• 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| (Read the staff posts before commenting!)

Reader mode
ProngoLuvr said:
tbh i wouldn't expect a big publication to tease a console announcement as vaguely as that unless they're really uncertain about what exactly the presentation is about, in which case they might be better off waiting until they can get more info. console leaks from big publications have been pretty consistent for every model of the Switch and for Nintendo's past systems, but that hasn't really shown up for this next system. Bloomberg's reporting on a 4K Switch may have been for the successor, but their current stance on its release date is just "we dunno, maybe 2024?" i think it's too late at this point for a 2023 release. i'd love to be wrong though.
Click to expand...
I see what you mean, but that is what I rhink's happening, I'm pretty sure they're just... Unsure. No more sure than we can be anyway.
 
Would be funny (actually sad) if people don’t want to talk due to nvidia hack and presumably putting themselves and their company at stake if they do and are discovered for it.

Yadda yadda accused of looking through illegally acquired files.


It wasn’t just hardware configurations that were leaked, but it was a lot of software files too. Which can put more than just AMD, Apple, Alphabet (Google), Microsoft, etc at risk but game companies such as EA, Ubisoft, Capcom, Take Two, PlayStation, etc. would be at risk for getting anywhere close to this or talking about it.


We focus so much on the NVN2 and the T239 config that was illegally acquired, but that isn’t the only thing that was illegally acquired….
 
Look over there said:
Ooh, gonna ramble a bit.
(but to answer your question, this sort of thing is usually both sides? Reader needs to be able to handle X version of protocol, the card itself needs to be so and so capable)
See the charts of benchmarks for both the internal NVMe SSD and the SD Card at 1:29 or 1:30? Here's my interpretation of what's going on...

SEQ = Sequential; you're working with one file and/or everything you want to work with is lined up for the best possible conditions.
Not surprisingly, the NVMe drive is crushing here.

Q should be Queue Depth, or the number of requests working simultaneously/in parallel. The NVMe drive shows some difference between Q1 and Q8, but doesn't seem to be hitting its theoretical max. Could be a CPU limit (I'm guessing power limit, as a desktop Zen 2 shouldn't have the same issue). The SD card on the other hand shows practically no difference between Q1 and Q8. It's only working one thing at a time?

RND = Random; I work with one file, finish, then jump to another file located elsewhere, and so forth.
Notice that at Queue Depth 1, the NVMe drive is only a few times faster than the SD card. Here's a question: how much of that difference is coming from the 'working with a given file' part and how much is coming from the 'I then jump around to the next file' part? :unsure:
I'm guessing more from the former than the latter, since at the end of the day, they're both using NAND flash. The medium itself can only be so responsive.
At Queue Depth 32, the NVMe drive pulls away again. So, jumping around randomly from a file to another one at a time isn't so hot. So the workaround is... why only work one at a time, when you can do a lot at once? As oldpuck mentioned before, cover up latency by employing parallelism. The SD card again seems to be stuck to just one thing at a time?


So to recap: When does an NVMe drive deliver on the hype (relative to something like an SD card)?
1. You are working with large, individual files. Here's something that the companies marketing these drives don't want to admit: this currently does not apply to the average consumer*. Granted, 'currently' does a heavy lifting here, because DirectStorage ought to be changing this in the future.
2. You can make lots of requests simultaneously/in parallel. Good idea! Also shouldn't be unique! SATA SSDs should also be to do this, as we know that in real life, there isn't much practical difference between NVMe and SATA for the average consumer. I assume, but don't know for sure, that UFS is also capable of this.

*corollary to this: the marketing push for PCIe gen 5 NVMe drives started this year. Guess what? Realistically, there's practically no meaningful improvement for the average consumer over existing gen 3 or 4 drives. Don't fall for the hype of ever increasing sequential speeds unless you know it'll do something for you.
Click to expand...

To add to this, it's probably worth going into how this matters (or doesn't) for games consoles, as the use-case is quite different than for PC (even for PC games).

First, though, it's worth clarifying exactly what those SSD performance measurements in the video are. There are four measurements being used:

SEQ1M Q8: Sequentially reading 1MB at a time with a queue depth of 8
SEQ1M Q1: Sequentially reading 1MB at a time with a queue depth of 1
RND4K Q32: Randomly reading 4KB at a time with a queue depth of 32
RND4K Q1: Randomly reading 4KB at a time with a queue depth of 1

As to why the 4KB and 1MB sizes are typically used for these tests, 4KB is the block size for consumer SSD controllers. That is, any typical SSD you're going to use in your PC is divided up into 4KB blocks, and when reading or writing data, the SSD is only going to be able to operate on 4KB at a minimum. Testing with 4KB blocks therefore represents a worst-case scenario for random data access. The 1MB size is simply large enough that you can use it as a best-case scenario for sequential data access.

The use of 4KB blocks is part of the reason why high performance SSDs require multiple GBs of RAM attached to the controller, to quickly look up what's in every block. A 1TB drive contains 256 million blocks of data, so you need a lot of RAM to keep all of that quickly accessible if you want to do well in those random read/write tests. An alternative is to drop the DRAM to hit a lower price, and use system RAM to cache part of the lookup table. This can allow for reasonable sequential read speeds but random reads and writes typically take a significant hit. The Kioxia BG4 SSD tested in the video is one of these DRAMless SSDs, so it's not too surprising to see pretty poor random performance.

Coming back to my original point, the design of the PS5's SSD is pretty instructive on how this all translates to games consoles. Anandtech has a good article on it here, but to summarise, part of the reason the PS5's SSD offers such high performance is that they designed it in a way that works for game consoles, but would be absolutely terrible as a PC SSD. Instead of either attaching a lot of DRAM to the SSD controller, or dropping the DRAM and using the system RAM as a cache for the lookup table, Sony took the third option: change the block size of the SSD. This isn't really an option on PC, but on a console Sony has control over the OS and developers will optimise around whatever hardware is available, so there's nothing stopping them.

Sony didn't just increase the block size by a factor of two or four, they increased it all the way to 128MB, a full 32,768 times the block size of a typical consumer SSD, and significantly larger even than the large 1MB chunks used in sequential SSD testing. Instead of 256 million blocks, the PS5 SSD only contains around 6,000 blocks of data, so Sony can fit the entire lookup table in a small pool of SRAM right on the controller.

This allows for very high sequential read speeds of 5.5GB/s, but likely pretty terrible random performance. In fact, with a 5.5GB/s peak and 128MB blocks, we can quite easily calculate the peak theoretical IOPS figure for the PS5 SSD: it's 41. No, I'm not missing a bunch of zeros or a letter K there, it's literally the peak read speed divided by the block size (keeping in mind they use different definitions of GB/MB, just to make things fun). When operating at full tilt, the PS5 SSD is reading about 41 full 128MB blocks of data a second, and the peak IOPS rate therefore can't be higher than 41 (as it has to read the full 128MB block for each request, even if it only needs 4KB of data from it). If you were running the test from the video on the PS5 SSD, the RND4K Q1 read result should come to about 168KB/s, or 0.17MB/s, which is about 100x lower than the SD card being measured.

Doesn't that mean that the PS5 SSD is bad and will perform worse than an SD card? On a PC maybe, but not in a games console. For the entire existence of the Playstation, both Sony's first party developers and third party developers have had to work around very slow seek times, literally waiting for a read head to line up to the data they need off some form of mechanical media, first optical discs, and then hard drives. Developers have decades of experience making sure their data is carefully laid out to ensure the most efficient sequential access, and console OS's have the ability to make guarantees about how that data is stored on disk in a way which Windows can't.

So when faced with the option of achieving 5.5GB/s with a traditional SSD approach with a 4KB block size and additional DRAM for the controller, or saving a bunch of money by using a much larger block size, forgoing the DRAM and just asking developers to keep doing what they're doing with carefully ordered data, they went with the latter. It's possible that developers may have had a bit of an easier time with, say a 1MB block size, but with a single PBR texture set coming to a few MBs, there's only so much to gain by going smaller with the block size. My guess is that Sony chose the 128MB size carefully based on feedback from developers.

To sum up, there are two things to learn here:

Firstly, if you're buying an SSD to use as the boot disk on your PC, make sure it has DRAM. PCIe 3.0 SSDs are pretty good value at the moment, and a good one will have a noticeable (if not enormous) benefit to real-world performance over a SATA SSD. PCIe 4.0 and 5.0 SSDs are overkill for general consumer use, but newer PCIe 4.0 drives can be more power efficient because they use newer controllers, so might be worthwhile if there's not much of a price difference or you're on a laptop.

Secondly, and more pertinently to this thread, random performance on an SSD doesn't matter that much in a games console. Games just don't need to read lots of 4KB chunks from random parts of the SSD all that often. Nintendo won't use a custom SSD like Sony, going for an off-the-shelf eMMC or UFS drive instead. Any option available to them (including the eMMC used in the original Switch units) will have far higher 4KB random read speeds than PS5's SSD. Ultimately it won't really matter, as developers have and will continue to manage data layouts to ensure they can make the most out of peak sequential read speeds.
 
ReddDreadtheLead said:
Would be funny (actually sad) if people don’t want to talk due to nvidia hack and presumably putting themselves and their company at stake if they do and are discovered for it.

Yadda yadda accused of looking through illegally acquired files.


It wasn’t just hardware configurations that were leaked, but it was a lot of software files too. Which can put more than just AMD, Apple, Alphabet (Google), Microsoft, etc at risk but game companies such as EA, Ubisoft, Capcom, Take Two, PlayStation, etc. would be at risk for getting anywhere close to this or talking about it.


We focus so much on the NVN2 and the T239 config that was illegally acquired, but that isn’t the only thing that was illegally acquired….
Click to expand...
Yeah, I assume a lot of personal employee data was found too. It honestly crosses the line into amoral territory.
 
D36 said:
Yeah, I assume a lot of personal employee data was found too. It honestly crosses the line into amoral territory.
Click to expand...
yeah, personnel records, including SSNs, were in the hack. It goes way past amoral. Blackmail artists trying to scam a company by putting its employees in danger. Gross.

At the point at which the contents of said hack were being actively discussed on Twitter, though, it seemed like willful ignorance to not discuss it here. All that said, almost everything discovered in said hack has been corroborated by public statements and code from Nvidia. But only one of those piece of information hit the mainstream gaming/tech press (the 8 core CPU cluster) - specifically because a lurker in this thread posted the find to Reddit. Everything else seems to have been too niche to get out there (save Nintendo Prime's video).
 
0
How do you guys feel about Nintendo adding a their own custom screen casting feature to Switch 2, that not only allows for the device to be played on any screen w/o a dock, but also offer synchronous dual screen gameplay in applicable instances like Wii U or DS on NSO?

I ask this bc I feel like the Switch platform's "gimmick" is versatility and providing no limits to the ways you can play and this could help build on that.
 
Kingpin722 said:
How do you guys feel about Nintendo adding a their own custom screen casting feature to Switch 2, that not only allows for the device to be played on any screen w/o a dock, but also offer synchronous dual screen gameplay in applicable instances like Wii U or DS on NSO?

I ask this bc I feel like the Switch platform's "gimmick" is versatility and providing no limits to the ways you can play and this could help build on that.
Click to expand...
I could see an streaming app for Windows and Mac that allowed one to play their Switch games on a computer without the use of emulation.
But knowing whether or not Nintendo's interested in that is another thing (you'd think they would be considering how this could be a way to stop some people from using emulators and pirating their games, but that's the same company that takes down ROMs being distributed online for games they don't even sell you anymore).
 
Kingpin722 said:
How do you guys feel about Nintendo adding a their own custom screen casting feature to Switch 2, that not only allows for the device to be played on any screen w/o a dock, but also offer synchronous dual screen gameplay in applicable instances like Wii U or DS on NSO?

I ask this bc I feel like the Switch platform's "gimmick" is versatility and providing no limits to the ways you can play and this could help build on that.
Click to expand...
I'm currently leaning towards that being the most likely big new feature for the reasons you mentioned as well as:
  • makes use of mobile for stuff like online communication potentially less tedious if players are likely to use mobile phones with their console anyway.
  • better justifies their investments in mobile games
  • saves money by allowing a second screen without having to actually pack it in with the console
  • helps sell remaining switch consoles if they let people use switches as screencast devices too
  • would give them an excuse to significantly upgrade their bluetooth/wifi which they need to do
  • logical endpoint of Labo stuff
  • would potentially allow for VR and/or AR
  • would be fairly easy to create a device that allows joycons to be used as backbone-like devices
 
Kingpin722 said:
How do you guys feel about Nintendo adding a their own custom screen casting feature to Switch 2, that not only allows for the device to be played on any screen w/o a dock, but also offer synchronous dual screen gameplay in applicable instances like Wii U or DS on NSO?

I ask this bc I feel like the Switch platform's "gimmick" is versatility and providing no limits to the ways you can play and this could help build on that.
Click to expand...
Screencasting comes up a lot. I don't know what the underlying solution was for the Wii U, but it was remarkably low latency. Assuming it was a custom protocol, dockless casting probably isn't in the cards, but I'd love to see something like this
 
TomNookYankees said:
  • makes use of mobile for stuff like online communication potentially less tedious if players are likely to use mobile phones with their console anyway.
Click to expand...
There are already several Nintendo published games (e.g. Dragon Quest Builders 2, Super Smash Bros Ultimate, etc.) that only allow voice chat via the Nintendo Switch Online app on smartphones.
TomNookYankees said:
  • would give them an excuse to significantly upgrade their bluetooth/wifi which they need to do
Click to expand...
Nintendo also needs to improve the placement of the Wi-Fi/Bluetooth antennae. Otherwise, I think the improvements that an upgraded Wi-Fi/Bluetooth chip brings are going to be undermined by the bad Wi-Fi/Bluetooth reception caused by the poor placement of the Wi-Fi/Bluetooth antennae.
 
oldpuck said:
Screencasting comes up a lot. I don't know what the underlying solution was for the Wii U, but it was remarkably low latency. Assuming it was a custom protocol, dockless casting probably isn't in the cards, but I'd love to see something like this
Click to expand...
The GamePad communicates with a Wii U console over a modified Wi-Fi protocol designed for low-latency transmission, establishing its connection with the console by using a variant of the WPS process, with proprietary transfer protocol and software co-developed with Broadcom.
Click to expand...
- from Wikipedia. Wireless screen casting's latency today doesn't seem low enough for something like gaming. Pretty sure AirPlay 2 still has a noticeable bit of a lag when casting an iPhone to a TV for example.

I think the most likely outcome of this would be a Chromecast-style device that plugs into the HDMI where Switch 2 could cast in "docked" mode to it. More portable than what we have now.
 
ezidro3 said:
I think the most likely outcome of this would be a Chromecast-style device that plugs into the HDMI where Switch 2 could cast in "docked" mode to it. More portable than what we have now.
Click to expand...
Perhaps a Lite model equipped with Drake could come with a Miracast (or a feature similar to Miracast) HDMI adapter included in the box?

And perhaps the max resolution for a Lite model equipped with Drake when using a Miracast (or a feature similar to Miracast) HDMI adapter is restricted to the max resolution of the display on the Lite model equipped with Drake (perhaps 720p 60 Hz as one example?)?
 
ItWasMeantToBe19 said:
I can't believe we're in June and people are still going on about the potential of it launching this year.

It's probably Fall 2024, sorry.
Click to expand...
Why apologize? No need to be rude.

We already know your stance, no need to be annoying about it.
 
0
Dakhil said:
Perhaps a Lite model equipped with Drake could come with a Miracast (or a feature similar to Miracast) HDMI adapter included in the box?

And perhaps the max resolution for a Lite model equipped with Drake when using a Miracast (or a feature similar to Miracast) HDMI adapter is restricted to the max resolution of the display on the Lite model equipped with Drake (perhaps 720p 60 Hz as one example?)?
Click to expand...
It defeats the purpose of the lite model.

The lite is supposed to be the cheap, cut down handheld only alternative. It won't have any features that can make it seem like the superior model, or even that the flagship model doesn't have.
 
ezidro3 said:
- from Wikipedia. Wireless screen casting's latency today doesn't seem low enough for something like gaming. Pretty sure AirPlay 2 still has a noticeable bit of a lag when casting an iPhone to a TV for example.

I think the most likely outcome of this would be a Chromecast-style device that plugs into the HDMI where Switch 2 could cast in "docked" mode to it. More portable than what we have now.
Click to expand...
Yeah current usual casting methods just aren’t there yet for gaming. They also weren’t made with gaming in mind per se so they probably never felt the need to as something like Airplay 2 works fine as is.
 
0
Hermii said:
It defeats the purpose of the lite model.

The lite is supposed to be the cheap, cut down handheld only alternative. It won't have any features that can make it seem like the superior model, or even that the flagship model doesn't have.
Click to expand...
Nintendo DS lite: i'm a joke to you?
 
Pokemaniac said:
When you look at what happened with NX, Nintendo actually managed to keep details about the thing mostly under wraps until only a few months before the unveiling trailer, and that was with a delay, even. Leaks are a fickle and unreliable source of information, and Nintendo's gotten pretty good at controlling them. Another factor worth considering is that it's likely that much of the third party lineup this thing will have in the launch window won't be true exclusives, with most of the games probably releasing on either Switch and/or PS and XBox as well, which will help to further limit the number of people that even need to touch this thing. The lack of third party leaks is perhaps somewhat notable, but just doesn't seem to be a strong determining factor for when this is releasing.

The main reasons I think 2023 is at least the target are that the hardware development timeline seems to line up reasonable well with that, from what we can guess and that Nintendo has been acting pretty erratic and extra mysterious this year. With Nintendo's second half still so up in the air, yet indicators pointing much more towards feast than famine, we're probably in for a bunch of announcements soon regardless of what's happening.
Click to expand...

The concept of the NX was largely leaked immediately after dev kits were sent out in late 2015. Third parties immediately went to the WSJ to tell them that the NX consisted of a home console and mobile unit, which they could infer based on the two modes on the NX devkits most likely.
 
kasparov77 said:
I could see an streaming app for Windows and Mac that allowed one to play their Switch games on a computer without the use of emulation.
But knowing whether or not Nintendo's interested in that is another thing (you'd think they would be considering how this could be a way to stop some people from using emulators and pirating their games, but that's the same company that takes down ROMs being distributed online for games they don't even sell you anymore).
Click to expand...
I mean is it really gonna stop some people & not just a very very tiny minority; who would then most likely go back to what they were doing because it doesn’t accomplish the same goals.
 
TomNookYankees said:
Did the original 1-2 switch game have people use their smartphones as controllers?
Click to expand...
no we used the Joy-Con for it severals minigames, 1-2 Switch is another of this tech demon kinda of games Nintendo make to showcase the console main apeal, like they did with Wii Sports and Nintendo Land
 
ItWasMeantToBe19 said:
The concept of the NX was largely leaked immediately after dev kits were sent out in late 2015. Third parties immediately went to the WSJ to tell them that the NX consisted of a home console and mobile unit, which they could infer based on the two modes on the NX devkits most likely.
Click to expand...
I said "mostly" for a reason, as that initial report was ultimately not very accurate and misleading. If you want to compare that to the current situation, look no further than the flurry of leaks in 2020 and 2021, in which Drake and Switch OLED information was hopelessly mixed together, and recent discoveries about the Switch OLED have only cast more doubt on which information was about each system.
 
Last edited:
Pokemaniac said:
I said "mostly" for a reason, as that initial report was ultimately not very accurate and misleading. If you want to compare that to the current situation, look no further than the flurry of leaks in 2020 and 2022, in which Drake and Switch OLED information was hopelessly mixed together, and recent discoveries about the Switch OLED have only cast more doubt on which information was about each system.
Click to expand...
What recent discoveries about the oled?
 
Hermii said:
What recent discoveries about the oled?
Click to expand...
That it's fully wired for 4k output, unlike the launch Switch. Some of the information from back then is unambiguously about Drake, but if there were some 4k Mariko devkits also floating around, then that could not only help explain the level of confusion, but also means that reports of a more powerful 4k Switch that'll be used like the New 3DS could easily have been about the Mariko system instead.
 
Pokemaniac said:
That it's fully wired for 4k output, unlike the launch Switch. Some of the information from back then is unambiguously about Drake, but if there were some 4k Mariko devkits also floating around, then that could not only help explain the level of confusion, but also means that reports of a more powerful 4k Switch that'll be used like the New 3DS could easily have been about the Mariko system instead.
Click to expand...
That's not recent, we knew that since launch. No dev would confuse even a max clocked Mariko for a 4k system. And Bloomberg specifically said new diss capable soc.
 
Hermii said:
The lite is supposed to be the cheap, cut down handheld only alternative. It won't have any features that can make it seem like the superior model, or even that the flagship model doesn't have.
Click to expand...
I don't know if Nintendo's original intention for the Nintendo Switch Lite was to have the Nintendo Switch Lite be exclusively a handheld console from the very beginning. There's a rumour from Nikkei about the Nintendo Switch Lite still being able to access TV mode. Perhaps Nintendo removed the Nintendo Switch Lite's ability to access TV mode via removing the PI3USB30532 chip present on the Nintendo Switch to simply ensure Nintendo can reach the $199.99 price point for the Nintendo Switch Lite and still earn profits, no matter how small? (I'm not sure, which is why I asked.)

And I've remembered Jon from Spawn Wave made a video last year talking about how some of his family members thought the Nintendo Switch Lite has the ability to connect to the dock to access TV mode since "Switch" is in the name. But his family members still didn't understand when Jon explained that the Nintendo Switch Lite can't connect to the dock to access TV mode.

I wonder if Nintendo received feedback about how some people in the general public were confused about how the Nintendo Switch Lite couldn't connect to a dock to access TV mode due to "Switch" being in the name.
(Personally, I think "Lite" is very self-explanatory. But then again, I'm not really part of the general public. So who knows?)
 
so i had a dream where the new console was shadow-dropped on Twitter, Switch Pro was the name and it retailed for $912.99 (lol)

launch titles were 4K versions of Lylat Wars & Mario 64 which somehow looked worse than the originals with Minecraft style graphics. Nintendoomed :cry:
😁
 
I think it would be smart, for a bunch of reasons, to ALLOW people to dock the (next) Switch Lite. But in the box? No way. It's all being externalised. It could even be an even more expensive dock unloading some of the chips and circuitry from the console to keep the console price down. Likely not going beyond handheld performance / resolutions because it's a naturally more thermally constrained device, and a bunch of other reasons.

I EXPECT the next Switch Lite to dock. It's in the name. I however expect that experience to be extremely, well. Lite...

As for further differentiation, it would likely put the Lite in a position where it's cheaper to buy on its own but the same price (despite fewer capabilities) than the flagship console, for a whole bunch of market and economic reasons.

250$ Lite + a 150$ conversion kit including Joy-Con 2, a Charging Grip, a USB C Cable, and a Switch Lite Deluxe Dock. And 150$ for all that is on the LOW END of expectations if they offer it, Joy-Con and docks aren't as cheap to produce as you'd think.

Uh, in short, there's reasons it CAN'T dock, but there's a lot of reasons it SHOULD in future, but also that this experience probably won't be optimal anyway.
 
Concernt said:
I think it would be smart, for a bunch of reasons, to ALLOW people to dock the (next) Switch Lite. But in the box? No way. It's all being externalised. It could even be an even more expensive dock unloading some of the chips and circuitry from the console to keep the console price down. Likely not going beyond handheld performance / resolutions because it's a naturally more thermally constrained device, and a bunch of other reasons.

I EXPECT the next Switch Lite to dock. It's in the name. I however expect that experience to be extremely, well. Lite...

As for further differentiation, it would likely put the Lite in a position where it's cheaper to buy on its own but the same price (despite fewer capabilities) than the flagship console, for a whole bunch of market and economic reasons.

250$ Lite + a 150$ conversion kit including Joy-Con 2, a Charging Grip, a USB C Cable, and a Switch Lite Deluxe Dock. And 150$ for all that is on the LOW END of expectations if they offer it, Joy-Con and docks aren't as cheap to produce as you'd think.

Uh, in short, there's reasons it CAN'T dock, but there's a lot of reasons it SHOULD in future, but also that this experience probably won't be optimal anyway.
Click to expand...
I don't think Nintendo would allow a mode designed for handheld to be blown up to potentially a huge screen in any circumstance. All the compromises would potentially become so much more apparent, and the casual observer who don't know what they're looking at could get a faulty impression of the console's capabilities. Especially not if they stick with a 720p panel.
 
Dakhil said:
(Personally, I think "Lite" is very self-explanatory. But then again, I'm not really part of the general public. So who knows?)
Click to expand...
Not so much for general public. Lite means first "lighter" or "smaller", a term already used from Nintendo for DS Lite which was the same console as DS but, in fact, lighter and smaller.
 
Hermii said:
That's not recent, we knew that since launch.
Click to expand...
We didn’t. We knew the chip in the dock supported 4k, but that was plausibly explained by pre-4k versions no longer being manufactured.

The discovery that the OLED model has wired up PCI lanes to support 4k is fairly recent. Unlike all previous discoveries with the OLED model this change only plausibly supports 4k and increases the complexity of the motherboard.

Unlike previous evidence there isn’t another plausible explanation. It seems highly likely that until late in it's design the OLED model was intended to support 4k output.


Hermii said:
No dev would confuse even a max clocked Mariko for a 4k system. And Bloomberg specifically said new diss capable soc.
Click to expand...
It didn’t. It said that the new device would achieve 4k with DLSS. The leap that it would use DLSS 2 and thus tensor cores was ours.

Nvidia produced a custom version of DLSS 1.0 (informally called DLSS 1.9) that ran on CUDA cores instead of tensor cores. It was only ever used in Control, IIRC, but that would run on the existing Switch, presuming it was ported to run on ARM. Mario Kart 4k would be possible with an upclocked Mariko and DLSS 1.0.

Nvidia also said that the device would have more memory. Nintendo did, in fact, update the core memory technology in the OLED model, and despite the fact that the OLED requires no custom development, shipped OLED devkits with increased memory.

We've never seriously discussed a DLSS 1.0 Switch, but such a device would absolutely square the circle of all the rumors. It matches both the Bloomberg report and other insiders talking about a 4k Pro as a minor upgrade that would be superceded by a true next gen successor. DLSS 1.0 required a custom model generated for each game for good results, it's likely that you would get some flagship titles that would get 4k enhancements, but it wouldn't be every game, and there would be no BC problems.

Nintendo decides to pursue a pro revision, designs it for 4k using DLSS technology, shipping software based devkits and early OLED devkits. At somepoint Nintendo decides to release the OLED model as an exactly capable device. Devs are thrown for a loop, but at that point, Nintendo could still theoretically open up performance via a software update, and DLSS/4k is still the company's future, so development continues on the original 4k titles.

At some point after the Bloomberg article, Nintendo pulls that strategy entirely, Zynga issues their denial, and it's all in on T239.

I'm not saying that's what happened, but it absolutely is possible
 
Hermii said:
I don't think Nintendo would allow a mode designed for handheld to be blown up to potentially a huge screen in any circumstance. All the compromises would potentially become so much more apparent, and the casual observer who don't know what they're looking at could get a faulty impression of the console's capabilities. Especially not if they stick with a 720p panel.
Click to expand...
They will almost CERTAINLY not stick with a 720p panel. The Series S and Switch are both clear examples of consumers not caring that much about resolution. Which SHOULD be the primary sacrifice for such a device.
 
oldpuck said:
We didn’t. We knew the chip in the dock supported 4k, but that was plausibly explained by pre-4k versions no longer being manufactured.

The discovery that the OLED model has wired up PCI lanes to support 4k is fairly recent. Unlike all previous discoveries with the OLED model this change only plausibly supports 4k and increases the complexity of the motherboard.

Unlike previous evidence there isn’t another plausible explanation. It seems highly likely that until late in it's design the OLED model was intended to support 4k output.



It didn’t. It said that the new device would achieve 4k with DLSS. The leap that it would use DLSS 2 and thus tensor cores was ours.

Nvidia produced a custom version of DLSS 1.0 (informally called DLSS 1.9) that ran on CUDA cores instead of tensor cores. It was only ever used in Control, IIRC, but that would run on the existing Switch, presuming it was ported to run on ARM. Mario Kart 4k would be possible with an upclocked Mariko and DLSS 1.0.

Nvidia also said that the device would have more memory. Nintendo did, in fact, update the core memory technology in the OLED model, and despite the fact that the OLED requires no custom development, shipped OLED devkits with increased memory.

We've never seriously discussed a DLSS 1.0 Switch, but such a device would absolutely square the circle of all the rumors. It matches both the Bloomberg report and other insiders talking about a 4k Pro as a minor upgrade that would be superceded by a true next gen successor. DLSS 1.0 required a custom model generated for each game for good results, it's likely that you would get some flagship titles that would get 4k enhancements, but it wouldn't be every game, and there would be no BC problems.

Nintendo decides to pursue a pro revision, designs it for 4k using DLSS technology, shipping software based devkits and early OLED devkits. At somepoint Nintendo decides to release the OLED model as an exactly capable device. Devs are thrown for a loop, but at that point, Nintendo could still theoretically open up performance via a software update, and DLSS/4k is still the company's future, so development continues on the original 4k titles.

At some point after the Bloomberg article, Nintendo pulls that strategy entirely, Zynga issues their denial, and it's all in on T239.

I'm not saying that's what happened, but it absolutely is possible
Click to expand...
I do like your theory. I don't necessarily believe it happened that way, but it's a good theory.

Another theory could be that the plan could have been at one point to put Drake in the swoled, for an iterative successor style device (per Iwata s IOS/ Android comments), and at some point the decision was made to rather do a more traditional console generation. Get ready some exclusive launch titles, change up the industrial design, new controller etc.
 
0
Thraktor said:
To add to this, it's probably worth going into how this matters (or doesn't) for games consoles, as the use-case is quite different than for PC (even for PC games).

First, though, it's worth clarifying exactly what those SSD performance measurements in the video are. There are four measurements being used:

SEQ1M Q8: Sequentially reading 1MB at a time with a queue depth of 8
SEQ1M Q1: Sequentially reading 1MB at a time with a queue depth of 1
RND4K Q32: Randomly reading 4KB at a time with a queue depth of 32
RND4K Q1: Randomly reading 4KB at a time with a queue depth of 1

As to why the 4KB and 1MB sizes are typically used for these tests, 4KB is the block size for consumer SSD controllers. That is, any typical SSD you're going to use in your PC is divided up into 4KB blocks, and when reading or writing data, the SSD is only going to be able to operate on 4KB at a minimum. Testing with 4KB blocks therefore represents a worst-case scenario for random data access. The 1MB size is simply large enough that you can use it as a best-case scenario for sequential data access.

The use of 4KB blocks is part of the reason why high performance SSDs require multiple GBs of RAM attached to the controller, to quickly look up what's in every block. A 1TB drive contains 256 million blocks of data, so you need a lot of RAM to keep all of that quickly accessible if you want to do well in those random read/write tests. An alternative is to drop the DRAM to hit a lower price, and use system RAM to cache part of the lookup table. This can allow for reasonable sequential read speeds but random reads and writes typically take a significant hit. The Kioxia BG4 SSD tested in the video is one of these DRAMless SSDs, so it's not too surprising to see pretty poor random performance.

Coming back to my original point, the design of the PS5's SSD is pretty instructive on how this all translates to games consoles. Anandtech has a good article on it here, but to summarise, part of the reason the PS5's SSD offers such high performance is that they designed it in a way that works for game consoles, but would be absolutely terrible as a PC SSD. Instead of either attaching a lot of DRAM to the SSD controller, or dropping the DRAM and using the system RAM as a cache for the lookup table, Sony took the third option: change the block size of the SSD. This isn't really an option on PC, but on a console Sony has control over the OS and developers will optimise around whatever hardware is available, so there's nothing stopping them.

Sony didn't just increase the block size by a factor of two or four, they increased it all the way to 128MB, a full 32,768 times the block size of a typical consumer SSD, and significantly larger even than the large 1MB chunks used in sequential SSD testing. Instead of 256 million blocks, the PS5 SSD only contains around 6,000 blocks of data, so Sony can fit the entire lookup table in a small pool of SRAM right on the controller.

This allows for very high sequential read speeds of 5.5GB/s, but likely pretty terrible random performance. In fact, with a 5.5GB/s peak and 128MB blocks, we can quite easily calculate the peak theoretical IOPS figure for the PS5 SSD: it's 41. No, I'm not missing a bunch of zeros or a letter K there, it's literally the peak read speed divided by the block size (keeping in mind they use different definitions of GB/MB, just to make things fun). When operating at full tilt, the PS5 SSD is reading about 41 full 128MB blocks of data a second, and the peak IOPS rate therefore can't be higher than 41 (as it has to read the full 128MB block for each request, even if it only needs 4KB of data from it). If you were running the test from the video on the PS5 SSD, the RND4K Q1 read result should come to about 168KB/s, or 0.17MB/s, which is about 100x lower than the SD card being measured.

Doesn't that mean that the PS5 SSD is bad and will perform worse than an SD card? On a PC maybe, but not in a games console. For the entire existence of the Playstation, both Sony's first party developers and third party developers have had to work around very slow seek times, literally waiting for a read head to line up to the data they need off some form of mechanical media, first optical discs, and then hard drives. Developers have decades of experience making sure their data is carefully laid out to ensure the most efficient sequential access, and console OS's have the ability to make guarantees about how that data is stored on disk in a way which Windows can't.

So when faced with the option of achieving 5.5GB/s with a traditional SSD approach with a 4KB block size and additional DRAM for the controller, or saving a bunch of money by using a much larger block size, forgoing the DRAM and just asking developers to keep doing what they're doing with carefully ordered data, they went with the latter. It's possible that developers may have had a bit of an easier time with, say a 1MB block size, but with a single PBR texture set coming to a few MBs, there's only so much to gain by going smaller with the block size. My guess is that Sony chose the 128MB size carefully based on feedback from developers.

To sum up, there are two things to learn here:

Firstly, if you're buying an SSD to use as the boot disk on your PC, make sure it has DRAM. PCIe 3.0 SSDs are pretty good value at the moment, and a good one will have a noticeable (if not enormous) benefit to real-world performance over a SATA SSD. PCIe 4.0 and 5.0 SSDs are overkill for general consumer use, but newer PCIe 4.0 drives can be more power efficient because they use newer controllers, so might be worthwhile if there's not much of a price difference or you're on a laptop.

Secondly, and more pertinently to this thread, random performance on an SSD doesn't matter that much in a games console. Games just don't need to read lots of 4KB chunks from random parts of the SSD all that often. Nintendo won't use a custom SSD like Sony, going for an off-the-shelf eMMC or UFS drive instead. Any option available to them (including the eMMC used in the original Switch units) will have far higher 4KB random read speeds than PS5's SSD. Ultimately it won't really matter, as developers have and will continue to manage data layouts to ensure they can make the most out of peak sequential read speeds.
Click to expand...
What a lovely and informative post, thank you. I did not know that about the PS5, curious if Nintendo would do any modifications for a switch succ to be very efficient about it. Of course, it would be more bespoke and not so much OTS, but still in this case.

like say, 4MB?
 
Pokemaniac said:
When you look at what happened with NX, Nintendo actually managed to keep details about the thing mostly under wraps until only a few months before the unveiling trailer, and that was with a delay, even. Leaks are a fickle and unreliable source of information, and Nintendo's gotten pretty good at controlling them. Another factor worth considering is that it's likely that much of the third party lineup this thing will have in the launch window won't be true exclusives, with most of the games probably releasing on either Switch and/or PS and XBox as well, which will help to further limit the number of people that even need to touch this thing. The lack of third party leaks is perhaps somewhat notable, but just doesn't seem to be a strong determining factor for when this is releasing.

The main reasons I think 2023 is at least the target are that the hardware development timeline seems to line up reasonable well with that, from what we can guess and that Nintendo has been acting pretty erratic and extra mysterious this year. With Nintendo's second half still so up in the air, yet indicators pointing much more towards feast than famine, we're probably in for a bunch of announcements soon regardless of what's happening.
Click to expand...
maybe the pandemic massively affected the games Nintendo is developing for Switch sucessor
 
Hermii said:
That's not recent, we knew that since launch. No dev would confuse even a max clocked Mariko for a 4k system. And Bloomberg specifically said new diss capable soc.
Click to expand...
We knew the dock was upgraded from launch, but that could have been for other reasons. The discovery that the console motherboard was upgraded to support it is more recent and less ambiguous.

A dev wouldn't confuse a Mariko system for a Drake one, but they wouldn't need to for the confusion to arise. The big touted feature from the most sources was 4k, and an overclocked Mariko capable of outputting at 4k would still be described as a 4k Switch, even if realistically it wouldn't natively hit that full resolution very often. With two 4k Switches, and one of them having the 4k capability disabled in software before launch, it's not hard to see how information about the two of them could get mixed together by less technical press, especially since I'm not super convinced a lot of the insiders had even heard of Drake at the time, in hindsight. When everyone was coming out of the woodwork in 2021 talking about "Switch Pro" devkits, the system many of them were describing sounded a whole lot more like an overclocked Mariko system than a Drake system.

Basically what I'm saying is, a lot of the rumors make more sense if Switch OLED was actually a "Switch Pro" late enough in development for some devkits to get sent out, but the final Switch OLED being distinctly not that caused much of that info to become associated with the details from very early Drake development materials which were likely starting to go out in a fairly similar timeframe.
 
I wouldn’t be surprised if Nintendo pulls an Xbox One S/X situation:
  • The cheaper model (One S/Switch OLED) will be able to output 4K but only for media streaming like Netflix, YouTube, etc.
  • The higher-end model (One X/Drake) will be able to play games in 4K and have enhancement patches designed for it to increase visual fidelity/frame-rate.

It’s (to me) the most likely possibility since the wired-up PCI lanes seem like a waste when Nintendo probably knew ahead of time that OLED was never going to target 4K gaming.
 
ReddDreadtheLead said:
What a lovely and informative post, thank you. I did not know that about the PS5, curious if Nintendo would do any modifications for a switch succ to be very efficient about it. Of course, it would be more bespoke and not so much OTS, but still in this case.

like say, 4MB?
Click to expand...
I don't think they even need to. Xbox doesn't seem to have any bespoke modifications and compares favorably to PS5 in loading. Drake would be limied by other aspects before storage issues anyway
 
Giancarlo said:
maybe the pandemic massively affected the games Nintendo is developing for Switch sucessor
Click to expand...
If Nintendo's major releases were having development troubles, one would expect some evidence of spacing out what they have to make it last longer, but that's not happening. If anything, they seem to be dumping out what they did have saved up.
ezidro3 said:
I wouldn’t be surprised if Nintendo pulls an Xbox One S/X situation:
  • The cheaper model (One S/Switch OLED) will be able to output 4K but only for media streaming like Netflix, YouTube, etc.
  • The higher-end model (One X/Drake) will be able to play games in 4K and have enhancement patches designed for it to increase visual fidelity/frame-rate.

It’s (to me) the most likely possibility since the wired-up PCI lanes seem like a waste when Nintendo probably knew ahead of time that OLED was never going to target 4K gaming.
Click to expand...
It was probably never going to do a ton of native 4k, but we aren't expecting a ton of that out of Drake either. Most likely they did want to have games running at > 1080p (For practical purposes, anything higher than 1080p means 4k output when you're dealing with TVs) on Mariko, but weren't satisfied and cut the feature fairly late. There still seem to be some vestigial signs of this in the OS, with an unused setting to boost the USB bandwidth devoted to video output. If the feature was ever going to be enabled on Switch OLED, there'd be no reason to hold that back, and it would have been enabled at launch.
 
Concernt said:
I think it would be smart, for a bunch of reasons, to ALLOW people to dock the (next) Switch Lite. But in the box? No way. It's all being externalised. It could even be an even more expensive dock unloading some of the chips and circuitry from the console to keep the console price down. Likely not going beyond handheld performance / resolutions because it's a naturally more thermally constrained device, and a bunch of other reasons.

I EXPECT the next Switch Lite to dock. It's in the name. I however expect that experience to be extremely, well. Lite...

As for further differentiation, it would likely put the Lite in a position where it's cheaper to buy on its own but the same price (despite fewer capabilities) than the flagship console, for a whole bunch of market and economic reasons.

250$ Lite + a 150$ conversion kit including Joy-Con 2, a Charging Grip, a USB C Cable, and a Switch Lite Deluxe Dock. And 150$ for all that is on the LOW END of expectations if they offer it, Joy-Con and docks aren't as cheap to produce as you'd think.

Uh, in short, there's reasons it CAN'T dock, but there's a lot of reasons it SHOULD in future, but also that this experience probably won't be optimal anyway.
Click to expand...
You don't even need to make a more expensive dock that only works on the Switch Lite 2.
A Switch lite 2 with the same form factor as the original switch lite but with docking capabilities would force buyers to buy one or more controllers or a joycon+grip separately.

Sell the lite at 250-275$. Joycons are 70-80$, A pro controller is about the same. Sell a dock for both the lite and regular model for around another 70-80$ and you got both models (assuming the regular one is 399$) costing about the same.
Difference being, the other one comes bundled with everything, has detachable controllers and better battery life.

I know which one I'm getting. Even if it costs slightly more.
 
oldpuck said:
Nvidia produced a custom version of DLSS 1.0 (informally called DLSS 1.9) that ran on CUDA cores instead of tensor cores. It was only ever used in Control, IIRC, but that would run on the existing Switch, presuming it was ported to run on ARM. Mario Kart 4k would be possible with an upclocked Mariko and DLSS 1.0.
Click to expand...
Would it? Even if it was a version that didn't need tensor cores, there's still just the matter of Switch being low end hardware. We rarely see Switch doing something even like FSR1 at 1080p resolution. To do something significantly more intensive like DLSS1 at 4K? Seems like that'd have to be one extremely turbo Mariko.
 
kasparov77 said:
You don't even need to make a more expensive dock that only works on the Switch Lite 2.
A Switch lite 2 with the same form factor as the original switch lite but with docking capabilities would force buyers to buy one or more controllers or a joycon+grip separately.

Sell the lite at 250-275$. Joycons are 70-80$, A pro controller is about the same. Sell a dock for both the lite and regular model for around another 70-80$ and you got both models (assuming the regular one is 399$) costing about the same.
Difference being, the other one comes bundled with everything, has detachable controllers and better battery life.

I know which one I'm getting. Even if it costs slightly more.
Click to expand...
Eh, but can they really do that? See, the Lite has a different formfactor and less thermal headroom, and I doubt they'll bother integrating the crossbar and other video related hardware for the CHANCE someone might use an OPTIONAL accessory. It makes far more sense for it to have its own, dedicated dock where they can offload some of those components to save money on the device itself. It's a device dedicated to handheld play. Why give it the circuitry to be anything but that, when you can charge extra for people to buy it seperately as part of the dock?

If the dock ISN'T a seperate unit to the flsgship console, I don't see it being able to dock at all. As Hermii pointed out, you're stretching the image thin at that point. The device is thermally constrained. Fine for handheld mode. But TV mode can stress even the flagship Switch's overbuilt cooling solution at times. Would they really risk it with a device that must inevitably be smaller and thus more constrained?

Unless they can make it into a money maker AND keep the price of the console itself down, what reason is there for them to even try?
 
0
ReddDreadtheLead said:
What a lovely and informative post, thank you. I did not know that about the PS5, curious if Nintendo would do any modifications for a switch succ to be very efficient about it. Of course, it would be more bespoke and not so much OTS, but still in this case.

like say, 4MB?
Click to expand...
Since it seems like the purpose of Sony going with the route they did was cost optimizing, I don't think that there's the same need for Nintendo here, be it either eMMC or eUFS.
...although, that does make me wonder, what's the cost of these drives for say, Microsoft, that would've prompted Sony to try option #3.
 
Reader mode
You must log in or register to reply here.
Please read this staff post before posting.

Furthermore, according to this follow-up post, all off-topic chat will be moderated.
Last edited:


Back
Top Bottom