lower than native 1080p. DLSS then brings that back up to 1080p, using prior frames, motion vectors, etc. That will use RAM bandwidth, and while I don't know how much, the overall would certainly be less than native 1080p.
So now a native 540p render with lower GPU and bandwidth loads vs 1080p. Now, imagine a dev having that amount of GPU resources available. They decide they want to up the settings of their renders, like from Medium to High. That requires additional GPU load. But how much additional RAM bandwidth would be required? Just as much? More? Less? If less, then that would help greatly in terms of bottlenecks.
SiG
Chain Chomp
I can see the ARM A720 as an upgrade path for a refreshed or "pro model" [REDACTED], but that also means cutting backwards compatibility unless there's some translation layer going on at the low level, and at that point, I think the GPU would be able to handle previous-Gen Switch shader-compilation translation in real-time.
They can compete now. You just don't see 8 A78s or 8 X# cpus because no product outside of consoles need that. Laptops aren't even using it, unfortunately, opting for 4 X# and 4 A#
Nvidia's best practice says that your assets should be tuned to your final output resolution after dlss. So your memory savings would come from the smaller buffers that make up the initial render. Higher quality assets will require more bandwidth
ertaboy356b
Resident Troll
Regarding 32-bit, can't they just emulate 32bit like how dosbox does it?
Shader translation is a CPU concern unless they build dedicated hardware into the GPU to handle it.
I don't expect Nintendo to do anything that would change the status quo of aarch32 until Switch 3, aside from perhaps not allowing its use in native Drake games.
DOSBOX is perhaps not the best example, but yes, CPU emulation is an option.
OP
OP
ertaboy356b
Resident Troll
What happens if non-compliant?
Probably that it can't be used, I imagine.
Mr Swine
Like Like
- Pronouns
- He
Seeing how Dread was nearing completion in Prime 3 and look how long time it took for the actual game to come out. I can see the save thing with Switch 2….
/s
Anyway I think we can rule out a Switch 2 this year as we will enter June in a couple of days and still no direct as long as the eye can see
There's no point in having a new cpu in a stealth revision. If you're going to inchrease performance, you can do so by clocking the old one faster. But probably, they would just use the efficiency for battery life.
ianm
Rattata
There is time till around June 28th for a direct, Nintendo announces it one or two days before. They have every opportunity to tease Switch 2 even for holiday this year. Nothing is off the table.
fiendcode
Like Like
We can rule it out because nothing was said in the annual report and shareholder's meeting. Drake will be next FY, not this one.
Ghostsonplanets
Ghost in the Sales
- Pronouns
- He/Him
Way past. Cortex X should be around Zen 3 - 4 performance rather than Zen 2 on the consoles.
Zemya
Shriekbat
Surely I might be wrong in the end, I want to, actually.
But for me it's looking quite bleak, the GaaS strategy smells like "throw as many as possible and if at least one sticks we have won". Sony is renowned for their top quality single player games so why make GaaS? They don't even look like they are gonna be the new Warframe or Destiny.
But this might be too out of topic.
OP
OP
And that claim turned out to be completely wrong.
MediaTek Partners With NVIDIA to Provide Full-Scale Product Roadmap to
MediaTek Partners With NVIDIA to Provide Full-Scale Product Roadmap to the Automotive Industry
corp.mediatek.com
eye0h
"it is known"
- Pronouns
- He/Him
They've been fixing leaky pipes maybeIt's amazing that there isn't even half a rumor about a Direct
OP
OP
Correct. Companies with the Cortex licence can't make any microarchitecture changes that Arm doesn't approve.
NVIDIA ACE for Games Sparks Life Into Virtual Characters With Generative AI
Custom Model Foundry Produces AI Models That Run in Cloud and PC TAIPEI, Taiwan, May 28, 2023 (GLOBE NEWSWIRE) - —COMPUTEX— NVIDIA today announced NVIDIA Avatar Cloud Engine (ACE) for Games, a custom AI model foundry service that transforms games by bringing intelligence to non-playable...
