• 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.

Nintendo Switch Dev Kit Stats Leaked? Cortex A57, 4GB RAM, 32GB Storage, Multi-Touch.

Status
Not open for further replies.

Mr Swine

Banned
If Switch uses ASTC for texture compression since it's in the TX1 hardware then devs can get in a whole lot more "higher" quality textures than what S3TC can do so while texture quality might suffer it won't be as bad as it would with S3TC
 
This was mentioned earlier as a joke, but what if the Switch Tegra SoC customization was scrapping the a53 cores and cramming WiiU Gekko CPU in their place? While this is wild speculation, that piece would solve the whole Switch puzzle for me, I know mixing Arm and PowerPC architectures seems weird, and I don't know if they could be used as extra gaming threads but intuitively at the very least it could manage OS and sound freeing the main 4 a57 from some burden plus.-

* Switch woulld get free GC, Wii and Wii U backwards compatibility
* Would explain a lot of weird Switch decisions (Clocks, handheld fan)
* It would kind of mirror Wii U mcm design
* Would explain those 500 man years fitting Gekko into Tegra X1

Now, while I thought this was scrapped when we got confirmation of Switch being Nvidia Tegra based, remember this Iwata quotes about NX.-

Absorbing the WiiU architecture adequately

* This quote is from Februry 2014, very close to the dates when the Nvidia deal was done, and would explain why the Switch SoC is basically a Maxwell based Tegra X1 used as a base to fit Gekko inside instead of waiting for later Nvidia designs.

Take this as a wild theory but it makes a ton of sense to me, I want to do a longer post about which I belive are the ideas behind Switch design but hate to write on tablet.

PD.- Kudos to M3d10n fot that UE4 finding.

What would Wii U/Wii/Gamecube BC give them when there's no optical drive? An easier time creating a Gamceube virtual console?
 

random25

Member
Nintendo games sure helped sell the WiiU. FYI the market has shown people want small bite sized games for on the go, and that is where phones have won.

Technically, they did. Just not on a large scale to make the final numbers respectable.

The recent significant hardware that got boost from Nintendo games is the 3DS.
 

z0m3le

Banned
This was mentioned earlier as a joke, but what if the Switch Tegra SoC customization was scrapping the a53 cores and cramming WiiU Gekko CPU in their place?

We heard reports of Wii U ports not working well at first. This wouldn't be the case if Espresso was on board. Switch is pretty much solved, the fan can probably be explained away with wanting to keep the device from throttling when docked in a warmer climate (100+ degree weather) remember it is going to be charging as well, so it will create extra heat, the fan might help it during these times.

The performance is good and there is a ton of room for overhead, upclocking the CPU in the future is very possible with this design.

Technically, they did. Just not on a large scale to make the final numbers respectable.

The recent significant hardware that got boost from Nintendo games is the 3DS.
Yep, Nintendo's hardware market this gen is around 75m, that isn't bad and if Switch can reach 50m while Nintendo expands into other branches of entertainment and chase the mobile market with their quality IPs, it can really replace the handheld market that they are substituting out for.
 
Got it. So for you every event to totally disconnected from every other event. Ok, and now back to the real world...



Nintendo is dealing with the reality that potential customers already own a smartphone that can play games, music, audiobooks, surf the web, chat with friends, browse Facebook, check their emails and bank account balances, and so on. Doing all that takes up a lot of time. Nintendo is hoping consumers have enough free time away from home after doing all that to justify spending $200-$300 more to fill it up.

Speaking for myself as a past mobile gamer, I have no interest in a dedicated mobile gaming device. I've fallen in love with audiobooks and podcasts. Those more than take up all the free time I have away from home. Even if I just had to play a game, there are more than enough good ones on my tablet that I buying a Switch still makes no sense.

In the end, with mobile devices there is a lot less time available for the Switch to fill up, which in turn makes it really hard to justify its price. Time is the variable people should focus in on, not games. If I spend an hour on NeoGaf, that is a hour I am not gaming. It doesn't matter how good the game is, an hour gaming is the same as an hour doing something else, and there is a finite number of free hours in a day.


