• Hey everyone, staff have documented a list of banned content and subject matter that we feel are not consistent with site values, and don't make sense to host discussion of on Famiboards. This list (and the relevant reasoning per item) is viewable here.
  • Do you have audio editing experience and want to help out with the Famiboards Discussion Club Podcast? If so, we're looking for help and would love to have you on the team! Just let us know in the Podcast Thread if you are interested!

StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (Read the staff posts before commenting!)

Reader mode
Pokemaniac said:
Storage doesn't really have anything to do with using an eGPU. They generally just care more about how fast you can push data through whatever port they're connected to, though the more pressing issue in a Switch-like device is that they'd probably interfere with the ability to just pull the system out of the dock whenever.
Click to expand...
As a person who doesn't know a lot about the technical side of the topic, I guess having a fast storage solution that allows a game to reload it's assets in a significantly short amount of time would benefit in the transition between TV and Handheld. They could even overclock the system when transitioning, it doesn't happen very often, so if it's just a momentaneous thing, why not?
 
Please use stronger sources to not disrupt the flow of conversation. The hardware thread especially has a higher standard of sourcing. -xghost777, meatbag, Barely Able
oldpuck said:
Great post all round. This absolutely isn't my field, so it's great to hear you break it down.

I know the "double upscale" gets pooh-poohed a lot but I don't think it's that weird. Games at non-native resolution have been an issue for decades, and are still being produced on the current consoles. But DLSS isn't just an upscaler - it's finding detail that doesn't exist in the native frame. It's more like DLSS is a cheaper path to a reasonable 1440p (or 4k, or 1080p) image, and if the resolution is sufficient, then why not enable more features in that base image?

All that said, I've been messing with a theory that I wanna throw out. Credit to Richard Leadbetter for putting this worm in my head.

Let’s go back to February of last year. Prior to the leak, we were trying to game out what a cut down, Nvidia specific Orin would look like and what hardware could be cut. One of the obvious candidates was the Deep Learning Accelerator, machine learning hardware that seemed useless in a console.

Next month, with the leak in full analysis mode, we discovered Orin’s “double rate” tensor cores. The obvious question is why doesn’t Drake have them. The obvious answer was, at the time, that Nintendo decided that the extra DLSS performance isn’t necessary.

Now, much further out, with additional information, I doubt both of these conclusions.

In the case of the tensor cores, they’re not truly “double rate.” Only a small number of instructions run at double the clock speed. I suspect because these are actually useful for model training, not model execution. That would mean they don’t accelerate DLSS at all, and Nvidia reserves them for their ML products purely for product differentiation.

The DLA is actually well documented, and functions much like tensor cores. They accelerate the same matrix operations as the tensor cores, and in fact, Nvidia’s machine learning tools let you run workloads that combine them into a single compute pool.

I think you’re seeing where this is going. Speculating: DLSS is tensor core limited. Speculating: DLSS can take advantage the of DLA. If so, Nvidia/Nintendo have an option for accelerating DLSS without making the GPU larger than it’s already substantial size.

Initially I thought the DLA was ruled out because it’s not in the list of blocks in the Nvidia driver. But upon review, the block isn’t listed on that place on Orin either. It does show up in other files for Orin, but Drake doesn’t have those files at all - not because the functionality is disabled but because the work was in progress when Nvidia made it private.

Totally speculative, but y’all are smarter than me so I thought I’d throw it out.
Click to expand...
That would be quite the twist! My first thought is to ask where the DLA structure is located within the Orin die. If it's an extra structure outside the GPU IP and not within the SM structure, I think there would incur extra latency and energy-usage? I dunno. If only we could test Orin with Windows OS and profile if the DLA structure is used by DLSS, that would be a good starting point.

BTW Oldpuck, as you're in touch with Rich, can you ask him to do, in his RTX 2050 video, multiple clock x power testings, so that we can determine more or less the power usage of SEC8N at given clocks based on a mobile/laptop consumer Ampere dGPU? Unless he's already doing this. If so, please disregard.
P4bl0 said:
www.express.co.uk

Nintendo Switch 2 specs - Nintendo's console is more powerful than you think

It looks like the Nintendo Switch 2 - which could be unveiled as early as this week - will be able to match the power of the Xbox Series S.
www.express.co.uk www.express.co.uk




We so back
Click to expand...
They're basically compiling the reports of Gamescom, 12GB report from Necrolipe and others rumors like the MoneyDJ one. Anyone who is reading this thread is already aware of all of them. Still, good to see reporting of it outside the gaming sphere.
 
Last edited:
oldpuck said:
Initially I thought the DLA was ruled out because it’s not in the list of blocks in the Nvidia driver. But upon review, the block isn’t listed on that place on Orin either. It does show up in other files for Orin, but Drake doesn’t have those files at all - not because the functionality is disabled but because the work was in progress when Nvidia made it private.
Click to expand...
In my notes I've got two places where DLA definitions show up publicly for Orin but not Drake: T234/T239, and T234/T239. There are from a June 2021 commit, and work was ongoing as you mentioned, so we don't know for sure that they're final, but at this stage at least they do imply some differences. The second one is also where the NVJPG, PVA, and camera blocks appear to be removed from T239.

For completeness: I never really bothered reporting on DLA-related findings from the leak because the info seemed incomplete and not important to me. I think the one file that mentions it for T239 is just an identical copy of T234's file for the same classes, including the PVA and camera stuff, so it looks like it just wasn't updated there.

P4bl0 said:
www.express.co.uk

Nintendo Switch 2 specs - Nintendo's console is more powerful than you think

It looks like the Nintendo Switch 2 - which could be unveiled as early as this week - will be able to match the power of the Xbox Series S.
www.express.co.uk www.express.co.uk




We so back
Click to expand...
I'm sorry, but what is this garbage? The body of the article is just repeating a collection of prior rumors and reporting, with the "as soon as this week" thrown into the subhead without a lick of justification. And the comparison to the Series S seems to be entirely based on the RAM size rumor.
 
Last edited:
0
P4bl0 said:
Maybe to get it out of the way and let Wonder shine longer? Who knows

Also maybe they want to announce before GTA VI trailer end of the month
Click to expand...

Like I said in the other thread it doesn't back up the claim.

Take that claim as a nothing burger until we get more sources confirming it.

Like others have said, no way they will announce it before SMBW releases.
 
0
P4bl0 said:
Liam Martin has worked for the Daily Express since 2016. As Gaming Editor, he's covered the launch of new consoles like the PS5 and Nintendo Switch,

He could know stuff we don’t.
Click to expand...
You don't know anything about the Express, then. They're known for shoddy reporting, various fact checking sites gives them a "mixed" rating for how factual they are. Don't get your hopes up.
 
Meelow said:
Well, I emailed him asking if he could back up his claim.

I'm going to hope that you don't have actual expectations for an announcement this week.
Click to expand...
I'll admit, it's a power move to say "as soon as this week" in an article published Friday afternoon.
 
Meelow said:
Well, I emailed him asking if he could back up his claim.

I'm going to hope that you don't have actual expectations for an announcement this week.
Click to expand...

Awesome, looking forward, I don’t have expectations, I’m open to any outcome, healthier
 
P4bl0 said:
Liam Martin has worked for the Daily Express since 2016. As Gaming Editor, he's covered the launch of new consoles like the PS5 and Nintendo Switch,

He could know stuff we don’t.
Click to expand...
Right he could, but also…

“The Nintendo Switch 2 version of the Matrix Awakens demo - which originally released to show off Unreal Engine 5 - is said to have been running on DLSS 3.1, instead of DLSS 3.5 on the PS5 and Xbox Series X.”

So, ya… doubt.
 
LiC said:
I'll admit, it's a power move to say "as soon as this week" in an article published Friday afternoon.
Click to expand...

Lmao true, he obviously meant this coming week but again. Nintendo isn't going to announce the Switch successor the same week as SMBW releases.

mjayer said:
Right he could, but also…

“The Nintendo Switch 2 version of the Matrix Awakens demo - which originally released to show off Unreal Engine 5 - is said to have been running on DLSS 3.1, instead of DLSS 3.5 on the PS5 and Xbox Series X.”

So, ya… doubt.
Click to expand...

Hahaha.
 
0
ILikeFeet said:
That doesn't mean he'd know anything
Click to expand...

Well you can make a lot of connects in 10 years in this field

mjayer said:
Right he could, but also…

“The Nintendo Switch 2 version of the Matrix Awakens demo - which originally released to show off Unreal Engine 5 - is said to have been running on DLSS 3.1, instead of DLSS 3.5 on the PS5 and Xbox Series X.”

So, ya… doubt.
Click to expand...

Maybe he is not a tech guy. DLSS versions sound very confusing at first mixed with Ray Tracing etc, as I said let’s give him benefit of the doubt cause that’s such a bold claim.
Lets see if he answers the email :)
 
P4bl0 said:
Well you can make a lot of connects in 10 years in this field



Maybe he is not a tech guy. DLSS versions sound very confusing at first mixed with Ray Tracing etc, as I said let’s give him benefit of the doubt cause that’s such a bold claim.
Lets see if he answers the email :)
Click to expand...
If he doesn't know technical details then is he really a useful source in regards to telling us anything we don't already know?
 
0
Thraktor said:
Thanks for posting this, there's a lot of interesting stuff there I wasn't aware of, and some very good explanations of things I didn't fully understand before.

I'll have a read through the papers when I get the chance, the Intel one particularly seems very interesting. If I'm understanding it correctly, the two filter paths which are not-quite-convolutional-networks are the parts that are actually performing the denoising, one for diffuse lighting and one for specular. Rather than having a neural network perform denoising directly, they seem to be using the neural network to decide what filter kernels at what positions should be used to appropriately denoise the image. Then the albedo and normal buffers don't need to be denoised, just antialiased, so they use a single layer filter. I would have assumed you would just dump all the noisy data into the neural network and let it sort it out, I never would have thought to use a neural network to control a more traditional algorithm like this.
Click to expand...
Yes, I believe that’s exactly the strategy. My current understanding of the process is that in every block of the decoder, two 1x1 convolutional layers pass over the activated output for that block to generate a set of per-pixel filter kernels to pass to the corresponding layers in the filter paths, and those filter paths are what actually do the denoising, like you are saying. The denoised specular and diffuse buffers are linearly combined with the filtered albedo into a composite at a sub-output resolution, which is bilinearly upsampled to the output resolution (both tone-mapped and linear versions of the composite are generated).

This would normally be missing some high frequency information. But that missing high frequency information is retained in the output block thanks to the skip connection from the input block. So the upsampled, tone-mapped version of the composite is concatenated with the features from the output block, which is used to generate one final set of per-pixel kernels. Those kernels are applied to the linear (not tone-mapped) version of the composite, generating a final output with the higher frequency information recovered.

There is some hand-waving in places that I am still working out. In particular, I’m still figuring out some of the aspects related to the temporal blending and history rejection. They also claim earlier in the paper that they provide the surface roughness to the network to prevent overblurring, but I can’t find where that’s actually used in the details of the network. Maybe it’s just one of the input channels?

Also, it’s explicitly a recurrent approach in that it just uses the previous output frame, like you and I had theorized about for DLSS a year or two ago. It feels nice to confirm the possibility!
 
Last edited:
LiC said:
In my notes I've got two places where DLA definitions show up publicly for Orin but not Drake: T234/T239, and T234/T239. There are from a June 2021 commit, and work was ongoing as you mentioned, so we don't know for sure that they're final, but at this stage at least they do imply some differences. The second one is also where the NVJPG, PVA, and camera blocks appear to be removed from T239.

For completeness: I never really bothered reporting on DLA-related findings from the leak because the info seemed incomplete and not important to me. I think the one file that mentions it for T239 is just an identical copy of T234's file for the same classes, including the PVA and camera stuff, so it looks like it just wasn't updated there.
Click to expand...
Follow-up: I'm double checking, and there are some auto-generated files for T239 in the leak that remove most mentions of DLA while adding the FDE.
 
I mean I doubt it’s anything but it might also be something. It’s true he may have contacts, and he’s obviously pulling a lot of other rumors together, it’s probably just that but you never know. I think it’s healthy to bet it’s nothing tho.
 
0
Ghostsonplanets said:
BTW Oldpuck, as you're in touch with Rich, can you ask him to do, in his RTX 2050 video, multiple clock x power testings, so that we can determine more or less the power usage of SEC8N at given clocks based on a mobile/laptop consumer Ampere dGPU? Unless he's already doing this. If so, please disregard.
Click to expand...
I believe the video is basically done and is in the final editing phase. But Rich did share with me some power draw data, and it’s very consistent with the Orin power estimator. Take that with a huge flake of maldon finishing salt though - even the laptops use GDDR6, and you can’t separate the power draw of VRAM from the overall GPU.

LiC said:
Follow-up: I'm double checking, and there are some auto-generated files for T239 in the leak that remove most mentions of DLA while adding the FDE.
Click to expand...
There goes that bubble! Thanks for checking.
 
0
P4bl0 said:
www.express.co.uk

Nintendo Switch 2 specs - Nintendo's console is more powerful than you think

It looks like the Nintendo Switch 2 - which could be unveiled as early as this week - will be able to match the power of the Xbox Series S.
www.express.co.uk www.express.co.uk




We so back
Click to expand...
Again...
37946f7477d6058dace1d612a7d22dce9cbd9a67-569x500.png
 
Dunno where this speculation about Switch NG coming in 2025 started but absolutely NOT
The hardware will not skip 2024 and we'll have before next revisions (or upgrades) for both Xbox Series / PlayStation 5

Switch 1 will remain as focus until the end of this FY, with a decline on it's 'momentum' starting H2 2024
 
necrolipe said:
Dunno where this speculation about Switch NG coming in 2025 started but absolutely NOT
The hardware will not skip 2024 and we'll have before next revisions (or upgrades) for both Xbox Series / PlayStation 5

Switch 1 will remain as focus until the end of this FY, with a decline on it's 'momentum' starting H2 2024
Click to expand...
Any idea when Nintendo could potentially reveal it? Is this year still potentially on the table?
 
justanotherperson said:
As a person who doesn't know a lot about the technical side of the topic, I guess having a fast storage solution that allows a game to reload it's assets in a significantly short amount of time would benefit in the transition between TV and Handheld. They could even overclock the system when transitioning, it doesn't happen very often, so if it's just a momentaneous thing, why not?
Click to expand...
Games would have to be explicitly built to be able to not just crash when you took away a whole GPU and VRAM pool while they were running. It's a much greater imposition than how games deal with docking and undocking today by just tweaking some graphics settings in response (and even that isn't strictly required).
 
Differences between T234 and T239
  • Index
    • Since I had it in front of me, here are the differences in hardware definitions between T234 and T239 (if we assume certain files were up to date), based on LIC definitions (no relation; stands for Legacy Interrupt Controller). This is a pretty dry list and more for my own reference, so peruse only with discretion.

    Hidden content is only available for registered users. Sharing it outside of Famiboards is subject to moderation.

    The only unsolved thing here is the PPC and PPC2DISPLAY defines for T239. Anyone have any guesses what those are?

    Edit: PPC is not a very unique acronym, but assuming what I'm looking at is the same usage, then PPC encompasses the PES (Shared Primitive Engine), so maybe it stands for Primitive Processing Cluster? Nvidia doesn't use that term publicly, but it could refer to that part of the pipeline. Although if that's the case, I still don't know what PPC2DISPLAY means, since the primitive pipeline is not sending anything to the display.
     
    Last edited:
  • Index
    • LiC said:
    Since I had it in front of me, here are the differences in hardware definitions between T234 and T239 (if we assume certain files were up to date), based on LIC definitions (no relation; stands for Legacy Interrupt Controller). This is a pretty dry list and more for my own reference, so peruse only with discretion.

    * Hidden text: cannot be quoted. *


    The only unsolved thing here is the PPC and PPC2DISPLAY defines for T239. Anyone have any guesses what those are?
    Click to expand...
    So what I've generally turned up in a few minutes searching (side note: PPC in relation to Nvidia is not a super searchable term, and brings up a lot of complaining about Nvidia not providing PowerPC drivers) is that there's apparently something called "NVIDIA VirtualLink PPC firmware" that you can supposedly update via the Nvidia control panel (I don't see it, but I'm on Linux right now and my GPU doesn't support VirtualLink). I've also found a random old forum post which also seems to connect the "PPC" abbreviation to USB.

    VirtualLink, for those unaware, was a failed attempt at standardizing a USB Type-C alt mode to provide power, video (in the form of DisplayPort), and data over a single cable for VR headsets. As far as I'm aware, the port only ever appeared on Turing GPUs, and was already dead and buried by the time Nvidia released their desktop Ampere cards, so this would probably be the only Ampere chip to have anything related to that. The VirtualLink workload seems actually kinda similar to how the Switch works (albeit with the power flowing in the opposite direction), so perhaps some of their VirtualLink work was salvaged for Drake?

    That said, I could be way off base. I did also find some stuff about "PPC" in relation to ACPI, but that seemed unrelated.
     
    Pokemaniac said:
    So what I've generally turned up in a few minutes searching (side note: PPC in relation to Nvidia is not a super searchable term, and brings up a lot of complaining about Nvidia not providing PowerPC drivers) is that there's apparently something called "NVIDIA VirtualLink PPC firmware" that you can supposedly update via the Nvidia control panel (I don't see it, but I'm on Linux right now and my GPU doesn't support VirtualLink). I've also found a random old forum post which also seems to connect the "PPC" abbreviation to USB.

    VirtualLink, for those unaware, was a failed attempt at standardizing a USB Type-C alt mode to provide power, video (in the form of DisplayPort), and data over a single cable for VR headsets. As far as I'm aware, the port only ever appeared on Turing GPUs, and was already dead and buried by the time Nvidia released their desktop Ampere cards, so this would probably be the only Ampere chip to have anything related to that. The VirtualLink workload seems actually kinda similar to how the Switch works (albeit with the power flowing in the opposite direction), so perhaps some of their VirtualLink work was salvaged for Drake?

    That said, I could be way off base. I did also find some stuff about "PPC" in relation to ACPI, but that seemed unrelated.
    Click to expand...
    There seem to be a lot of different definitions, and the existence of PowerPC definitely makes it not very searchable. But I did just find one places that spells it out as "primitive processor clusters" as I suspected, being a sub-unit of the GPC encompassing the PES. So I think that's what the T239 defines are probably for. Then again, there's another possibility in Port Policy Controller, which is related to USB ports, and could explain the "PPC2DISPLAY" define.
     
    LiC said:
    Edit: PPC is not a very unique acronym, but assuming what I'm looking at is the same usage, then PPC encompasses the PES (Shared Primitive Engine), so maybe it stands for Primitive Processing Cluster? Nvidia doesn't use that term publicly, but it could refer to that part of the pipeline. Although if that's the case, I still don't know what PPC2DISPLAY means, since the primitive pipeline is not sending anything to the display.
    Click to expand...
    Must be related to USB C and USB C Display. Port Policy Controller? could be.
     
    LiC said:
    There seem to be a lot of different definitions, and the existence of PowerPC definitely makes it not very searchable. But I did just find one places that spells it out as "primitive processor clusters" as I suspected, being a sub-unit of the GPC encompassing the PES. So I think that's what the T239 defines are probably for. Then again, there's another possibility in Port Policy Controller, which is related to USB ports, and could explain the "PPC2DISPLAY" define.
    Click to expand...
    It's possible that "Port Policy Controller" could be the VirtualLink "PPC", but there's probably no way to tell for sure without comparing against one of the 20 series cards.
     
    ertaboy356b said:
    Must be related to USB C and USB C Display. Port Policy Controller? could be.
    Click to expand...
    Yeah, on second thought, I think that's the more likely definition. The files I'm looking do also have definitions related to PPC = Primitive Processing Cluster, but the existence of "PPC2DISPLAY," and the fact these are defines T239 has but T234 doesn't, make me think that it's Port Policy Controller.

    Pokemaniac said:
    It's possible that "Port Policy Controller" could be the VirtualLink "PPC", but there's probably no way to tell for sure without comparing against one of the 20 series cards.
    Click to expand...
    Seems likely. PPC looks to be a generic microcontroller and part of the overall the architecture on dGPUs since Volta, but they probably built VirtualLink on top of it, hence the need for a "VirtualLink PPC firmware."
     
    Last edited:
    Found some more PPC2DISPLAY defines it seems like T234 might have had at one point which were later removed, and it also had some for PPC2PCIE, so it's definitely the USB port-related one and not the primitive processing one.
     
    Pokemaniac said:
    Games would have to be explicitly built to be able to not just crash when you took away a whole GPU and VRAM pool while they were running. It's a much greater imposition than how games deal with docking and undocking today by just tweaking some graphics settings in response (and even that isn't strictly required).
    Click to expand...
    But I could see it still to be possible to do something like this. Just not for the normal model. Something like a Pro Dock with more power, but protected by a mechanism to enable the switch. Thus you get say PS5 power in Pro Dock, but it always is a few seconds process to do the switch. This would never fly in the mass market version, but people who say, they never play undocked or just want brilliant graphics would accept it.
     
    invwar said:
    But I could see it still to be possible to do something like this. Just not for the normal model. Something like a Pro Dock with more power, but protected by a mechanism to enable the switch. Thus you get say PS5 power in Pro Dock, but it always is a few seconds process to do the switch. This would never fly in the mass market version, but people who say, they never play undocked or just want brilliant graphics would accept it.
    Click to expand...
    This isn’t ever going to happen, the amount of RnD cost on the return would be terrible. Ps5 and Xbox series x already selling at a loss, as if Nintendo will manufacture a ”pro dock” to compete.
     
    Giancarlo said:
    no,Super Mario Bros Wonder will keep Nintendo Switch healthy for a least half of 2024, then Nintendo drop another system seller(who could be it?) and Nintendo wont need to worry about launching it next hardware in 2024, they can do it in 2025/2026, dont forget Nintendo will focus on Switch until march 2025.
    Click to expand...
    Nah 2D Mario sells, but it also peaks very early.
     
    0
    You know, i'm in the hunt for an ultrawide monitor for both productivity and gaming, and the more i research, the less i want the Succ to have an OLED screen.

    Hell, it has made me wary of the Switch OLED, despite wanting one. The risk of burn in and the lack of a timefram in wich it can happen for sure means to me that it's a technology that isn't ready yet, and it could even be merely transitional before the real deal, whatever it is, actually happens.

    Give me a bright IPS screen with a nicely tuned HDR mode and call it a day. I don't want to be looking obsessively for burn in after 6 months, i want to enjoy my Succ without a care in the world.
     
    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