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

[Alovon11]

Are you guys leaning more towards an early 2023 for the Switch revision?

If this Switch revision is a customized Orin NX, would there be enough time to produce enough for a late 2022 relase?

Honestly, at this rate, anything past March 2022 is possible now XD
Most likely times are Summer or Fall 2022 IMHO as 2023 would be a bit late time-wise relative to Orin and the Devkits.

 

[Pokemaniac]

On UFS, UFS memory cards definitely seem to be rare as hen's teeth right now.

Nintendo's licensing partnership with SanDisk as the "official" microSD for Switch would be in jeopardy, but if SanDisk isn't ready to provide what's necessary, so be it.

That said, even Samsung has seemingly run dry of UFS memory card inventory, since I can't find one for sale except on eBay. Perhaps that would change if they had a vendor looking to use them like Nintendo. But with so few electronics manufacturers using them and me not being sure that Samsung would go the licensed vendor route like SanDisk, Nintendo could potentially do "private label" production with Samsung and sell cards branded as solely a Nintendo product, which would practically be like having proprietary media without it actually being proprietary.

But with that rarity, it seems like UFS is up in the air unless Samsung manufactures more for Nintendo's needs. And it seems unlikely that they would go eUFS internally and microSD externally, wanting to get as close to parity read speeds across both storage options. Nor do I see them going UHS-II only for microSD.

One of the larger benefits of UFS memory cards is that Samsung sells a 128GB UFS card for $30, so it's not any more expensive than UHS-I microSD storage currently is, but with UHS-II speeds. Its only drawback in Samsung is currently the only manufacturer.

I think the idea behind UFS cards is that in theory, it should be easy to pivot older eUFS manufacturing capacity over to making them. The problem is that they've never been in demand enough for that to actually happen.

Are you guys leaning more towards an early 2023 for the Switch revision?

If this Switch revision is a customized Orin NX, would there be enough time to produce enough for a late 2022 relase?

Second half of 2022 seems like the most likely target right now, with a possibility of it slipping into 2023 if something goes wrong.

 

[Terrell]

Honestly if nintendo cared about power again they would just release a normal console, I think switch dane will be more powerful than the base model by a decent amount but I don't think they 100% care about having the most top of the line in power. After the GameCube they stopped caring about that and more care about making the device as cheap as possible and power comes second.

Wii U was actually made explicitly to be the most high-powered device they could get from the PowerPC architecture that they needed to make Wii backwards compatible without headaches and was designed that way on purpose. Switch was one of the most powerful devices in its form factor/power envelope when it was released (not bleeding edge, given the circumstance of when Tegra X1 was first produced, but no slouch, either).

And while I think people suggesting a $450-500 piece of hardware are pushing the limits of what consumers will pay for the type of device it is, I believe it's a mischaracterization to say that they build hardware on the cheap or don't care about power. They most certainly do care about it, within the context of the form factor that clearly worked out for them, because that's the niche they carved for themselves that currently no one else seems intent on following them into.
... well, no one except for Valve because, as someone pointed out earlier, Switch has started to become the sales leader for indie titles (which Steam used to be without question) and needed a place in that space to compete on equal footing. Despite their attempt to say that Switch made zero impact on the decisions surrounding Deck, no one in their right mind takes Valve's transparent statement at face value, which is why it continues to elicit comparison (undoubtedly much to Valve's chagrin).

But that indicates the #1 reason hardware horsepower is going to matter within the confines of the form factor: If Sony or Microsoft or anyone else decide to make a play for the space Nintendo has carved out for themselves in the dedicated hardware games industry, they will want to fight for it, because it's all they've got now. So that means heading competition off at the pass, even if they don't see any competition coming or even if corporations vying for the same dollar in the same space say they're not in competition when so many people clearly see otherwise.

Are you guys leaning more towards an early 2023 for the Switch revision?

If this Switch revision is a customized Orin NX, would there be enough time to produce enough for a late 2022 relase?

My personal deadline has been March 2023, but the hard "deadline" would be March 2024. At this point, the only reason to push it back is to make sure that there's software ready to go.

 

[japtor]

I apologize for the error regarding NVMe.

I looked up something, you mentioned, on Amazon and I don't seem to see micro SD Express, that makes me think it's not a widely used standard.
As for the M.2's I understand the prices are good although I have some doubts about the actual size and how they might make it "easy" to fit in a slot.

SD Express in general is barely a thing on the market itself, like the manufacturer side parts were only recently starting to come out, so yeah I wouldn't expect micro SD Express yet. When I said early, I meant really early. The spec has been around for a while but this year was the first there's been anything concrete production wise, so maybe there'll be more SD Express stuff (and micro) next year.

I think the idea behind UFS cards is that in theory, it should be easy to pivot older eUFS manufacturing capacity over to making them. The problem is that they've never been in demand enough for that to actually happen.

The only place I've seen to even buy one right now is the 128GB one straight from Samsung, which has an amusing review saying it doesn't work with the Galaxy S20. Hell the little header note says they're "only compatible with Samsung Notebook". If Samsung won't even commit to and help push it themselves with their volume of phones I can't see Nintendo taking a chance.

 

[ShadowFox08]

Are you guys leaning more towards an early 2023 for the Switch revision?

If this Switch revision is a customized Orin NX, would there be enough time to produce enough for a late 2022 relase?

I think it could go either way. I don't see us getting switch 2 until Q4 2022 at the earliest. But Q1 2023 to Q4 2023 is more than possible for 2-3 reasons:
1. They don't want to screw over OLED owners by releasing another console a year or less after release
2. We have a transistor shortage that is heavily predicted to last for the entirety of 2022
3. Save costs and they may not think it's necessary to release before 5 years because switch is selling well..

But if it does get released in 2023, I imagine the chip being taped out later then projected and not even using 8nm at that point.

I'm going to guess somewhere between Q4 2022 to end of Q1 2023. Not like they have to release the new console in March or anything. I do think it's very likely they will release botw 2 with switch 2 (as well as release on OG switch models) and 2023 may seem a bit drawn out to release botw 2.

 

[Branduil]

I don't see much reason to release in 2022 when Switch is still going strong and they won't even be able to make enough of them for demand. Better to build up inventory and get one more big holiday out of Switch before releasing in 2023.

 

[Terrell]

The only place I've seen to even buy one right now is the 128GB one straight from Samsung, which has an amusing review saying it doesn't work with the Galaxy S20. Hell the little header note says they're "only compatible with Samsung Notebook". If Samsung won't even commit to and help push it themselves with their volume of phones I can't see Nintendo taking a chance.

Oh wow, they're back in stock for a change.

The main reason is that Nintendo isn't obligated to follow the rest of the consumer electronics industry.

UFS has the same problem as UHS-II, UHS-III and Express microSD: limited devices where its advantages are necessary. Up until this point, the only use-case that makes UHS-II or higher mandatory is videography and that business has not adapted to UFS due to existing SD cards in peoples' possession (because UHS-II predated UFS by at least 5 years) and inertia. UFS was designed for a use case that didn't want to change and when it did, it went with CFexpress due to not needing a super-small form factor like other devices, being 4-lane and using PCIe/NVMe, meaning super-duper low latency for mirrorless video camera capture. UHS-I was fast enough to load music and pictures and other such things on smartphones and still cameras, so there was no incentive to push beyond that standard for most other consumer electronics.

But as fast read/write speeds are going to become necessary in the video game space and Nintendo and Valve are the only ones operating in the portable device space currently, you either have to choose between fast removable storage or far more fast internal storage. Valve clearly picked option #2 as its primary response (you can load games off the SD card on Deck, but they suffer load time increases and likely performance issues like you see with some Switch games).

With UHS-I SD cards already running up against the limits they impose on smooth gameplay for sub-1080p games, Nintendo will have to choose between UHS-II or higher SD cards or UFS cards. UHS-II and UHS-III card readers are quite expensive, as are the cards themselves. Meanwhile, I can get a 128GB UFS card with UHS-II speed from a reputable manufacturer for $30, which for those read/write speeds is a steal compared to what you'd pay for a similar UHS-II card. And who knows what the bill of materials for a UFS card reader is.

Lastly, gaming is one of the few spaces in consumer electronics where you can use a peripheral that no other device does or can and get away with it (see example: proprietary SSD add-ons for Xbox Series), so long as you aren't price-gouging (see example: PS Vita memory cards). If it's offered at a great price, so much the better.

 

[ReddDreadtheLead]

I;ve said this for a while. Steam deck is a defensive move. I think there is indeed a segment of PC users who appearciate the flexibility Switch provides with indie titles and unchallenged Switch would eat into Steam's appeal as an indie platform. But it's hardly a direct competitor in the way some people have positioned it as and they'll end up disappointed by it both in terms of its reach and valve's ambition.

Steam deck's reveal and delayed launch also puts it in an awkward position where by the time it's available in any real numbers, Switch 2 would very likely have been announced and on its way, likely at the same cost if not slightly cheaper as the cheapest steamdeck SKU with more features and power.

The only way the ecosystem could remain appealing to the people positioning it as a Switch killer is with a Steamdeck Pro, which is just as unlikely given the first one isn't even out. someone compared the deck to Blackberry's iPad killer, which had great specs but launched late and by the time it was ready the technical advantages vis-a-vis iPads had evaporated. I wouldn't go that far, but it's market positioning as a direct Switch competitor isn't really there. I'd go as far to say it's not really trying to compete with Switch, some fanboys just want it that way because they dislike Nintendo succeeding.

And absolutely correct on the point that's it's a laptop PC in a Switch shell. it's direct competitor are GPD wins of this world, and the OS layer will be substantially heavier than the Switch 2s that a lot of memory and CPU power will be wasted servicing it as a windows device.

I was mainly pointing out that 12GB for a switch is probably fine if the OS is super lean like the current switch.

Steam Deck(Valve) was thinking lower but went with higher for longevity reasons. Switch probably doesn’t really need 16GB of RAM (though devs would love it), while SD would need more for its package relatively because of what it is.

Not really about the competition stuff which is a different matter.


I think 12GB is a better than fine amount for this device. 8GB is fine as well. 16GB is probably overkill for such a device honestly.

 

[Skittzo]

I don't see much reason to release in 2022 when Switch is still going strong and they won't even be able to make enough of them for demand. Better to build up inventory and get one more big holiday out of Switch before releasing in 2023.

Building up inventory costs extra money and they won't be able to make enough no matter when it launches.

 

[ILikeFeet]

games have been using RTXGI already, but the first game with infinite scrolling volumes is about to be released. ray traced probe grids are one of those things that could, theoretically, be accelerated by the minimal amount of RT cores in Dane

ICARUS - Prepare for Beta Weekend 7 | With NVIDIA RTXGI, Photo mode and Giveaways - Steam News

This weekend brings beautiful graphics and bountiful prizes.

store.steampowered.com

 

[Ozzie]

Curious question, so this Dane is supposed to have DLSS and much better CPU. Does it compare to Steam Deck on a similar level in terms of overall power or is it stronger or perhaps have a better advantage due to DLSS?

 

[Alovon11]

Curious question, so this Dane is supposed to have DLSS and much better CPU. Does it compare to Steam Deck on a similar level in terms of overall power or is it stronger or perhaps have a better advantage due to DLSS?

By the output? Waaaaaaaayyy stronger.
The only thing that the SD may nudge it (emphasis on may) out on is Single Threaded CPU performance but Dane will have 2-8more cores than SD (6 to 12 Cores is the range now thanks to Orin NX).

And GPU wise Dane may beat it before DLSS if the clocks are right.

The only thing Dane will loose to SD likely is GPU horsepower in portable mode, but even then NIS exists so even without DLSS that gives they mode a boost even if they disable DLSS in portable mode.

 

[Dekuman]

By the output? Waaaaaaaayyy stronger.
The only thing that the SD may nudge it (emphasis on may) out on is Single Threaded CPU performance but Dane will have 2-8more cores than SD (6 to 12 Cores is the range now thanks to Orin NX).

And GPU wise Dane may beat it before DLSS if the clocks are right.

The only thing Dane will loose to SD likely is GPU horsepower in portable mode, but even then NIS exists so even without DLSS that gives they mode a boost even if they disable DLSS in portable mode.

You forgot the most important part.

Dane Switch will be smaller and have longer battery life

 

[Alovon11]

You forgot the most important part.

Dane Switch will be smaller and have longer battery life

At the very least way lighter and thinner, yeah, and battery life on average should be higher.

 

[LiC]

It's obviously impossible that the next Switch won't support SD cards and advertise them as the main way to expand storage. They're too ubiquitous and I would wager the vast majority of Switch owners don't even know what the alternatives are (in addition to most already having a card they'll want to continue using). Now, it's not impossible for them to support SD + another format, but that's extra money and likely extra space in the form factor. If Nintendo cares about fast storage in the next model, and the price point is going up anyway, they should just spend more on the speed and capacity of the internal storage. Releasing multiple SKUs with different capacities, like the Wii U, is an option.

Also, I don't see a lot of discussion about the game cartridges, but I think they're a pretty good example of why inertia is going to dictate most of the storage features of the next model. Even the current carts, which are smaller and slower than SD cards, are still too expensive for a lot of publishers almost 5 years in. It doesn't seem like a jump forward to new, more expensive storage technology is in the cards for any part of the Switch hardware proposition.

 

[fwd-bwd]

I think the idea behind UFS cards is that in theory, it should be easy to pivot older eUFS manufacturing capacity over to making them. The problem is that they've never been in demand enough for that to actually happen.

This. The demand is the real issue for UFS Card, not supply. Some manufacturers were pinning their hopes on the demand from drones and phones with high-res cameras, but that hasn't panned out because drones are not of mass market and high-end phones already have massive internal storage. Nintendo can potentially become the largest single consumer of UFS Cards.

Nintendo's licensing partnership with SanDisk as the "official" microSD for Switch would be in jeopardy, but if SanDisk isn't ready to provide what's necessary, so be it.

SanDisk is a Western Digital brand; the latter manufactures eUFS storages. SanDisk probably can keep the partnership if WD is willing to refashion their eUFS storages into UFS Cards. Aside from Samsung and SanDisk/WD, there are other potential partners: Micron/Crucial, Kioxia/Toshiba, SK Hynix, Phison, Yangtze, etc.

Also, I don't see a lot of discussion about the game cartridges, but I think they're a pretty good example of why inertia is going to dictate most of the storage features of the next model. Even the current carts, which are smaller and slower than SD cards, are still too expensive for a lot of publishers almost 5 years in. It doesn't seem like a jump forward to new, more expensive storage technology is in the cards for any part of the Switch hardware proposition.

That's a fair assessment. If Nintendo would not increase the speed of game cartridge (to keep the cost low for 3rd parties), installation to internal storage might become mandatory; in that case keeping the slow microSD is fine. On the other hand, if Nintendo is interested in an eventual digital-only SKU (no cartridge reader), they might need a removable storage solution such as proprietary drives or branded UFS Cards.

 

[Terrell]

Also, I don't see a lot of discussion about the game cartridges, but I think they're a pretty good example of why inertia is going to dictate most of the storage features of the next model. Even the current carts, which are smaller and slower than SD cards, are still too expensive for a lot of publishers almost 5 years in. It doesn't seem like a jump forward to new, more expensive storage technology is in the cards for any part of the Switch hardware proposition.

There was a suggestion that Nintendo and Micron were working on 3D NAND to increase capacity for cheaper than the current price of a 64GB game card. That (to me at least) shows Nintendo is interested in working to improve costs, if not performance, in that component, as well. And best I can figure, game cards perform at a faster read speed than SD cards, probably benefitting from being read-only in some fashion.

EDIT: Oops, Macronix, not Micron.

 

[Dakhil]

It doesn't seem like a jump forward to new, more expensive storage technology is in the cards for any part of the Switch hardware proposition.

There's a report and a rumour about Nintendo's the first customer for Macronix's 48-layer 3D NAND memory, which Nintendo's sampling, and is speculated to be used for Game Cards with a capacity of 64 GB or larger. Of course, no one knows if the report and the rumour are going to materialise.

~

Anyway, I think 256 GB of capacity is the bare minimum if Nintendo mandates that all games (digital or physical) be installed in the internal flash storage, considering games have only increased, not decreased, in size, and if Nintendo plans on having a feature similar to Quick Resume, which does take up a good chunk of the internal flash storage's capacity.