The problem with this is that you are associating your interests with the interests of others. You need to take a step back and look at the big picture because not everyone is in your situation. And an hour of gaming isnt like an hour of doing something else. People like to play good games and when a game is good and accessible they play it more often.
 

ggx2ac

Member
This was mentioned earlier as a joke, but what if the Switch Tegra SoC customization was scrapping the a53 cores and cramming WiiU Gekko CPU in their place? While this is wild speculation, that piece would solve the whole Switch puzzle for me, I know mixing Arm and PowerPC architectures seems weird, and I don't know if they could be used as extra gaming threads but intuitively at the very least it could manage OS and sound freeing the main 4 a57 from some burden plus.-

* Switch woulld get free GC, Wii and Wii U backwards compatibility
* Would explain a lot of weird Switch decisions (Clocks, handheld fan)
* It would kind of mirror Wii U mcm design
* Would explain those 500 man years fitting Gekko into Tegra X1

Now, while I thought this was scrapped when we got confirmation of Switch being Nvidia Tegra based, remember this Iwata quotes about NX.-

Absorbing the WiiU architecture adequately

* This quote is from Februry 2014, very close to the dates when the Nvidia deal was done, and would explain why the Switch SoC is basically a Maxwell based Tegra X1 used as a base to fit Gekko inside instead of waiting for later Nvidia designs.

Take this as a wild theory but it makes a ton of sense to me, I want to do a longer post about which I belive are the ideas behind Switch design but hate to write on tablet.

PD.- Kudos to M3d10n fot that UE4 finding.

Would using Denver be a more easier solution? Having it deal with PPC since it was originally made to do both x86 and ARM whereas instead they could do ARM and PPC?
 

FyreWulff

Member
I think including the original chips of the previous consoles would make sense with an optical drive. But without an optical drive you might as well do all your BC via emulation.
 

Mokujin

Member
That would require IBM and AMD to be involved with the project, and nothing has ever indicated that they were.

I don't think Amd involvement would be required and Gekko design wouldn't need changes like they did from Wii to Wii U, I would even speculate that Nintendo owns WiiU Gekko "Espresso" design not needing much from IBM either.

But either way, just me speculating.

I think including the original chips of the previous consoles would make sense with an optical drive. But without an optical drive you might as well do all your BC via emulation.

Having the actual chip saves a ton of hours to the emulation department, plus you avoid any emulation performance issues (which may be a problem with Switch low cpu clockspeeds),and it would even bring maybe wii and wiiU titles into the mix.

What would Wii U/Wii/Gamecube BC give them when there's no optical drive? An easier time creating a Gamceube virtual console?

Of course I'm talking about GC (Wii,WiiU?) Virtual Console, not exactly backwards compatibiity, but having the actual chip in there makes it weird for me calling it emulation.
 

Peltz

Member
I think including the original chips of the previous consoles would make sense with an optical drive. But without an optical drive you might as well do all your BC via emulation.
Eh... I'm fine with a clean break in the library. No BC and I'm presonallu going with no VC. So I'd prefer they not include previous hardware to keep costs as low as possible.
 

LordRaptor

Member
Got it. So for you every event to totally disconnected from every other event. Ok, and now back to the real world...

"The real world" of "expert predictions" in gaming are the consensus of "core gamers" spectacularly failing to predict what actually ends up happening with unerring accuracy.
 

z0m3le

Banned
https://youtu.be/n50_WSl0MVY?t=2m30s Not sure if this was already talked about, but it is a pretty interesting note about NVN and the Switch's architecture from developers talking to Pachter.
“I’ve actually heard from developers that Nintendo is the easiest of the big three to develop for. “
It bodes really well, Nintendo has always seemed to have an issue with dev tools, and that quote should put everyone at ease, of course Shield wouldn't get the ports they do if Nvidia wasn't good with dev tools.
 

RootCause

Member
Speaking of emulation. Considering how much stronger this is, in both cpu, and gpu over the 3ds.

Do you guys think it can handle 3ds emulation?
I'd love for it to have bc for digital copies.
 

bomblord1

Banned
Speaking of emulation. Considering how much stronger this is, in both cpu, and gpu over the 3ds.

Do you guys think it can handle 3ds emulation?
I'd love for it to have bc for digital copies.

If it can handle GameCube emulation I don't see how it couldn't handle 3DS
 

Zedark

Member
https://youtu.be/n50_WSl0MVY?t=2m30s Not sure if this was already talked about, but it is a pretty interesting note about NVN and the Switch's architecture from developers talking to Pachter.

Fuck, Switch confirmed to be a hell to development for.

Seriously though, pretty strange way of putting things. It is easy to development for but it all depends on the system's capabilities. Wouldn't the first statement consider the power issue in its assessment?

Edit: on the other hand, one could always say that for examplethe Xbox 360 is easy to dev for, even though it is now underpowered I guess. It is probably more oriented at the dev tools aspect of it.
 

z0m3le

Banned
Fuck, Switch confirmed to be a hell to development for.

Seriously though, pretty strange way of putting things. It is easy to development for but it all depends on the system's capabilities. Wouldn't the first statement consider the power issue in its assessment?

Well we think we know the CPU performance, NVN could be a dream to develop on, but the CPU could still cause trouble, it's like FF15, pretty great except for what isn't.
 
https://youtu.be/n50_WSl0MVY?t=2m30s Not sure if this was already talked about, but it is a pretty interesting note about NVN and the Switch's architecture from developers talking to Pachter.
It bodes really well, Nintendo has always seemed to have an issue with dev tools, and that quote should put everyone at ease, of course Shield wouldn't get the ports they do if Nvidia wasn't good with dev tools.

I think this coupled with the fact that they are using very modern architecture (moreso than PS4/XB1 really) will be a big help in making up for the low clock rates. The CPU still sounds like an issue though we don't know the configuration for sure yet.
 

z0m3le

Banned
That's nice to know.
I hope Nintendo considers adding bc emulation for 3ds.

I don't know if Nintendo has ever done BC emulation. They don't mind letting you play your previous library through hardware emulation, but software emulation, which is what we are talking about here (Gamecube VC being the main factor of discussion) has always been a per game emulation, and Nintendo charges for the work that goes into emulating that game through software in a close to perfect state. However Wii U VC let you download games you digitally owned on Wii, at a discounted price. So maybe if a 3DS VC of LOZ:ALBW is released on Switch, we can get it for $1 or $2 instead of whatever they might charge (knowing Nintendo, I'd say $15)

I think this coupled with the fact that they are using very modern architecture (moreso than PS4/XB1 really) will be a big help in making up for the low clock rates. The CPU still sounds like an issue though we don't know the configuration for sure yet.

There is a lot of overhead for clocks, I wouldn't be surprised if developers complained enough that Nintendo bumps the CPU clock speeds, 1.2ghz to 1.5ghz should be fine even on a battery.
 

RootCause

Member
I don't know if Nintendo has ever done BC emulation. They don't mind letting you play your previous library through hardware emulation, but software emulation, which is what we are talking about here (Gamecube VC being the main factor of discussion) has always been a per game emulation, and Nintendo charges for the work that goes into emulating that game through software in a close to perfect state. However Wii U VC let you download games you digitally owned on Wii, at a discounted price. So maybe if a 3DS VC of LOZ:ALBW is released on Switch, we can get it for $1 or $2 instead of whatever they might charge (knowing Nintendo, I'd say $15)

I wouldn't mind paying an upgrade fee for 3ds tittles. I'll skip the likes of Mario Kart, and Smash since those will make their way to the switch.
 
https://youtu.be/n50_WSl0MVY?t=2m30s Not sure if this was already talked about, but it is a pretty interesting note about NVN and the Switch's architecture from developers talking to Pachter.
It bodes really well, Nintendo has always seemed to have an issue with dev tools, and that quote should put everyone at ease, of course Shield wouldn't get the ports they do if Nvidia wasn't good with dev tools.

Very interesting. Nice to hear some good news regarding the Switch after the last couple days.
 

Persona7

Banned
Does anyone think gamecube games will just be quick ports instead of emulation? Otherwise I am guessing simple emulation at original resolution but upscaled.
 

AlStrong

Member
I'm not expert either but I'm really not sure the costs would change all that much by being clocked higher... it's weird.

