Nintendo Switch Tech Speculation discussion

[Rodéric]

Nintendo would do well with a Monster Hunter on release or in the release window.
Pokemon should help too, maybe some other 3DS franchises too.

 

[Rootax]

As a Home & Mobile console all-in-one.

Maybe they've changed their tune, because when the trailer was revealed, it was a described as the next home console, on Nintendo website. Like, it was not the 3ds successor. I still think it's the worst of both world. You can guess a poor battery life, and underpowered for a home console. And, again with the multiple pads needed... 249dollars/euro could be nice, but It would be without the gamepad pro (the only one with a true d-pad...). Aaah sorry I always say the same thing about the Switch. I'm disappointed (right now).

 

[Deleted member 13524]

At the very least I would think the Tegra chip in the Switch is matched to a 128 bit memory bus, processed on 16nm FinFet, and clocks north of 1Ghz. This is one powerful portable console.

That would require two LPDDR4 x64 packages, or 4 x32 ones. That's probably very expensive...

 

[Entropy]

That would require two LPDDR4 x64 packages, or 4 x32 ones. That's probably very expensive...

It's the standard configuration for Parker (nVidias 16nmFF TegraX1 successor).

 

[AlNom]

At the very least I would think the Tegra chip in the Switch is matched to a 128 bit memory bus, processed on 16nm FinFet, and clocks north of 1Ghz.

Why?

 

[Goodtwin]

Why?

Pure speculation on my part. Time will tell.

 

[AlNom]

Seems doubtful they'd blow past the leaked dev kit specs by doubling up on the bus width or increasing clock speeds, both of which have implications to die size, TDP & power consumption. Shifting from 20nm to 16nm is mostly a win for power consumption with miniscule density increase (short of massive re-architecting).

If they shaved off the A53's, that would further save die space.

 

[Goodtwin]

Seems doubtful they'd blow past the leaked dev kit specs by doubling up on the bus width or increasing clock speeds, both of which have implications to die size, TDP & power consumption. Shifting from 20nm to 16nm is mostly a win for power consumption with miniscule density increase (short of massive re-architecting).

If they shaved off the A53's, that would further save die space.

Is it impossible to match up a standard Tegra X1 to a 128 bit memory bus? Eurogamer has mentioned the dev kits had audible fan noise, the Shield console is nearly silent. The move to 16nm Finfet allows the Tegra Parker to boost clocks up to 1.5Ghz. The Nvidia CEO said they began working with Nintendo about two years ago, seems like it development of the custom Tegra for switch would line up nicely with the development of the Tegra Parker. Who knows, maybe it will indeed be a nearly bone stock Tegra X1, but I can be optimistic until we know for sure.

 

[Deleted member 13524]

Seems doubtful they'd blow past the leaked dev kit specs

Honestly, does anyone know if this leak appeared anywhere else other than a single GAF post from a totally random guy who basically dumped the specs of a Tegra X1?

AFAIK that leak is just as plausible as Goodtwin's speculation.

 

[Goodtwin]

Does anyone know if the Tegra X1 is bandwidth starved? On paper it looks bad, and you would think its a huge bottleneck for a 512Gflop GPU rendering games like Doom BFG and Trine 2 at 1080p, but it pulled those off on the shield console with fantastic results, and this is running on the Android API. I know the 2nd Gen Maxwell incorporated improved compression, but holy crap.

 

[AlNom]

Is it impossible to match up a standard Tegra X1 to a 128 bit memory bus?

The memory bus takes up chip perimeter, so it depends.

The move to 16nm Finfet allows the Tegra Parker to boost clocks up to 1.5Ghz.

For something that goes into a vehicle...

 

[Rootax]

Does anyone know if the Tegra X1 is bandwidth starved? On paper it looks bad, and you would think its a huge bottleneck for a 512Gflop GPU rendering games like Doom BFG and Trine 2 at 1080p, but it pulled those off on the shield console with fantastic results, and this is running on the Android API. I know the 2nd Gen Maxwell incorporated improved compression, but holy crap.

In some digital foundry video reviews, most games on tegra x1/nvidia shield android tv show slowdowns with alpha effects, a no good texture filtering. I guess it is bandwidth related ?

 

[mpg1]

I think the power of this thing is somewhat irrelevant. Just because it could possibly run AAA multiplats...how many people are going to use it for that?

It just needs to be able to run Nintendo games.

 

[Pixel]

You want to create a greate 1st impression, and put your best face forward in a reveal trailer and fact they showed two games that both exist on last gen consoles I think shows the nature of 3rd party support for the Switch. Basically they will get ports if the game is coming to PS3/Xbox360. And this makes sense economically until Switch gets a very large install base and proves its install base will buy all kinds of assortments of 3rd party games. We saw the lastgen remake Skyrim Special Edition and NBA2K17 (more than likely a port of the ps3/xbox360 version). Not looking good for ports unless they are also on PS3/Xbox360 at some point.

 

[bunnybug]

Honestly, does anyone know if this leak appeared anywhere else other than a single GAF post from a totally random guy who basically dumped the specs of a Tegra X1?

AFAIK that leak is just as plausible as Goodtwin's speculation.

nope the leak was orginaly from eurogamer. which there sources were pretty sure it's based off TX1. here is poster that works fro digital foundry just recently said what's he's hearing about switch.

http://www.neogaf.com/forum/showpost.php?p=224771160&postcount=316

 

[Goodtwin]

I don't think anyone disputes the idea that the TX1 was used in the developnent kits, but that doesn't mean it's final hardware. I remember Shin'en said that development kits became more capable as the release of Wii U drew near. I think TX1 gives us a ballpark idea of performance, but I fail to see how Nvidia spent hundreds of man hour years slightly modifying a TX1.

Sent from my SM-G360V using Tapatalk

 

[bunnybug]

I don't think anyone disputes the idea that the TX1 was used in the developnent kits, but that doesn't mean it's final hardware. I remember Shin'en said that development kits became more capable as the release of Wii U drew near. I think TX1 gives us a ballpark idea of performance, but I fail to see how Nvidia spent hundreds of man hour years slightly modifying a TX1.

Sent from my SM-G360V using Tapatalk

from nivida pr

But creating a device so fun required some serious engineering. The development encompassed 500 man-years of effort across every facet of creating a new gaming platform: algorithms, computer architecture, system design, system software, APIs, game engines and peripherals. They all had to be rethought and redesigned for Nintendo to deliver the best experience for gamers, whether they’re in the living room or on the move.

I'm think this, plus over exaggerated PR. I really don't see why developers would make up something like being pretty sure final hardware will be based on X1.

 

[AlNom]

I remember Shin'en said that development kits became more capable as the release of Wii U drew near.

I don't know the context, but that could be related to Nintendo exposing more features (API/SDK) rather than AMD spinning up new HW features every now and then.

 

[Rikimaru]

Early dev hardware almost always has lower clocks.

 

[mpg1]

nvidia ceo ALWAYS uses the "hundred man years of engineering" phrase when talking about products...nothing special that he he now uses it for Switch..