Thraktor
"[✄]. [✄]. [✄]. [✄]." -Microsoft
- Pronouns
- He/Him
I don't think this is necessarily the case. As I understand it, ARM's reference cores are typically completed long before they're publicly announced, to give chip designers like Qualcomm & Mediatek time to design SoCs around them. For example, the Neoverse V2 core was announced in September last year, by which point Grace had probably already taped out using the core. At a guess, the designs are probably finalised at least a year or more before a public announcement, given typical chip design timescales. Leaving the official announcement to ~6 months before chips appear using the cores is likely intended to give space for the previous designs to have some time in the market first, as they wouldn't want to announce the X4 and A720 cores before phones with X3 and A715 cores hit the market.
That being said, I would still agree this is too late for use in T239, even aside from the issue of no Aarch32 support.
And they are preparing to go to war with Steam because of the the deck being able to emulate switch games
Concernt
Optimism is non-negotiable
- Pronouns
- She/Her
There are rumours about a "digital event", weirdly enough, but nothing specifically about a direct.It's amazing that there isn't even half a rumor about a Direct
Concernt
Optimism is non-negotiable
- Pronouns
- She/Her
the problem is that Dolphin(the emulator) was going to be lauched on steam and well.... its nintendo
eye0h
"it is known"
- Pronouns
- He/Him
I think the sheer number of people emulating TOTK almost 2 weeks before release really pissed them off LOL
ReddDreadtheLead
#TeamLate2025WithAPotentialForEarly2026
- Pronouns
- He/Him
Er, no.
i think the X4 already outperforms those CPUs in single threaded by a noticeable margin, and when combined in multi threading should be close or above it by now in the optimal situation, ie it has the right and appropriate amount of caches and clocked right but all while consuming a lot less. Even with the Zen 2 having SMT enabled.
Though these are better for bursty loads, they worked to make it better for sustained loads.
So after a few gens (11 gens after the A78 so 2030-2031 Switch succ succ), Nintendo could go with an X and an A7 core in the same setup for efficiency and performance.
They now allow up to 14 cores in a cluster and 32MB L3.
Nintendo should be able to get another generational leap if they decide to stick with this Nvidia x ARM relationship. Maybe not as big as Drake over the switch, but definitely a noticeable gen over gen leap again.
I wonder if Nintendo will go for a 10 core CPU for the next next system. Or if they’ll stick to 8.
In this case, it’s 4Xn + 4A7n + 2A5n for the OS.
Or, 8A7n+ 2A5n since that’s possible on ARM v9
man that dialogue sucked. but hey, RTXDI!Correct. Companies with the Cortex licence can't make any microarchitecture changes that Arm doesn't approve.
NVIDIA ACE for Games Sparks Life Into Virtual Characters With Generative AI
Custom Model Foundry Produces AI Models That Run in Cloud and PC TAIPEI, Taiwan, May 28, 2023 (GLOBE NEWSWIRE) - —COMPUTEX— NVIDIA today announced NVIDIA Avatar Cloud Engine (ACE) for Games, a custom AI model foundry service that transforms games by bringing intelligence to non-playable...investor.nvidia.com
Arm GPUs built on new 5th Gen GPU architecture
This blog provides details about the new Arm GPUs, including Immortalis-G720, Mali-G720, and Mali-G620, that are all built on the new 5th Gen GPU architecture.
community.arm.com
ARM is working with Epic to bring the desktop renderer to mobile. that's a step closer to lumen and nanite supported on mobile
NintenDuvo
Bob-omb
- Pronouns
- He him
Where is the rumor about a digital event???
OP
OP
Hajeman
Handheld
- Pronouns
- Mr.
Who didn't angry when peoples can play their hardly developed games without paying. But TOTK already got guinness world record for the fastest selling Nintendo games ever (pokemon sv is not 1st party game, it's 2nd party)
NintenDuvo
Bob-omb
- Pronouns
- He him
Ethan Gach is Kotaku's senior editor.
Speaking about rumours:
Regarding the showcases anyone can make that claim unless this person is known to be a trusted insider? and in regards to the hardware did i miss something where Nintendo said no new hardware this year??
Erathix
Famiro
- Pronouns
- He/Him/They/Them
Again, off topic but you're right, it is quite bleak and there's no other way to put it. I will make this my final rant in this thread but damn I can't help but vent. No TL
R.These companies keep chasing the dragon and throwing so many different ideas at the wall hoping so badly they will be the next Fortnite or the next Destiny. Yet they fail, so many fail and they keep trying to pursue it desperately to the point of sheer madness. All they obsess over is the revenue of a game like that.
The odds are 1%. You have a 1% chance that your game will be as big as those games. To have a live service that can be as continuously big over a decade and keep making money without losing momentum to the point where it's an unstoppable juggernaut and not much can move the needle downwards (I mean, Destiny has made a lot of questionable and awful decisions and that game is still going strong and gaining players.)
A lot of these publishers don't even understand why these games are popular. It isn't the model itself at all, it's the gameplay. The gameplay and even the story, lore, etc, how it all works together, that is what makes these games successful. Their gameplay is so strong that people want so badly to overlook the worst elements of the games, the more predatory elements, because they enjoy the act of playing the game.
The funny thing is, when obsessively trying to make a game like that they forget the most important aspect: to make the game actually good and worth playing first, then start creating a plan to be predatory and suck all soul out of the game. The first part will hook enough suckers it softens the blow of the second and will even make the second part have a higher success rate.
They make these games with zero direction, designing the cash shop and ways to monetise first then building the game, trying very, very pathetically to leap onto every trend imaginable in the multiplayer space to make something utterly unoriginal and lifeless, then they try to find a style for it, which is usually very try-hard and see-through, not to mention eye-roll inducing and/or bland as hell. Slap some name onto it and, oh look! We have our new live service games we think you will definitely bite onto and invest your time and money in.
Except, thankfully, more often than not people don't bite and the game ends up falling on its face from launch onwards and it goes downhill from there. Another issue is the fact that in the desperation, they will turn studios with zero experience in making a live service and push them into making one. This shows greatly when a developer has no experience or interest in making a game like this, the game will end up with a million problems on launch and they have a roadmap that is very unambitious and devoid of content even by live service standards. This then spirals as players try to find a reason to continue playing the game before dropping it as there's absolutely nothing else to do in it and a non-existent drive to add anything that will motivate engagement with the game. If people decide the game is not worth investing in at all and vote with their wallets, they cannot continue to support the game and the Devs will inevitably cease the service, rendering all money that any gamer foolish enough to give to said live service, completely mute and a waste. Burning the money would have more use than that.
So when Sony says they want to make 12 of these things and expect their revenue to consist a lot more of live service revenue...I think they are delusional and haven't even looked at a lot of other publisher's who have tried and are slowly realising the risks associated with it. Even EA is trying to make things right again and make more Single Player titles after going through the obsessed live service phase and stating with deluded confidence that Single Player games were dying.
PlayStation is a brand heavily associated for and routinely praised for Single-Player video games that consistently win GOTY awards and are always part of the conversation in the industry. To redirect focus on not just Multi-Player games, but Live Service titles, in such a bold, all-in kind of way, is bleak. Knowing the talent they have and how Dev cycles are getting longer and more strained, that is years of other games we could've gotten for this console flushed down the toilet to make projects that let's face it are destined for failure.
Nobody is giving a shit about something as cringe and as tone-deaf ironic as Fairgame$. Talk to people and they have no idea what Concord is meant to be from that CGI trailer and don't have interest in it. When people realise Marathon is an extraction shooter rather than an FPS with a Single Player and a Multi-Player component, they tune out, we had Rainbow Six: Extraction and that failed, we have Tarkov, COD, Bungie has a high pedigree for their gameplay but can you seriously design a live service focused so heavily on just one mode without bothering to make other modes so players who don't care about Extraction modes in their video games can engage in any other way? Oooo, discount Splatoon from Square Enix, because when I think Square, the first thing I think about is how great their live serv...oh wait, The Avengers, Babylon's Fall, FFVII: The First Soldier (granted, mobile title, different league but still), Chocobo Racing GP.
Yes, such success, much wow. I am quivering so hard for that live service success money right now. Nintendo is sitting in the corner laughing about the audacity to try and make Splatoon from Wish.
What about Factions? I smell troubled development, straining to see how we can go about making this a 10 year thing and how many skins we can make? Raise your hand if it's endless Ellie and Joel ones, it probably is!
Helldivers 2? Throw it in the pile of countless other squad based survival shooter games in the vein of Left 4 Dead trying to turn themselves into live services. I doubt anyone is begging for the sequel to Helldivers to be another one on the assembly line.
But go on, make Anthem after Anthem after Anthem, chase that dragon, believe you have a vision so grand that you can make the next Fortnite, lightning in a bottle viral games that are rare and don't happen often nor can be capitalised on to such degree's often to the point where people are simply fine with accepting it for what it really is.
A lot of these, if we're lucky all of them, are going to fail because they are so tired, derivative, soulless and boring. Blending into each other like a big ball of bullshit being marketed as the top shit to get excited about.
If Nintendo did a direct with the majority of their first party exclusives being solely live services, everyone would lose their shit online. The worst direct of all time.
If anyone thought that the above was a good future for gaming, they have probably grown up in this environment of all these games being forced down their throats that they have never seen what gaming was like in the past to know that this isn't okay and isn't the direction you would want the industry to keep pursuing.
I hope for Nintendo that as they evolve with their consoles that they don't lose sight of the goal of making games first and foremost, fun and engaging entertainment. Look at the Indie scene and wonder why Indie's are taking more of a foothold in the industry. It's not just to fill in the gaps between longer AAA development. It's the fact that a lot of them are making games for fun, they are making games that are simply GAMES. Top to bottom, start to finish, pay once and that is it, you're done, you can play everything they have to offer over and over again for life, they will not shut down and they do not demand more money from you. Hell, games like Hollow Knight make you wish you could give them more money because at the price they sell it for it is a legitimate steal for how much passion was put into it.
Games are always better when you can feel the clear love and passion the Devs had making it, when you play it and know that this was exactly what they wanted from beginning to end. When you actually care to stick around for the credits because you want to see all the human beings behind the game that you lost yourself in because it was genuinely brilliant. That is a game.
I hate games that reek of corporate, where it feels like the Devs felt obligated to make it but nowhere near happy doing it, where every decision surrounding it stinks of other hands poking in and altering things, tinkering things, to adhere to how they believe games should be. To reduce them solely to being about money, nothing else, there can be nothing else, product is product, that's it, they can't be anything more. Just shut up, take it, pay up, don't think, don't feel, don't play for fun, games are meant to be your second job, to make ME richer.
Can't wait to see how this whole Summer plays out with all the different showcases and seeing which ones are the most corporate and which ones have the most genuine games that come out of them. PS Showcase crossed off the list and set the bar very low, let's see how others fare.
Erathix
Famiro
- Pronouns
- He/Him/They/Them
Ahhh yes, Kotaku, beacon of video game journalism and trusted source of leaks.
I. Don't. Buy. It.
ziggyrivers
thunder
- Pronouns
- He/him/his
Ethan Gach is Kotaku's senior editor.
Speaking about rumours:
lol Nintendo always has at least one digital showcase. They also almost always have up to two big showcases. It’s just safe guessing.
And that rumor makes me wanna grab it with a grain of salta because Nintendo never said that a new console would release after 2024
TomNookYankees
Koopa
- Pronouns
- He/Him
The part that seems weird is to throw in "coming weeks and months", suggesting that either Ethan is hedging his bets by using a phrase that includes the entire summer or is implying that only one of these will be within June and the other will be later in the summer.
eye0h
"it is known"
- Pronouns
- He/Him
Speaking about rumours:
I love this dude he's the biggest litmus test for journalists and content creators since fucking zippo it's awesome
Hosermess
Koopa
I mean, at some point they will announce something for the 2H. But a digital event could be anything. A Direct, a Treehouse, another MP4 delay announcement. They could also repeat TotK's 10 minute gameplay showcase but this time with Pikmin 4. It's cool he's got "sources", but when the basic reaction is "and?" was it really necessary to tweet it out? Free and easy engagement I guess.
Last edited:
kirbyfan407
Twinkle Popo
- Pronouns
- He/Him/His
Ethan Gach is Kotaku's senior editor.
Speaking about rumours:
The Ethan Gach tweet reads like an intentional joke to me.
Tentacle-tropes
Kremling
No they are not. From the latest update on that it was Valve who contacted Nintendo. There is no gain for Valve even attempting to shield this.
KeijiKG
Moblin
I don't really put much stock in the new hardware reveal being in a Direct.It's amazing that there isn't even half a rumor about a Direct
NineTailSage
Bob-omb
Not to dive to deep into the Nintendo DMCA to Valve story but the timing is pretty interesting since the Steam Deck running Dolphin emulated games in of itself represents the marketing and selling point of the space where Redacted will occupy...
As others have mentioned I'm sure Nintendo hearing stories of their most recent masterpiece TotK being emulated on day 1 was just salt in the wound of whatever marketing strategies they have been coming up with to sell Redacted once its released.
As others have mentioned I'm sure Nintendo hearing stories of their most recent masterpiece TotK being emulated on day 1 was just salt in the wound of whatever marketing strategies they have been coming up with to sell Redacted once its released.
It means he doesn't know a damn thing and isn't really worth much attention lmao
D
Deleted member 887
Guest
I want to single out a recurring theme in this video, rather than talking about the 4060 Ti overall.
There are areas where this card just falls over in performance relative to its predecessors, as memory demands scale up. That's because Nvidia has cut the memory bus and compensated with a larger cache, but there will always be workloads that simply aren't cache friendly, and no increase in cache size can overcome it.
When talking about the "advantages" that Ada brings to the table, its increased cache is often mentioned, and in isolation, yes, more cache is an improvement. But in the context of the overall Ada design, it simply isn't cost effective to continue pushing wider and wider memory busses gen-on-gen. And because of that, the L2 cache has to get larger to compensate.
Truth to be told, lower res targets tend to work well on low-bandwidth-high-cache designs, so REDACTED is probably a great target for this particular tradeoff, but all else being equal, I'd rather have the bandwidth over the cache. Cache also tends to just be physically large, so on a space constrained device, the perf/die area trade off, bandwidth might be worth it (even if on the perf/watt tradeoff tends to favor cache).
AMD has done some really excellent marketing to transform this arguable deficit into a feature with Infinity Cache. So much so, that there is this narrative that the Playstation and Xbox consoles are hobbled because they don't have it, despite the fact that the Series X and PS5 are bandwidth rich environments. Dropping Infinity Cache and just Throwing Bandwidth At It was probably a net performance win for PS5 over desktop RDNA 2.
Designing for any system is about artfully balancing all of these various bits to suit your target use case as well as possible. Nvidia and Nintendo seem to be in a similar place to Sony and AMD, where they're getting a custom solution halfway between one arch and the next. There is this narrative that Sony got "RDNA 1.5" but if you look at the bandwidth story specifically, you can see how Sony actually got something better than either gen, by being very selective about what it got, and targeting the console space very well. (Mark Cerny is very good at what he does).
I just really want to counter any sort of checklist narrative about REDACTED. There is the obvious version where Nintendo "wins" if it gets Ada and "loses" if it gets Ampere, which I think everyone in this thread understands is silly. But there is a subtler version where Ada is a list of features to backport, and Nintendo's "score" goes up the more it gets. In the case of cache/memory bandwidth, not only are not all features desirable for REDACTED, they might not even be features.
ReddDreadtheLead
#TeamLate2025WithAPotentialForEarly2026
- Pronouns
- He/Him
4060TI has really low cache though, so its hit rate is pretty poor relative to its GPU size and it’ll depend more on the GDDR memory. 4090 has a gargantuan cache that enables it to have a much better hit rate.
For a product like Redacted, more tiers of cache can help over a larger cache setup at reducing the bandwidth requirements. But also helping it manage the latency better.
The PS5 is only RDNA1 with RT so it was already going to be in the basis of RDNA 1 development, likewise for SX/S but more RDNA2 added.
Concernt
Optimism is non-negotiable
- Pronouns
- She/Her
Yeah, they never said that.
ReddDreadtheLead
#TeamLate2025WithAPotentialForEarly2026
- Pronouns
- He/Him
yeah
Last edited:
