- Pronouns
- He/Him
As promised - it is just criminal that Nintendo knows Diablo IV exists and doesn’t have hardware released yet that can play it. Nintendo just shot themselves in the foot, etc.
-
How about that Indie World and Partner Showcase? WestEgg, Irene, VolcanicDynamo, and Yakuza Superfan Derachi Discuss the latest on the newest episode of the Famiboards Discussion Club! Check it out here!
StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)
- Thread starter Dakhil
- Start date
As promised - it is just criminal that Nintendo knows Diablo IV exists and doesn’t have hardware released yet that can play it. Nintendo just shot themselves in the foot, etc.
Angel Whispers
Socials at @angelpagecdlmhq
Nintendo should have asked Microsoft for every Blizzard game in the next 10 years instead of some weird fish
- Pronouns
- He/Him
Absolutely.
Maxrunner
Like Like
You don't even need the sensor bar, gyro only is not a good solution for full aiming since it starts to loose it's initial position.
eye0h
"aggressively persue a multiplatform strategy"
- Pronouns
- He/Him
It is LOL
A late response to the "Nintendo is gonna Nintendo" is not warranted.
I'm currently trying to get my mother to download Stella Glow on my 3ds that I have at her house because some company that is completely normal just decided to shut down the 3ds and WiiU eshop.
Now, I have no idea how big the game is, so I can't give her an estimate of how long it will take to download, so I checked! It's 11k blocks. Wtf is a block? I opened the internet and converted to MB, because Nintendo is a totally normal company.
Update from normal-land! I told her to download it now, but I realized it wasn't a good idea. Can I change the download method? Of course not! I have to make her close the lid, wait a few seconds and NOW she can touch the sleep mode button. Normal!
I'm currently trying to get my mother to download Stella Glow on my 3ds that I have at her house because some company that is completely normal just decided to shut down the 3ds and WiiU eshop.
Now, I have no idea how big the game is, so I can't give her an estimate of how long it will take to download, so I checked! It's 11k blocks. Wtf is a block? I opened the internet and converted to MB, because Nintendo is a totally normal company.
Update from normal-land! I told her to download it now, but I realized it wasn't a good idea. Can I change the download method? Of course not! I have to make her close the lid, wait a few seconds and NOW she can touch the sleep mode button. Normal!
Last edited:
Dizzle
Like Like
- Pronouns
- He/Him
What's this, your tight 5?
Thraktor
"[✄]. [✄]. [✄]. [✄]." -Microsoft
- Pronouns
- He/Him
Yeah, I could see Aula originally being planned as a DSi/n3DS style mid-gen upgrade with a modest boost in performance. The 4K support may well have been something which was only an incidental feature, just added because they had all the hardware there to output 4K so they may as well enable it (like Xbox One S), then latched onto by journalists and others as if it was a key feature. Between Switch selling far better than expected in 2020 and 2021, and the impact of COVID on development making the work involved in supporting extra performance profiles less appealing, they may have decided that the increase in performance wasn't worth it and they pivoted to just releasing it as the OLED model instead.
I agree that it's not 100% conclusive, and to play devil's advocate it's also technically possible that T239 has a separate four lane DisplayPort interface that we don't know about, but I've been trying to piece together the reports of a "4K" model being cancelled with the apparent continued development of T239, and it increasingly seems like the obvious answer is that they were always completely different things.
chocolate_supra
Must proceed at a high velocity #TeamDecade
- Pronouns
- He/Him
NGL, I went back and read it again and couldn't help but hear it in Seinfeld's voice.
D
Deleted member 887
Guest
What is the DEAL with DOWNLOADS?
LinkURL
Bob-omb
- Pronouns
- He/him
With a minimally decent camera it would be possible for the joycon itself to identify the TV without the need for a sensor bar.
Thraktor
"[✄]. [✄]. [✄]. [✄]." -Microsoft
- Pronouns
- He/Him
Okay, I've completely gone down the rabbit hole of the Switch DisplayPort lanes mystery today, so I figure I may as well delve as deep as I can. For anyone who's reading this trying to make sense of my madness, please read this post, the second part of this post and then this post.
In the last post, I posited that, although the TX1 SoC supports four lanes of DisplayPort 1.2a, and the PI3USB30532 USB-C crossbar switch used in the original Switch supports sending four lanes of DisplayPort 1.2 over USB-C, I didn't know if all four lanes were actually wired up. I've subsequently realised that with an image of the Switch's motherboard and a pinout for the PI3USB30532, I can actually answer this question myself. A quick search showed me that someone had very helpfully posted high-res photos of a desoldered Switch motherboard online, and combining this with a pinout diagram of the crossbar switch I can definitively say that the original Switch only has DisplayPort lanes DP0 and DP1 wired up between the SoC and the crossbar switch, and lanes DP2 and DP3 are unconnected.
That is, although the SoC and USB-C port on the original Switch both supported the four lanes of DisplayPort data necessary for 4K output, it's not possible for it to actually output 4K, as only two of the lanes are wired up. This is an entirely sensible decision considering Nintendo never intended to support 4K, and the original dock wouldn't have been able to output a 4K signal anyway. Wiring those lanes up would have just added needless complexity to the motherboard.
So, how about the "external_display_full_dp_lanes" configuration flag which suggests that Aula (aka the OLED Switch) is the only model which can use all four DisplayPort lanes supported by TX1/Mariko? Well, the helpful person who posted the desoldered o.g. Switch board photos is even more helpful than I thought, as they also posted desoldered OLED model board photos! Shoutout to KreativDax if they ever see this!
We can do the exact same thing with the OLED model's motherboard, which also uses the PI3USB30532 USB-C crossbar switch, and you can quite clearly see all four DisplayPort lanes wired up between the Mariko SoC and the USB-C crossbar Switch. That is, from a hardware point of view, everything in the Switch OLED model seems to be wired up to support output of 4K resolution video over the USB-C port. We've also known since it released that the upgraded dock that comes with the OLED model supports HDMI 2.0 output, which means the entire Switch OLED system appears to be 4K capable, seemingly just without software support.
Coming back to the Switch OS info, the "external_display_full_dp_lanes" flag indicates what I expected; that Aula (Switch OLED) is the only model physically capable of using all four DisplayPort lanes for external displays. The others either only have two lanes wired up (o.g. Switch) or have no external display support at all (Switch Lite). Having a flag in the OS for this is useful if you want to be able to output something over those lanes (ie 4K video), as software will need to know if its running on hardware that can do so or not. This flag was added in OS update 10.0, which I believe is the first update that added any references to Aula.
The other relevant Switch OS item is the "4kdp_preferred_over_usb30" that I originally posted about. From the name, we can infer that this toggles between using all four lanes of DisplayPort to send a 4K signal, or using only two DisplayPort lanes, and sending USB 3 data at the same time. This was added in OS update 12.0, which was also when the first references to CrdA (ie Cradle Aula, the Switch OLED dock) were added. This toggle makes little sense if the Switch model in question doesn't have four lanes of DisplayPort wired up to the USB-C port, which we've found out is the case for the original model. However, it makes a lot of sense for a model where you've wired up all four DP lanes (Aula), and when the OS now supports a dock which can take a 4K signal and output it to a TV (CrdA).
For quite a while I've thought that the use of HDMI 2.0-capable hardware in the OLED dock was simply a matter of it being cheaper or easier to source than the chip they were originally using, and the HDMI 2.0 support was incidental. I'm not so sure about that now, though. It was plausible that the apparent 4K support in the dock was incidental, but I don't think it's plausible that the 4K support they added to Switch OLED itself is incidental. The only plausible reason to wire up all four lanes of the SoC's DisplayPort to the USB-C port is if you intend to output resolutions higher than 2 lanes can handle. For standard TV resolutions, that means 4K. They chose not to wire these up on the original Switch, and as I mentioned above, that's a sensible decision if you don't intend on supporting 4K. Going from only wiring up two lanes on the OG model to wiring up all four lanes on the OLED model is a clear indication that they intended to use those two additional lanes for something. On top of this they added software into Switch's OS related to the use of those four lanes when adding Aula support, with one setting specifically relating to 4K output.
The more I look at it, the more I think that the OLED model is the "4K Switch", with Nintendo dropping 4K output and higher clocks late enough in development that the motherboard had already been finalised and much of the OS support had already been added. The hardware seems to have everything necessary to output at 4K, and at least some of the software support seems to have been in place. It explains why we were hearing about developers having dev kits for a 4K capable Switch all the way back in 2020, and why as many as 11 developers were willing to talk about it in 2021, as the release would have been relatively imminent, meaning many dev kits would have been out there. It also explains why developers would report that a 4K Switch was cancelled, because functionally the "4K Switch" was cancelled from a game development point of view, even if the hardware ended up in people's hands as the OLED model. Furthermore, it explains why Nintendo, who don't usually comment on rumours, explicitly stated that no third parties had 4K Switch dev kits, because by the time they said that they'd already dropped the plans and recalled the dev kits (or even just disabled 4K output via a software update).
TLDR: The original Switch wasn't physically capable of outputting 4K via its USB-C port, but Nintendo made specific changes to the motherboard of the OLED model to support 4K output. We already know that the OLED dock was also upgraded with hardware capable of 4K output. Software changes relating to 4K output were also added to the OS alongside support for this hardware. Therefore, I believe that the Switch OLED model is the elusive Switch Pro/Switch 4K, and that they only decided to drop 4K support and higher clocks at a late stage after the hardware had been finalised.
Edit: As a final clarifying point, although I'm talking about the technical capability to output 4K video in this post, I don't think the device was actually intended to render games at or near 4K resolution. The hardware would have been the same Mariko chip we got in the OLED model but with higher clocks, meaning noticeable but not massive performance improvements at best. I suspect that the 4K output was a secondary benefit, perhaps for a small number of games or for video streaming, like the Xbox One S. The public perception of it as a "Switch 4K" was likely due to the press focussing on the 4K output as an easy thing for non-technical audiences to latch onto, even if it was never a focus for Nintendo.
In the last post, I posited that, although the TX1 SoC supports four lanes of DisplayPort 1.2a, and the PI3USB30532 USB-C crossbar switch used in the original Switch supports sending four lanes of DisplayPort 1.2 over USB-C, I didn't know if all four lanes were actually wired up. I've subsequently realised that with an image of the Switch's motherboard and a pinout for the PI3USB30532, I can actually answer this question myself. A quick search showed me that someone had very helpfully posted high-res photos of a desoldered Switch motherboard online, and combining this with a pinout diagram of the crossbar switch I can definitively say that the original Switch only has DisplayPort lanes DP0 and DP1 wired up between the SoC and the crossbar switch, and lanes DP2 and DP3 are unconnected.
That is, although the SoC and USB-C port on the original Switch both supported the four lanes of DisplayPort data necessary for 4K output, it's not possible for it to actually output 4K, as only two of the lanes are wired up. This is an entirely sensible decision considering Nintendo never intended to support 4K, and the original dock wouldn't have been able to output a 4K signal anyway. Wiring those lanes up would have just added needless complexity to the motherboard.
So, how about the "external_display_full_dp_lanes" configuration flag which suggests that Aula (aka the OLED Switch) is the only model which can use all four DisplayPort lanes supported by TX1/Mariko? Well, the helpful person who posted the desoldered o.g. Switch board photos is even more helpful than I thought, as they also posted desoldered OLED model board photos! Shoutout to KreativDax if they ever see this!
We can do the exact same thing with the OLED model's motherboard, which also uses the PI3USB30532 USB-C crossbar switch, and you can quite clearly see all four DisplayPort lanes wired up between the Mariko SoC and the USB-C crossbar Switch. That is, from a hardware point of view, everything in the Switch OLED model seems to be wired up to support output of 4K resolution video over the USB-C port. We've also known since it released that the upgraded dock that comes with the OLED model supports HDMI 2.0 output, which means the entire Switch OLED system appears to be 4K capable, seemingly just without software support.
Coming back to the Switch OS info, the "external_display_full_dp_lanes" flag indicates what I expected; that Aula (Switch OLED) is the only model physically capable of using all four DisplayPort lanes for external displays. The others either only have two lanes wired up (o.g. Switch) or have no external display support at all (Switch Lite). Having a flag in the OS for this is useful if you want to be able to output something over those lanes (ie 4K video), as software will need to know if its running on hardware that can do so or not. This flag was added in OS update 10.0, which I believe is the first update that added any references to Aula.
The other relevant Switch OS item is the "4kdp_preferred_over_usb30" that I originally posted about. From the name, we can infer that this toggles between using all four lanes of DisplayPort to send a 4K signal, or using only two DisplayPort lanes, and sending USB 3 data at the same time. This was added in OS update 12.0, which was also when the first references to CrdA (ie Cradle Aula, the Switch OLED dock) were added. This toggle makes little sense if the Switch model in question doesn't have four lanes of DisplayPort wired up to the USB-C port, which we've found out is the case for the original model. However, it makes a lot of sense for a model where you've wired up all four DP lanes (Aula), and when the OS now supports a dock which can take a 4K signal and output it to a TV (CrdA).
For quite a while I've thought that the use of HDMI 2.0-capable hardware in the OLED dock was simply a matter of it being cheaper or easier to source than the chip they were originally using, and the HDMI 2.0 support was incidental. I'm not so sure about that now, though. It was plausible that the apparent 4K support in the dock was incidental, but I don't think it's plausible that the 4K support they added to Switch OLED itself is incidental. The only plausible reason to wire up all four lanes of the SoC's DisplayPort to the USB-C port is if you intend to output resolutions higher than 2 lanes can handle. For standard TV resolutions, that means 4K. They chose not to wire these up on the original Switch, and as I mentioned above, that's a sensible decision if you don't intend on supporting 4K. Going from only wiring up two lanes on the OG model to wiring up all four lanes on the OLED model is a clear indication that they intended to use those two additional lanes for something. On top of this they added software into Switch's OS related to the use of those four lanes when adding Aula support, with one setting specifically relating to 4K output.
The more I look at it, the more I think that the OLED model is the "4K Switch", with Nintendo dropping 4K output and higher clocks late enough in development that the motherboard had already been finalised and much of the OS support had already been added. The hardware seems to have everything necessary to output at 4K, and at least some of the software support seems to have been in place. It explains why we were hearing about developers having dev kits for a 4K capable Switch all the way back in 2020, and why as many as 11 developers were willing to talk about it in 2021, as the release would have been relatively imminent, meaning many dev kits would have been out there. It also explains why developers would report that a 4K Switch was cancelled, because functionally the "4K Switch" was cancelled from a game development point of view, even if the hardware ended up in people's hands as the OLED model. Furthermore, it explains why Nintendo, who don't usually comment on rumours, explicitly stated that no third parties had 4K Switch dev kits, because by the time they said that they'd already dropped the plans and recalled the dev kits (or even just disabled 4K output via a software update).
TLDR: The original Switch wasn't physically capable of outputting 4K via its USB-C port, but Nintendo made specific changes to the motherboard of the OLED model to support 4K output. We already know that the OLED dock was also upgraded with hardware capable of 4K output. Software changes relating to 4K output were also added to the OS alongside support for this hardware. Therefore, I believe that the Switch OLED model is the elusive Switch Pro/Switch 4K, and that they only decided to drop 4K support and higher clocks at a late stage after the hardware had been finalised.
Edit: As a final clarifying point, although I'm talking about the technical capability to output 4K video in this post, I don't think the device was actually intended to render games at or near 4K resolution. The hardware would have been the same Mariko chip we got in the OLED model but with higher clocks, meaning noticeable but not massive performance improvements at best. I suspect that the 4K output was a secondary benefit, perhaps for a small number of games or for video streaming, like the Xbox One S. The public perception of it as a "Switch 4K" was likely due to the press focussing on the 4K output as an easy thing for non-technical audiences to latch onto, even if it was never a focus for Nintendo.
Last edited:
Foxuzamakk
Cappy
- Pronouns
- He/him
Thing is, console development is getting longer and longer, they can't afford to wait. They most likely will have it be cross-gen but they are gonna need to get use to the next hardware as soon as possible if they want to get anything out within 5 years
Goodtwin
Like Like
Nintendo Switch Pro was likely going to be a "New" Nintendo Switch with more in common with the DSi and New 3DS rather than a PS4 Pro or Xbox One X. Personally I see the Tegra X2 Parker SOC as the likely candidate thanks to it 128bit memory bus and higher clocks speeds. I just don't see a mid gen refresh getting something as exotic as Drake. The journalist got details of the true next gen Switch mixed up with the New Switch, and this is what caused a lot of confusion.
Thraktor
"[✄]. [✄]. [✄]. [✄]." -Microsoft
- Pronouns
- He/Him
Yeah, and I think it also would explain the seeming contradiction between Nintendo cancelling a 4K Switch and there not being an angry outpouring of developers complaining about the cancellation. This "4K Switch" would have been powered by Mariko, with presumably higher clocks. At the absolute highest achievable clocks on Mariko you would have been looking at a system perhaps 50% more powerful than the original Switch, with about 30% more RAM bandwidth. There's no way that anyone was developing games exclusively for Aula without supporting the original Switch with performance so close between the two. Which means that when Nintendo cancelled it, it wasn't "Hey, you know that exclusive port of Metro Exodus you were working on that never would have run on the OG Switch? That effort's all wasted now", it was "You know that game you're making for Switch, where you were going to support a slightly higher performance profile for new hardware? You don't need to support that profile any more." If anything the cancellation just meant less work for developers as they had fewer hardware targets to worry about.
I should also emphasise that I don't think it was ever Nintendo's intention to actually have Switch games running at, or even near 4K on this thing. It's just an overclocked Mariko, after all. I suspect that 4K output was just a secondary feature which they didn't expect games to utilise (like on the Xbox One S), but it was picked up by the press because 4K is an easy buzz-word for people without technical knowledge to latch onto.
there's your usecase for tensor cores: identifying the tv
John Omaha
Moblin
I wouldn't be surprised if it was just a die-shrunk X1 with higher clocks.
that's what Mariko is. they can't die shrink it any further without, essentially, making a new chip
Thraktor
"[✄]. [✄]. [✄]. [✄]." -Microsoft
- Pronouns
- He/Him
From the Switch OS we know that Aula was always intended to use Mariko, and the Switch OLED motherboard (which has the additional DisplayPort lanes necessary for 4K) is only compatible with Mariko, so I doubt Parker was ever in the picture. Besides, Mariko can clock at least as high as Parker and doesn't have those big inefficient Denver cores.
I agree that it was intended as a modest performance improvement (although the performance improvement of the n3DS was actually quite large in some areas).
kvetcha
hoopy frood
- Pronouns
- he/him
I do wonder if it’s feasible to have single-frame markers in the screen that could act as tracking points for the camera.
- Pronouns
- She/Her
I wouldn't bet on that necessarily being the case for GameFreak and Nintendo.
I don't think the infrared camera could pick that up since tvs don't output various IR wavelengths like it does visible wavelengths (other than hotspots). so that's a no-go
ezidro3
Moblin
- Pronouns
- he/him
I would not be surprised if by the time Drake releases, some media apps either launch or get updated to enable 4K output and this gets retroactively added to the OLED model
kvetcha
hoopy frood
- Pronouns
- he/him
I’m curious what the actual wavelengths are that the camera can pick up.
OP
OP
Just to add, the Tegra X2's GPU's likely very similar, if not identical, to the Tegra X1's GPU, especially with the Tegra X2's GPU not supporting DP4a instructions, despite Nvidia advertising the Tegra X2's GPU as Pascal based, and Nvidia mentioning about introducing DP4a with Pascal.
LiC
Member
This is great analysis of the hardware (though I'd be interested to know whether the Mariko board in the V2 was wired like the V1 or like the OLED?), but I think the bigger picture is missing a few things. The 11 developers bit was reported in 2021 after the OLED was announced, where Mochizuki doubled down on the existence of a 4K Switch which all these developers were working on and expecting to release games for (which presumably means exclusive or enhanced games) "in or after the second half of 2022." While we all mostly agree that there most have been some conflation of something to explain the incorrect reporting, this part in particular opposes the idea of 4K OLED plans. And even if things got mixed up, you'd think that somebody at that point would have been able to straighten it out after the fact, with so many apparent sources. It also doesn't make sense that the story of developer tools for new hardware would have become about 4K support in the first place, since developers don't reverse engineer firmware and only know what Nintendo tells them in SDK documentation. If 4K was just some incidental feature that games couldn't render at, and wasn't a focus for Nintendo, then how did that become something multiple sources talked to Bloomberg about, to the point that it was treated as a real system feature instead of a mostly irrelevant technical detail? How would that it make it through, but not the easily communicable fact that it was using the same TX1+ chip with higher clocks (both things which SDK documentation would have stated explicitly)? And that not only didn't make it through, but ended up explicitly contravened by Bloomberg's March reporting that the 2021 system would use a new Nvidia chip. Again, there was probably some conflation somewhere, but "does the thing you have use the same old chip or a new one" is an easy question to get right if you have 11 or more sources. I'm also not sure Nintendo even would have undertaken the project of distributing separate SDKs and tools to a limited audience (which we've never seen leaked, despite the fact that lots of developer supposedly had it, and despite the fact that it would have been cancelled out from under them) for such a revision.
I could go on, but I'm not organizing my thoughts well already. Basically, the combination of DisplayPort settings, SoC support, and dock support is interesting, but the notion of "4K Mariko OLED" as a product doesn't quite fit with other things we know, and is not enough to untangle the mess of reporting around 2021 hardware (let alone the vast gulf of bullshit we need to cross to understand its mutation into the reporting on 2022-2023 hardware). I can still see a more minimal version of your explanation, where Nintendo did intentionally add 4K support to the OLED model and firmware, but had no plans to expose it as a real system feature, in which case it wouldn't have merited a limited audience of third party tool distribution and wouldn't have caused any of the reporting we got.
- Pronouns
- She/Her
I think, possibly, it may have been a case of a last minute pushback. In theory the chip being ready in March to May could hit shelves later the same year, but something slipped and the whole house of cards came down with it, and now they're slowly winding up again to try and get it out this year. Maybe they HAD planned to launch it with TOTK, but then TOTK slipped and development refocused on finishing the Switch version while the next gen patch planned for launch got shelved for the time being. This had a knock on effect in confidence within the company and they elected to ride out 2022 and early 2023 with their initial batch of what would have been cross gen titles (someone might kill me for this, but the rough technical state of first party published Nintendo Switch games lately really has my eyebrow raised, Bayonetta 3 and Pokémon Scarlet/Violet especially seem begging for better hardware).
Now we sit here in late March... With nothing for August. Nothing for September. Nothing for anything after July, and Pikmin 4 is NOT a keystone title in the calendar. Tears of the Kingdom in early May, an empty June, and an OK July. That doesn't sound like the plan you'd want to have if you're a company with a hurting stock price and a slowly stagnating, if dominant, console.
Near enough a year since the TOTK delay was announced, a year to realign goals, reassure publishing partners, and make sure the factories are ready (which according to supposed leaks from said factories, they are). Now would be about the time they're preparing marketing material, if not finalising it, for a marketing campaign launching no later than August (I can't see them going from July 21st to September 1st with nothing to announce and nothing to release, and I don't think that announcement will be "more games for Nintendo Switch 1", or we would know about some of them by now.)
I'd also like to bring up something perhaps a little in the financial weeds, but earlier pre-orders is generally financially favourable. The earlier you can announce and get pre-orders up without ruining your marketing cycle the better. Basically, pre-orders provide a company with liquid cash up front before the product is even released, and provides data as to how a game will perform. These are extremely useful to a company. Even if nowadays Nintendo doesn't get all the money up front, there's still people buying Game Vouchers and eShop credit and linking credit cards so they can be assured of a safe cash infusion on launch day at the very least.
Not to belabour the point, but, if they had Switch 1 games worth pre-ordering past July... We'd probably know about them by now!
Chicken Little ain't got nothing on this thread.
- Pronouns
- She/Her
That would:
A. Be extremely welcome news, in my opinion.
B. Line up with Samsung appearing in the leaked documentation.
Look over there
Like Like
Well, that'd be a surprise considering that would presumably mean Nvidia secured capacity on that node solely for the REDACTED.
...it's also technically the worst of the options my gut would find doable
(I'm still not sold on 8LPP being able to deliver an acceptable portable battery life, so I'm left with 7nm generation and newer nodes... and 5LPP's slightly worse than the N7 nodes)
Edit:
Coping time: it's not that much worse in the big picture. And it can be a worthwhile tradeoff if it's in the form of a nice package deal... and/or Samsung understands the need to price appropriately
...it's also technically the worst of the options my gut would find doable
(I'm still not sold on 8LPP being able to deliver an acceptable portable battery life, so I'm left with 7nm generation and newer nodes... and 5LPP's slightly worse than the N7 nodes
)Edit:
Coping time: it's not that much worse in the big picture. And it can be a worthwhile tradeoff if it's in the form of a nice package deal... and/or Samsung understands the need to price appropriately
Last edited:
LiC
Member
From scrolling through that Twitter feed, it looks like they just post a lot about phones and mobile SoCs, and any correct predictions in there could easily be swimming in a sea of guesses and misses. Let's not just trust someone is a reliable Samsung leaker because they're Korean.
The node needs to meet performance targets. If it meets them, it meets them. The fact that another (less available, more expensive) node could allow exceeding performance targets is pretty much irrelevant to Nintendo's decision-making.Well, that'd be a surprise considering that would presumably mean Nvidia secured capacity on that node solely for the REDACTED.
...it's also technically the worst of the options my gut would find doable
(I'm still not sold on 8LPP being able to deliver an acceptable portable battery life, so I'm left with 7nm generation and newer nodes... and 5LPP's slightly worse than the N7 nodes
Edit:
Coping time: it's not that much worse in the big picture. And it can be a worthwhile tradeoff if it's in the form of a nice package deal... and/or Samsung understands the need to price appropiately
- Pronouns
- She/Her
Leaked documentation points to 9W on the GPU in TV mode, and a Samsung fabrication. 5LPP meets the criteria without exception. No "may have just been copied over", no "magic secret sauce super efficient 8nm". Just a good fit. Sue me, I'm pumped on the basis that this really fits the pieces together.
Kise Ryota
kurokocchi
SEC 5LPP... meh
I hope not
I hope not
OP
OP
As I've mentioned before, Nintendo's and Nvidia's decision to have Drake's GPU have 1 MB of L2 cache instead of 4 MB of L2 cache like with Orin's GPU makes more sense with Samsung as the foundry partner for reasons explained here.
Zedark
The guy with the ToV avatar
Does anyone have a (link to a) map/explanation of the different process nodes and how they compare in terms of density and performance (in a quantitative sense)? Might be useful for perspective in case this person is legit (don't know of them, but I don't follow the mobile phone scene at all, so that means literally nothing).
Edit: Also, how does 5LPP compare to 5LPE? I can't find even a mention of 5LPP anywhere on Google, let alone solid numbers...
Edit 2: This might help:
I've found 5LPE roughly matches 7nm TSMC, as well.
Edit: Also, how does 5LPP compare to 5LPE? I can't find even a mention of 5LPP anywhere on Google, let alone solid numbers...
Edit 2: This might help:
I've found 5LPE roughly matches 7nm TSMC, as well.
Last edited:
HankyPanky
Like Like
This node makes a lot of sense. Isn't the most expensive option, seems to fit with all leaks & speculation and would likely give the required performance/efficiency for Drake. Could be better but Nintendo were always going to take 'good enough' to keep the price down. Especially if this is being marketed as the next Switch & not a pro revision, it needs to come in somewhat affordably.
Orin/Ampere 8LPP
Drake 5LPP
presumably the process of using a node in the same family? is much easier/more affordable than changing foundry altogether.
Orin/Ampere 8LPP
Drake 5LPP
presumably the process of using a node in the same family? is much easier/more affordable than changing foundry altogether.
Last edited:
- Pronouns
- He/Him
Even 349.- would already be a great price point.
Z0m3le
Bob-omb
@Thraktor Great work, I've mentioned in the past that Mariko has known clocks that were never used, 1.267GHz for 648GFLOPs when docked and a CPU clock close to the boosted 1.78GHz clock when loading. From what I had heard and mentioned in the past, they couldn't meet expectations of performance, and decided to scrap the higher clocks, to note this was not really a battery decision as the higher GPU clocks were for docked mode, I have no information on if the CPU higher clock was also for docked, the OLED model is already an upgrade for handheld users based on the OLED screen, and I think the docked clocks would have allowed Switch games to hit their dynamic resolution's maximum settings and push more stable frame rates when docked. It's too bad that they never opened these clocks, but technically they could do so even now.
This is what I've been expecting for a while now, it lines up well with what I heard in the past about Samsung picking up the contract, I mentioned years ago that I had heard that Samsung made a deal with Nintendo for at least the initial production, to be a partner on 4 or 5 components for this device, however this information is prior to the pandemic, and I had heard nothing about it since, not hearing an update in 3 and a half years or so, does make you wonder if things had changed, and maybe they have considering the chip shortage. This is also loosely indicated in the Nvidia hack, that Samsung was the foundry for Drake, confirmed in NVN files for initializing iirc? It also makes those DLSS test clocks possible. I very much believe those are real.
2TFLOPs handheld
3.456TFLOPs docked
As for the CPU clock, it's possible we will see 2GHz on 7 cores, with 1 reserved for the OS at a lower clock to save power, however my guess at the clock has remained an estimation, and I think it's very possible to be at 1.8GHz.
I also want to thank Thraktor's post from months ago, breaking down an estimation on costs for different process nodes on Drake, pointing out that Samsung 5nm would likely be cheaper than Samsung's 8nm process node. I do think it being ~10% better than Samsung 4LPX node is good, but I'd suggest the actual node is 4N, which I believe is based on 5LPP which I don't think exists anymore.
5nm LPP should offer ~60% higher clocks for both GPU and CPU... TSMC 5nm would have offered as much as 75% higher clocks.
This is what I've been expecting for a while now, it lines up well with what I heard in the past about Samsung picking up the contract, I mentioned years ago that I had heard that Samsung made a deal with Nintendo for at least the initial production, to be a partner on 4 or 5 components for this device, however this information is prior to the pandemic, and I had heard nothing about it since, not hearing an update in 3 and a half years or so, does make you wonder if things had changed, and maybe they have considering the chip shortage. This is also loosely indicated in the Nvidia hack, that Samsung was the foundry for Drake, confirmed in NVN files for initializing iirc? It also makes those DLSS test clocks possible. I very much believe those are real.
2TFLOPs handheld
3.456TFLOPs docked
As for the CPU clock, it's possible we will see 2GHz on 7 cores, with 1 reserved for the OS at a lower clock to save power, however my guess at the clock has remained an estimation, and I think it's very possible to be at 1.8GHz.
I also want to thank Thraktor's post from months ago, breaking down an estimation on costs for different process nodes on Drake, pointing out that Samsung 5nm would likely be cheaper than Samsung's 8nm process node. I do think it being ~10% better than Samsung 4LPX node is good, but I'd suggest the actual node is 4N, which I believe is based on 5LPP which I don't think exists anymore.
If it was 8nm, it would have been 1.3TFLOPs and the CPU would have likely been 1.2-1.4GHz.
5nm LPP should offer ~60% higher clocks for both GPU and CPU... TSMC 5nm would have offered as much as 75% higher clocks.
I'm not sure this really resolves the Bloomberg reporting so much as possibly add a bit more flavor as to how the conflation may have happened. Aside from the initial preparing for 4k report, those all pretty unambiguously had some Drake details in the mix.
That said, it is interesting that this capability seemingly managed to survive to the retail version. By the time Aula began to show up in the retail firmware, it had already seemingly been stripped of any additional power it may have once had, so it's possible this is just something that was just considered harmless enough that it wasn't worth redoing the motherboard over. Whether they ever communicated this capability to third parties is up for debate, though if they did, it may help to explain a few details around the margins (though certainly not any events that supposedly happened after the system's release).
It's too bad. 4k output is the one thing that could have made Switch OLED worth buying for me.
Lelouch0612
Growing Install Base
- Pronouns
- He/Him
Does the 5LPP node offer an easy die-shrink path ?
I know Samsung 8nm did not and it was one of issue with that node (in terms of future revisions).
I know Samsung 8nm did not and it was one of issue with that node (in terms of future revisions).
Z0m3le
Bob-omb
I'm not sure they will die shrink this chip, it should offer ~3.5hours in handheld mode, based on a slightly larger (4315mah to 5000mah), though cheaper and more easy to come by battery.
Current Switch games, even when patched, would offer far more battery life, 8 to 10 hours is certainly reasonable, even longer for the average Switch game.
Lelouch0612
Growing Install Base
- Pronouns
- He/Him
They would eventually need one to make a Lite, and offer an entry level SKU.
Last edited by a moderator:
