• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
  • Hi Guest. We've rebooted and consolidated our Communities section, so be sure to check it out and subscribe to some threads. Thanks!

News Rumor Hype Nintendo Switch 2 details, release date (January 2021), and new information on Zelda Breath of the Darkness

NutJobJim

Member
Apr 15, 2007
10,629
2,236
1,415
London UK
Don't own a Switch yet and all these rumours about a more powerful one drive me crazy. Have no idea if I should wait past black friday or not
I doubt a more powerful Switch is coming for at least one year minimum given that they just recently revised the original hardware to improve the battery and also launched Switch Lite.

Nintendo software and hardware generally retain a lot of their value so if, and that's a big IF, Nintendo launch a Switch Pro in a year or two you will easily be able to sell or trade your Switch in and pay a small amount to upgrade.

You may end up losing $100 or so but then you will have had over a year to play the Switch in that time. Hanging on forever doesn't make much sense unless you're especially broke.
 
  • Like
Reactions: PrimeX

LordOfChaos

Member
Mar 31, 2014
9,195
1,275
710
Carmel CPU is still like Denver?

Also, how high clocks should we expect with Tegra X1 at 7nm? 2 GHz CPU and 1 GHz GPU would be nuts, as long as games get patched (otherwise there's no point).

I think we would also need higher clocked DRAM to feed those hungry cores. Something like LPDDR4X/LPDDR5 maybe. 25.6GB/s ain't gonna cut it.

It seems similar, but Nvidia hasn't disclosed that many details to people not buying it. It's a 10 wide superscalar engine (!), which makes it the only ARM custom core to go wider than Apple (7-wide) and is in fact wider than anything we have on desktop. At the same time though it shares Denvers binary translation nature, the ARM core is only 2-wide, the VLIW engine is 10.


Now if you have the tools you could write code directly to the VLIW core too, but this would break ARM compatibility.

So it's the same sort of situation where a very ambitious core design is often limited by how good the binary translation engine is. Denver had clear problems there, Carmel must have made improvements but I don't know how much.

2GHz on 7nm should be doable, question is power draw on battery, at least docked it should be perfectly fine to hit that, maybe a little under undocked, though with the Switch they kept the CPU clock speed the same docked or un. I think it should at least be able to do current docked mode GPU power in undocked mode, which would be a nice 'automatic' win for the system that can just tell unpatched games it's running docked. And then with a new higher performance docked mode with GPU clock speed 800-1000Mhz.
 
Last edited:
Oct 16, 2017
254
363
270
Now we’re going somewhere. Nintendo has a patent for a “supplemental computing device” that can boost a gaming system using its own hardware. “The game console may couple to a supplemental computing device that includes resources that the game console may also utilize to increase the speed or quality of a user's gaming experience.”

If Nintendo has been playing with that idea in the labs, it’s not hard to imagine them releasing a Super Dock that could boost resolutions (even to 4K?) and frame rates over the current docked Switch’s capabilities.

This has lots of advantages. It doesn’t require a new console generation and all the uncertainty that can entail (Nintendo can’t be eager to risk another Wii U). If you want the power, you can pay for it; if not, just buy the basic Switch… and maybe pick up the Super Dock later. It’s pretty easy to market and up-sell.

I have no idea if this will happen, but it seems like Nintendo’s thought about it.
There is just one issue with "superdocks"...


We have SEGA and its history of the SEGA ICU tower of hardware. We are still making fun of it to this day. Having extra power from attachments sounded nice in theory, but that only ever worked for PC. In Consoles, that path leads to tears and disappointment.

The fact is such extra power split the player base, and that very few games would be made to take advantage of that extra power because there would be less people who own the extension. SEGA died a slow death because of this. You get more power, but the games suffer for it.
 
Last edited:
  • Thoughtful
Reactions: Negotiator

Negotiator

Member
Jun 28, 2011
4,318
884
720
It seems similar, but Nvidia hasn't disclosed that many details to people not buying it. It's a 10 wide superscalar engine (!), which makes it the only ARM custom core to go wider than Apple (7-wide) and is in fact wider than anything we have on desktop. At the same time though it shares Denvers binary translation nature, the ARM core is only 2-wide, the VLIW engine is 10.


Now if you have the tools you could write code directly to the VLIW core too, but this would break ARM compatibility.

So it's the same sort of situation where a very ambitious core design is often limited by how good the binary translation engine is. Denver had clear problems there, Carmel must have made improvements but I don't know how much.

2GHz on 7nm should be doable.
Yeah, VLIW is a pain in the ass, unless you're a top-notch programmer. It's the same reason ATi abandoned VLIW (TeraScale) in favor of RISC (GCN) and Intel abandoned Itanium in favor of x86-64. Hell, even nVidia had issues with VLIW (GeForce FX series).

Personally, I think VLIW will make a comeback in the future, when we will have AI-assisted compilers that will be able to offer human coding efficiency or even better. It will be necessary to extend Moore's law, but for now hardware complexity is preferable compared to software complexity.
 

iconmaster

Member
Jul 18, 2013
4,384
6,500
850
There is just one issue with "superdocks"...

We have SEGA and its history of the SEGA ICU tower of hardware. We are still making fun of it to this day. Having extra power from attachments sounded nice in theory, but that only ever worked for PC. In Consoles, that path leads to tears and disappointment.

The fact is such extra power split the player base, and that very few games would be made to take advantage of that extra power because there would be less people who own the extension. SEGA died a slow death because of this. You get more power, but the games suffer for it.
That’s worth bringing up, but consider the difference “more power” means to 2D versus 3D games. In 2D, games don’t readily scale with the hardware. Either you have multiple background layers, large sprites, transparency layers and more colors – or you don’t. In 3D games, this variability is already present within games. Resolutions are dynamic and frame rates fluctuate.

Trying to run a more advanced 2D game on weaker 2D hardware necessitates manually culling content, like Rare porting Donkey Kong Country to Game Boy. This isn’t the case with 3D.

(I do think Nintendo probably has SEGA foremost in mind as it considers its options, so that’s probably a good reason why it won’t happen – even if I think it’d play out differently.)

I think they’d want to wait for 4K-capable hardware and market it entirely as an add-on for those with 4K displays, even if different games used the power differently. It seems to me that’d be easily distinguished in the market and would only split the base as much as PS4 Pro and Xbox One X have split theirs.

Probably at that point you’re talking as Zannegan Zannegan suggests about a true console which only requires the Switch portable for purposes of saves and controller sync.
 
Last edited:
  • Like
Reactions: Zannegan

Negotiator

Member
Jun 28, 2011
4,318
884
720
There's another reason console hardware add-ons have been abandoned: modern console hardware is too power-hungry, therefore it needs active cooling.

Old school console chips like SuperFX/SVP or even the N64 expansion memory pack didn't consume a lot of watts.

Besides, it's technically impossible to offer a GDDR5/6 add-on, since those memory chips require to be soldered (hence why they're so fast). DIMM only supports DDRx, not GDDRx.

Imagine if PS4 Pro was a hardware add-on (assuming it's technically feasible)... you would have 2 Jet Engines vs only 1! Nobody in their right mind wants that in a living room, trust me. :)

TL;DR: times have changed. 100W+ consoles didn't exist back then.
 
  • Thoughtful
Reactions: iconmaster

iconmaster

Member
Jul 18, 2013
4,384
6,500
850
As I understand it, they basically can't have a supplemental computing device that boosts the Switch's base power because the USB port at the bottom of the Switch doesn't have the bandwidth to do that AND send video to the TV.
Btw, it doesn't solve every issue involved but my assumption was the dock would have its own video-out. The Switch would be feeding it game data only. This is likely slower than pulling data from the cartridge and undoubtedly slower than RAM access. (The Switch supposedly has 25.6 GB/s memory bandwidth and USB-C is more like 1.25 GB/s, but I have no idea if these numbers are apples and oranges.)

Edit: thanks for the Gbps correction N Negotiator
 
Last edited:
  • Like
Reactions: Zannegan

Negotiator

Member
Jun 28, 2011
4,318
884
720
Btw, it doesn't solve every issue involved but my assumption was the dock would have its own video-out. The Switch would be feeding it game data only. This is likely slower than pulling data from the cartridge and undoubtedly slower than RAM access. (The Switch supposedly has 25.6 GB/s memory bandwidth and USB-C is more like 10 GB/s, but I have no idea if these numbers are apples and oranges.)
10 Gbps, not GB/s. 1 Byte = 8 bits.

It can go up to 40 Gbps with Thunderbolt 3 (USB 4.0).

Switch has a USB-C interface that runs at 5 Gbps (1080p60 + USB Ethernet 100/1000 Mbps).