 

[LiC]

There was a suggestion that Nintendo and Micron were working on 3D NAND to increase capacity for cheaper than the current price of a 64GB game card. That (to me at least) shows Nintendo is interested in working to improve costs, if not performance, in that component, as well. And best I can figure, game cards perform at a faster read speed than SD cards, probably benefitting from being read-only in some fashion.

EDIT: Oops, Macronix, not Micron.

Physical copies of Switch games are known to have noticeably longer load times than digital (whether internal or SD, though internal is fastest).

 

[Terrell]

Physical copies of Switch games are known to have noticeably longer load times than digital (whether internal or SD, though internal is fastest).

Looks like it depends on which SD card you use, as is usual. Cheap out and you’ll get worse performance.

Anyways, if Nintendo is willing to make improvements with game cards, the whole discussion changes.

And on further reading, the use of 3D NAND for game cards is being disputed because game cards currently use Macronix’s XtraROM tech, and NAND is a volatile flash memory solution. So it seems if Nintendo is going to do anything to improve read speeds, it will have to involve an evolution in Macronix’s ROM technology or another supplier who can. Whether the speed bottleneck is the ROM itself or the game card reader, it will need to be resolved. SD cards aren’t fast enough for what’s to come, so I guess the ROM solution for game cards needs fixing, as well, unless we see mandatory installs (Eww).

 

[ShadowFox08]

I was mainly pointing out that 12GB for a switch is probably fine if the OS is super lean like the current switch.

Steam Deck(Valve) was thinking lower but went with higher for longevity reasons. Switch probably doesn’t really need 16GB of RAM (though devs would love it), while SD would need more for its package relatively because of what it is.

Not really about the competition stuff which is a different matter.


I think 12GB is a better than fine amount for this device. 8GB is fine as well. 16GB is probably overkill for such a device honestly.

Switch 2/Dane definitely doesn't need 16GB Ram. Devs will be more than happy with 12, with 8 being the bare minimum.

Curious question, so this Dane is supposed to have DLSS and much better CPU. Does it compare to Steam Deck on a similar level in terms of overall power or is it stronger or perhaps have a better advantage due to DLSS?

Dane should have DLSS and the A78 CPUs will help it bridge that CPU power gap with Xbox series s to make it similar to switch vs PS4 base (a factor of 3-3.5x).

Someone correct me if I'm wrong but Steam Deck CPU uses pretty much the same CPU as the x box series and PS5, but only half as many cores (4 vs 8) and between 2.4-3.6GHz speed (let's just 3GHz average). That's actually pretty amazing to fit in a small form factor with 15 watts if true. And X series/PS5 CPU is supposedly equal to A78s in single thread performance per GHz. if we take out 1 core for running the OS, I think if we put really conservative estimates of 7 A78 cores running at 1Ghz on Dane/Switch 2 Vs Steam Deck's 3 cres running at 3Ghz each, Steam Deck will edge out in performance by 30-50% for single thread performance. Switch 2 could go higher to 1.2-1.5 and match without a hitch. I don't know about multi thread performance though, which I heard AMD really does better at.

Other than that, Dane GPU should match 1.6 TFLOPs RDNA2 GPU on Steam Deck, or get close to it without too much of a hitch, in docked mode. And then we aren't even counting GPU efficiency per FLOP from Dane's newer hardware as well as DLSS, which should put it well above.

At the very least way lighter and thinner, yeah, and battery life on average should be higher.

Who knows about battery life... I hope it's at least as much as V1 switch (3hrs for demanding games). I don't think m like we know Steam Deck's true battery life for games that really push it, just the range if you to 8 hrs... Which is really vague. Important to note that SD runs in 1 and 1.6 TFLOPs profiles..

But to be fair Steam Deck is running at 1-1.6 TFLOPs in handheld (and with a big battery) vs Dane likely running it at 1 TFLOP or lower on handheld.

 

[ILikeFeet]

Someone correct me if I'm wrong but Steam Deck CPU uses pretty much the same CPU as the x box series and PS5, but only half as many cores (4 vs 8) and between 2.4-3.6GHz speed (let's just 3GHz average). That's actually pretty amazing to fit in a small form factor with 15 watts if true. And X series/PS5 CPU is supposedly equal to A78s in single thread performance per GHz. I think if we put really conservative estimates of 8 A78 cores on Dane/Switch 2 Vs Steam Deck's 4 cores running at 3Ghz, Steam Deck will edge out in performance by 50% for single thread performance. I don't know about multi thread performance though.

I suspect the Steam Deck is gonna be more variable than people realize. they say 15W for the APU, but that's shared between the CPU and GPU. unless you allow for max clocks on both (which is gonna destroy the battery), one is gonna suffer for the other.

 

[ShadowFox08]

I suspect the Steam Deck is gonna be more variable than people realize. they say 15W for the APU, but that's shared between the CPU and GPU. unless you allow for max clocks on both (which is gonna destroy the battery), one is gonna suffer for the other.

Yeah we'll see the max clockspeeds at 15 watts for them. it would be impressive if they can run at 1.6 TFLOPs GPU and 3Ghz CPU at 15 watts. It's interesting they went with 4 CPU cores instead of 8 (with lower clocks with 8). Maybe a spacing issue?

In terms of performance efficiency, despite being at 8nm and a different architecture, the ampere matches really quite well with the RDNA2 at 7nm.

Whatever GPU and CPU performance combo SD realistically can achieve at 7nm without throttling at 15 watts... It will then give us/narrow us down a better idea of what we can expect from Dane at 8nm. There's still a lot we don't know about Dane like how RT (especially how many cores) and DLSS to some degree, will affect power draw and physical space, as well as the switch hardware components outside of the SOC.

 

[Dakhil]

And on further reading, the use of 3D NAND for game cards is being disputed because game cards currently use Macronix’s XtraROM tech, and NAND is a volatile flash memory solution.

One potential problem is that Macronix's Gaming Machine XtraROM, which seems to be used for the Game Cards, seem to have a max capacity of 32 GB, which happens to be the highest capacity Game Card for the Nintendo Switch so far. So I don't think Macronix's Gaming Machine XtraROM can be used if Nintendo wants Game Cards with capacities greater than 32 GB.

But Anandtech mentioned that Macronix designs and manufactures memory with reliability and durability in mind since Macronix's memory's generally used for specialised devices (e.g. defibrillators, drones, video game cartridges, watches, etc.). But speaking about NAND, Macronix's SLC NAND seems to have been designed with reliability and durability with mind, especially with Macronix's NAND flash memory not only passing all the stringent AEC-Q100 reliability standards, but also be the industry's first fully compliant AEC-Q100 flash memory for automotive applications. And Macronix has been researching on how to make NAND flash memory last longer.

So I imagine Macronix's NAND flash memory is probably more reliable and durable than NAND flash memory from other manufacturers. Of course, I don't know if the report and the rumour will ultimately materialise.

 

[Terrell]

One potential problem is that Macronix's Gaming Machine XtraROM, which seems to be used for the Game Cards, seem to have a max capacity of 32 GB, which happens to be the highest capacity Game Card for the Nintendo Switch so far. So I don't think Macronix's Gaming Machine XtraROM can be used if Nintendo wants Game Cards with capacities greater than 32 GB.

But Anandtech mentioned that Macronix designs and manufactures memory with reliability and durability in mind since Macronix's memory's generally used for specialised devices (e.g. defibrillators, drones, video game cartridges, watches, etc.). But speaking about NAND, Macronix's SLC NAND seems to have been designed with reliability and durability with mind, especially with Macronix's NAND flash memory not only passing all the stringent AEC-Q100 reliability standards, but also be the industry's first fully compliant AEC-Q100 flash memory for automotive applications. And Macronix has been researching on how to make NAND flash memory last longer.

So I imagine Macronix's NAND flash memory is probably more reliable and durable than NAND flash memory from other manufacturers. Of course, I don't know if the report and the rumour will ultimately materialise.

The primary thing here is that NAND is still flash memory, meaning the data on it is meant to be transient and will never compete with ROM’s ability to retain data for longer periods of time and without risk of write functionality being opened up. Read-only tech does have advantages that Nintendo may be unwilling to give up.

 

[Dakhil]

The primary thing here is that NAND is still flash memory, meaning the data on it is meant to be transient and will never compete with ROM’s ability to retain data for longer periods of time and without risk of write functionality being opened up. Read-only tech does have advantages that Nintendo may be unwilling to give up.

Interestingly enough, Macronix mentioned that the MX30LFxxS series of Macronix's SLC NAND flash memory features an advanced security feature called Permanently Block Lock (PBL), which could set a block or a sector of the NAND flash memory to be permanently read only.

I know that Nintendo probably won't use Macronix's SLC NAND flash memory. But considering Nintendo has been using Macronix's memory since the SNES, and assuming the report and the rumour about Nintendo planning on using a 48-layer 3D NAND memory for Game Cards with capacities larger than 32 GB still materialises, I imagine Nintendo could work with Macronix on bringing a feature similar to PBL to the 48-layer 3D NAND memory used for the Game Cards larger than 32 GB in capacity.

 

[LiC]

Looks like it depends on which SD card you use, as is usual. Cheap out and you’ll get worse performance.

Anyways, if Nintendo is willing to make improvements with game cards, the whole discussion changes.

And on further reading, the use of 3D NAND for game cards is being disputed because game cards currently use Macronix’s XtraROM tech, and NAND is a volatile flash memory solution. So it seems if Nintendo is going to do anything to improve read speeds, it will have to involve an evolution in Macronix’s ROM technology or another supplier who can. Whether the speed bottleneck is the ROM itself or the game card reader, it will need to be resolved. SD cards aren’t fast enough for what’s to come, so I guess the ROM solution for game cards needs fixing, as well, unless we see mandatory installs (Eww).

I'm not aware of a comparison that's been done where cartridge beats any SD card for loading times. The overall picture is that the speed of the SD card doesn't matter, since the Switch doesn't actually read faster from even the most high end cards. Internal > SD > cartridge is how it is for all cases that have been tested.

 

[Dakhil]

I'm not aware of a comparison that's been done where cartridge beats any SD card for loading times. The overall picture is that the speed of the SD card doesn't matter, since the Switch doesn't actually read faster from even the most high end cards. Internal > SD > cartridge is how it is for all cases that have been tested.

Thraktor mentioned on ResetEra that the eMMC 5.1 chip runs at ~100 MB/s and the Game Cards run at 25 MB/s (the Game Cards have a 8-bit bus width running at a frequency of 25 MHz).

 

[Terrell]

I'm not aware of a comparison that's been done where cartridge beats any SD card for loading times. The overall picture is that the speed of the SD card doesn't matter, since the Switch doesn't actually read faster from even the most high end cards. Internal > SD > cartridge is how it is for all cases that have been tested.

Sandisk did a test back in 2017 where their Extreme Pro SD card loaded data faster than other UHS-I SD cards in their lineup. So yes, the card you use makes a difference for speed. But point taken that game cards are slowest, my mistake.



This also indicates again that it’s something that will need to be resolved if we (or Nintendo) don’t want to see mandatory installs moving forward.

 

[ILikeFeet]

I'm not aware of a comparison that's been done where cartridge beats any SD card for loading times. The overall picture is that the speed of the SD card doesn't matter, since the Switch doesn't actually read faster from even the most high end cards. Internal > SD > cartridge is how it is for all cases that have been tested.

hell, I don't think there's a significant difference in those loadings unlike the other systems. Nintendo should just let the SD card and internal storage go balls out

 

[Terrell]

So there’s a slim possibility that Nintendo has OLED panel supply figured out…?

Apple's 2023 iPhones Could See A Move Towards BOE Displays

While all of the talk about the 2023 iPhone has so far centered around the expected switch to a custom 5G modem and away from one built by Qualcomm, there might be further change in terms of the display, too.

www.redmondpie.com

The link suggests BOE will be able to manufacture 10 times the amount of OLED panels Samsung is currently producing to satisfy future iPhone production, which should help with supply for Nintendo in one way or another.

 

[ILikeFeet]

So there’s a slim possibility that Nintendo has OLED panel supply figured out…?

Apple's 2023 iPhones Could See A Move Towards BOE Displays

While all of the talk about the 2023 iPhone has so far centered around the expected switch to a custom 5G modem and away from one built by Qualcomm, there might be further change in terms of the display, too.

www.redmondpie.com

The link suggests BOE will be able to manufacture 10 times the amount of OLED panels Samsung is currently producing to satisfy future iPhone production, which should help with supply for Nintendo in one way or another.

I never thought supply was an issue. there are many OLED suppliers out there that Nintendo can hit up in a pinch

 

[Terrell]

I never thought supply was an issue. there are many OLED suppliers out there that Nintendo can hit up in a pinch

Not as many as you’d think, especially not in terms of the quality we come to associate with OLED, anyways. There’s a reason that Samsung is considered the leader in OLED, despite how few substrates they can produce right now.

 

[Look over there]

Someone correct me if I'm wrong but Steam Deck CPU uses pretty much the same CPU as the x box series and PS5, but only half as many cores (4 vs 8) and between 2.4-3.6GHz speed (let's just 3GHz average). That's actually pretty amazing to fit in a small form factor with 15 watts if true. And X series/PS5 CPU is supposedly equal to A78s in single thread performance per GHz. if we take out 1 core for running the OS, I think if we put really conservative estimates of 7 A78 cores running at 1Ghz on Dane/Switch 2 Vs Steam Deck's 3 cres running at 3Ghz each, Steam Deck will edge out in performance by 30-50% for single thread performance. Switch 2 could go higher to 1.2-1.5 and match without a hitch. I don't know about multi thread performance though, which I heard AMD really does better at.

The presumed A78s in single thread per ghz are pretty close to/potentially even with regular Zen 2, going by those Geekbench 5 scores. They're ahead of the chip presumed to be the X series/PS5 CPU, because they're handicapped Zen 2s (having a quarter of the L3 cache) like the other Zen 2 APUs. Plus they're also using GDDR instead of regular DDR ram, which probably hinders them in CPU-bound/memory latency sensitive tasks. The Steam Deck will probably use the same handicapped Zen 2, based on AMD's tendencies. Although, the Steam Deck will be using LPDDR instead of GDDR.
Multi-threading's.... tough to say. If you want a generic, overall rule of thumb, I guess that you can treat the 4C/8T CPU as... close but not quite a peer to a 6C/6T CPU in multi threaded tasks? (based on 6C/12T seen as approximately a peer to a 8C/8T*) But that combines both gaming and non-gaming tasks. For gaming performance specifically, it varies from game to game. Some games don't even need a bunch of threads in the first place and end up preferring having SMT disabled in order to push higher clock rates (because enabling SMT eats some power on its own). For other games, enabling SMT is more for helping with dips/1% lows than it is for pushing average FPS up. It's tricky, and part of that is just due to how much variance there is in PC configurations, so there isn't like a lot of optimization for set hardware configs like in console space.

*based on my recollection of reactions to Intel i7 9700 being 8C/8T versus the i7 8700 being 6C/12T

Yeah we'll see the max clockspeeds at 15 watts for them. it would be impressive if they can run at 1.6 TFLOPs GPU and 3Ghz CPU at 15 watts. It's interesting they went with 4 CPU cores instead of 8 (with lower clocks with 8). Maybe a spacing issue?

Probably a yields concern moreso than it is die space. I think that for these Zen 2 laptop chips, AMD designs one die each. For example, Renoir (the 4000 series Zen 2 laptop chips) is one die, covering from 4 to 8 cores. IIRC, the way Zen 2 is designed, you get pairs of CCX (Core Complexes), with each CCX containing 4 cores. So a 8 core Zen 2 requires both CCX to be completely functional. A 6 core Zen 2 requires both CCX to have at least 3 cores each working. A 4 core Zen 2 typically requires just one of either CCX to be fully working (with the exception of the Ryzen 3100, which asks for both CCX to have 2 workings cores each).
It's probably the case that Van Gogh follows the typical Zen 2 APU design, so the die itself may still physically have two CCX. Then a Steam Deck would need just one fully functioning CCX. Why not 6 cores then? Maybe it's a power concern there. Or maybe Valve has a preference for cutting out the extra latency for communication between two CCXs.

 

[ILikeFeet]

Not as many as you’d think, especially not in terms of the quality we come to associate with OLED, anyways. There’s a reason that Samsung is considered the leader in OLED, despite how few substrates they can produce right now.

Nintendo has shown in the past what a goes for a passing panel. there's a lot of companies that can fill in some of Nintendo's orders. people won't like it when the screen lottery comes back, but I predict it definitely will

 

[Helix]

Honestly, at this rate, anything past March 2022 is possible now XD
Most likely times are Summer or Fall 2022 IMHO as 2023 would be a bit late time-wise relative to Orin and the Devkits.

hmmm, idk. I feel like Summer 2022 is an aggressive estimated timeframe. it's possible that Nintendo may have locked in chip supplies before and with restrictions slowly easing up, manufacturing can resume at least some decent capacity but I guess you have a point in that many of the devkits have been out there for quite a while, I'm still leaning on early 2023 tbh

 

[Alovon11]

hmmm, idk. I feel like Summer 2022 is an aggressive estimated timeframe. it's possible that Nintendo may have locked in chip supplies before and with restrictions slowly easing up, manufacturing can resume at least some decent capacity but I guess you have a point in that many of the devkits have been out there for quite a while, I'm still leaning on early 2023 tbh

This thing is being released with BOTW2, BOTW2 is still being released in 2022 as of latest info from Nintendo

 

[Dakhil]

Honestly, at this rate, anything past March 2022 is possible now XD
Most likely times are Summer or Fall 2022 IMHO as 2023 would be a bit late time-wise relative to Orin and the Devkits.

I think it depends on when Nintendo starts mass production, which would probably be reported by well known news organizations (e.g. Bloomberg, Nikkei, Wall Street Journal, etc.). And I think summer 2022 is probably unlikely if Nintendo decides to start mass manufacturing on March 2022, assuming mass production takes ~6 months.

So there’s a slim possibility that Nintendo has OLED panel supply figured out…?

Apple's 2023 iPhones Could See A Move Towards BOE Displays

While all of the talk about the 2023 iPhone has so far centered around the expected switch to a custom 5G modem and away from one built by Qualcomm, there might be further change in terms of the display, too.

www.redmondpie.com

The link suggests BOE will be able to manufacture 10 times the amount of OLED panels Samsung is currently producing to satisfy future iPhone production, which should help with supply for Nintendo in one way or another.

Depends on if BOE can consistently meet Apple's strict quality control standards, considering BOE's failure to do so is the reason why Apple used primarily Samsung's and LG's OLED displays for the iPhone 12 and iPhone 13 models for 2020 and a good chunk of 2021. And so far, BOE has only received conditional approval from Apple to produce ~15 million OLED displays, which is ~10% of the OLED displays used for the iPhone 12 and the iPhone 13 models.

Not as many as you’d think, especially not in terms of the quality we come to associate with OLED, anyways.

Nintendo's definitely not as stringent as Apple and probably other smartphone manufacturers when display quality is concerned, going by how the Nintendo 3DS and the Nintendo Switch (2019) have suffered from screen lottery.

 

[Helix]

This thing is being released with BOTW2, BOTW2 is still being released in 2022 as of latest info from Nintendo

we don't know if that is set in stone though, they said they were aiming for that and besides if they were planning to launch by Summer 2022 then we would have likely heard them ramp production now which they haven't. Fall 2022 or Early 2023 is where my headspace is at.

 

[TonyBaduy]

Summer 2022 is when I think it'll be announced, most likely before E3 so devs can show their games on it.

 

[Alovon11]

we don't know if that is set in stone though, they said they were aiming for that and besides if they were planning to launch by Summer 2022 then we would have likely heard them ramp production now which they haven't. Fall 2022 or Early 2023 is where my headspace is at.

Well my stance is if NVIDIA can pretty much hide Orin going into mass production from us, they can hide Dane (Orin comes out literally in 2 months)

So I feel that point is a bit moot, and anyway if they are intending to launch it with BOTW2 and BOTW2 is still slated for 2022, there comes a point where it is literally too late to delay and that point is likely soon, or already has passed.

You can't just push a system intended for 2022 out into 2023 if it's already 2022/nearly 2022 for a system like this.

Valve can do it as they are the only ones selling the Steam Deck.

Nintendo? They have countless deals and license agreements with Retailers, Advertising Firms, Ad-spots in other people's productions.etc likely in the works or already agreed on for occurring in 2022 around the season they intend to release it (which that should be clear at this point as BOTW2 is targeting 2022)

So the only scenario where BOTW2 does not come out with Switch 2/Orin/Next is where BOTW2 is delayed into 2023, not Dane with it.

It's just too late to push things back with how much work goes into marketing and retailer agreements for Dane itself.

 

[Terrell]

Nintendo's definitely not as stringent as Apple and probably other smartphone manufacturers when display quality is concerned, going by how the Nintendo 3DS and the Nintendo Switch (2019) have suffered from screen lottery.

OLED manufacturing differences aren’t just about whether displays are too warm or too cool (the main focus of the “display lottery”). It’s also more of a “won’t last long enough to live through the manufacturer warranty” problem. This is why the bulk of consumer electronics with OLED have panels made by a handful of manufacturers, and some of them don’t compete in every size because they want to focus quality control on a specific market segment (some only make TV panels, some only smartwatch/IoT panels) and in the “medium-sized” panel business, the business that consists of smartphones and game consoles and thus has the highest demand for durability and reliability outside of the TV market, there’s even less competition because of stringent quality demands from the major players there (Including Samsung and LG themselves, who sell their own devices in that market segment).

In any case, Apple potentially leaning on BOE for more OLED panels and less on Samsung and LG opens up Nintendo to take advantage of any vacated production capacity for themselves.

 

[Pokemaniac]

Looks like it depends on which SD card you use, as is usual. Cheap out and you’ll get worse performance.

Anyways, if Nintendo is willing to make improvements with game cards, the whole discussion changes.

And on further reading, the use of 3D NAND for game cards is being disputed because game cards currently use Macronix’s XtraROM tech, and NAND is a volatile flash memory solution. So it seems if Nintendo is going to do anything to improve read speeds, it will have to involve an evolution in Macronix’s ROM technology or another supplier who can. Whether the speed bottleneck is the ROM itself or the game card reader, it will need to be resolved. SD cards aren’t fast enough for what’s to come, so I guess the ROM solution for game cards needs fixing, as well, unless we see mandatory installs (Eww).

IIRC XtraROM isn't actually true ROM, and is actually modified NAND.

 

[Terrell]

IIRC XtraROM isn't actually true ROM, and is actually modified NAND.

It depends on the type, there’s 3 types. Nintendo is using ASIC XtraROM, which doesn’t sound like modified NAND, but there IS an NAND-ish variety of XtraROM.

 

[Mercury_Sagit]

Well my stance is if NVIDIA can pretty much hide Orin going into mass production from us, they can hide Dane (Orin comes out literally in 2 months)

So I feel that point is a bit moot, and anyway if they are intending to launch it with BOTW2 and BOTW2 is still slated for 2022, there comes a point where it is literally too late to delay and that point is likely soon, or already has passed.

You can't just push a system intended for 2022 out into 2023 if it's already 2022/nearly 2022 for a system like this.

Valve can do it as they are the only ones selling the Steam Deck.

Nintendo? They have countless deals and license agreements with Retailers, Advertising Firms, Ad-spots in other people's productions.etc likely in the works or already agreed on for occurring in 2022 around the season they intend to release it (which that should be clear at this point as BOTW2 is targeting 2022)

So the only scenario where BOTW2 does not come out with Switch 2/Orin/Next is where BOTW2 is delayed into 2023, not Dane with it.

It's just too late to push things back with how much work goes into marketing and retailer agreements for Dane itself.

One point regarding announcement-to-release window is the demand. That for Dane will be magnitudes higher than for Orin boards so I don't think the 2 months window for the latter is a good indicator for the former. Nevertheless, I'm still in the late 2022 release camp.

 

[brainchild]

games have been using RTXGI already, but the first game with infinite scrolling volumes is about to be released. ray traced probe grids are one of those things that could, theoretically, be accelerated by the minimal amount of RT cores in Dane

ICARUS - Prepare for Beta Weekend 7 | With NVIDIA RTXGI, Photo mode and Giveaways - Steam News

This weekend brings beautiful graphics and bountiful prizes.

store.steampowered.com

Not commenting on Dane, but I will say that the Infinite Scrolling Volumes technique is definitely going to be a part of the indirect lighting solution for my project. It's really clever and efficient.

EDIT:

It's honestly not all that conceptually different from how global illumination is handled in Breath of the Wild (probe sampling is directly tied to player position), though the actual implementation is very different and the quality is obviously much higher.

 

[Mr Swine]

Looks like it depends on which SD card you use, as is usual. Cheap out and you’ll get worse performance.

Anyways, if Nintendo is willing to make improvements with game cards, the whole discussion changes.

And on further reading, the use of 3D NAND for game cards is being disputed because game cards currently use Macronix’s XtraROM tech, and NAND is a volatile flash memory solution. So it seems if Nintendo is going to do anything to improve read speeds, it will have to involve an evolution in Macronix’s ROM technology or another supplier who can. Whether the speed bottleneck is the ROM itself or the game card reader, it will need to be resolved. SD cards aren’t fast enough for what’s to come, so I guess the ROM solution for game cards needs fixing, as well, unless we see mandatory installs (Eww).

Didn’t Nintendo have a job listing on someone that has worked with compression last year? I think Nintendo can go the Sony route with some advanced compression freeing up/ speeding up data transfers, bandwidth and storage.

 

[ILikeFeet]

Not commenting on Dane, but I will say that the Infinite Scrolling Volumes technique is definitely going to be a part of the indirect lighting solution for my project. It's really clever and efficient.

EDIT:

It's honestly not all that conceptually different from how global illumination is handled in Breath of the Wild (probe sampling is directly tied to player position), though the actual implementation is very different and the quality is obviously much higher.

BotW is one of the games I cited that could potentially be accelerated since it's already using a probe-based solution

 

[Terrell]

Didn’t Nintendo have a job listing on someone that has worked with compression last year? I think Nintendo can go the Sony route with some advanced compression freeing up/ speeding up data transfers, bandwidth and storage.

I think you’re talking about Oodle and its Kraken and new Texture tools. First, the good news:

On game consoles we're everywhere: Sony PS5, Microsoft Xbox Series X, Microsoft XboxOne, Sony PS4, Nintendo Switch, and tons of older hardware (Wii, Wii-U, Xbox 360, PS3, PSP, PS Vita, 3DS, etc).

Nintendo is likely fully aware of this solution, as is everyone else.

The bad news is the thing that maybe sets PS5 apart is that they developed a separate I/O controller that has Oodle hardware acceleration. I can’t say if any other hardware maker has done this, someone else may be able to chime in.

In addition, Sony and MS, with the move to SSDs, have eliminated the necessity for duplicated data in a software package for easier read access, which has already allowed for diminished software package sizes. With hardware acceleration, Sony is able to achieve both lower package sizes AND a speed boost with data loading.

Since this tech is no secret, whether or not this hardware acceleration is in Dane is going to be up to Nintendo and Nvidia to implement. But it does not completely resolve the speed issue with game cards, SD cards or even the internal eMMC, though it would absolutely be a big help.

 

[ILikeFeet]

I don't see the need for dedicated decompression hardware when Nvidia has work regarding decompression using the GPU