It'll have an effect on the yield of chips. Due to the chemical processes involved in fabbing chips, not all are made equal, which is why you basically end up with a range of speed grade bins ("binning") on PC. Essentially, there are a number of chips that operate at X frequency at Y voltage. Some chips can hit the target frequency at a lower voltage while others require a higher voltage. Naturally, the latter would lead to higher power consumption ala overclocking.

For consoles that operate at a fixed target, they basically have to throw away all the chips that can't operate at the target frequency at the target power consumption. So from there, one might see why clocks aren't necessarily pushing the envelope as they might on PC.

What would you say are the trade-offs/gains in the proposed scenario? Is there any reason to do such a thing? I ask because, in numerics, it brings us back up to what the devkits "could be" and what we would have thought but it seems like a weird way to go about it.

In terms of 2SMs/1GHz vs 3SMs, 768MHz? The chip would end up being larger (which can also have implications to the chip layout since it needs to be rectangular), but you can cut power a fair bit if that bumps them into the next step below in voltage.

Being larger does impact yield per wafer, but they'd have to weigh the two - A) die size & dies per wafer vs B) functional chips hitting target performance & power.
 
mobile games are great. But they can't replace tactile-button-input style gaming. You just can't do it with a touchscreen. They lack the feel and the latency, and when they do put buttons up, force you to obscure the screen with your hands.

This assumes that a mobile game needs to be the same type of game as a console game. Pokemon GO had no problem selling, and doesn't have the problems you mentioned. The fact remains that if you spend time even on non-gaming activities like Gaf, Facebook, Reddit, that is now competing for your tactile-button-input style gaming time.

So imagine that you spend most of your free time on a mobile phone on Facebook and playing Pokemon GO. Why does it matter to you that you can't play some button mashing game, even if you would enjoy it? Your time is already filled up. You could use that $200-$300 dollars you would spend on a Switch on other things...like for instance a better mobile phone which is a necessity while the Switch is not.

The problem with this is that you are associating your interests with the interests of others. You need to take a step back and look at the big picture because not everyone is in your situation. And an hour of gaming isnt like an hour of doing something else. People like to play good games and when a game is good and accessible they play it more often.

I am explicitly not saying I am like everyone else. That is why I listed a whole range of other activities that a person could do on a mobile device. It's the people who claim that only a certain type of game can satisfactorily pass a given person's time who are the ones assuming how others will spend their time.

The reality is that there used to be very few options for a person to enjoyably spend their time when stuck at the DMV, or on a bus, or waiting for their car to be repaired...and so on. Gaming was a major one of those very few options. Now we have many more things available to us. It is obvious that a good portion of those people who use to use gaming to pass their time have now found other mobile activities. Granted there will still be those who prefer traditional gaming, but those will be increasingly fewer people as more alternative options become available.

"The real world" of "expert predictions" in gaming are the consensus of "core gamers" spectacularly failing to predict what actually ends up happening with unerring accuracy.
Like I said. Got it. Why are you even commenting on Gaf then?

According to you it's pointless to discuss the future. There was no way to know that the XB1 was going to be hobbled by its initial designs. Who could have predicted that releasing Titanfall 2 in between Call of Duty and Battlefield would negatively affect sales? Hmmm...Maybe a No Man's Sky game that has failed to show any compelling gameplay just might not sell well. All those discussions...totally useless.
 

Schnozberry

Member
In terms of 2SMs/1GHz vs 3SMs, 768MHz? The chip would end up being larger (which can also have implications to the chip layout since it needs to be rectangular), but you can cut power a fair bit if that bumps them into the next step below in voltage.

Being larger does impact yield per wafer, but they'd have to weigh the two - A) die size & dies per wafer vs B) functional chips hitting target performance & power.

I believe the Tegra X1 was 121mm^2 on 20nm. I don't know exactly how large an additional SM would make the chip though.
 

z0m3le

Banned
Very interesting. Nice to hear some good news regarding the Switch after the last couple days.

The sky was falling because Switch fell ~40gflops outside our low estimates. It was a bit ridiculous TBH, the big surprise was the CPU clocks, but if there is issues there, I think Nintendo will upclock that part, we are talking maybe .5watt for 1.2ghz clocks, which might sound like a lot, but you'd be gaining quite a lot of performance, there might also be 4 A53 cores locked away from developers, Nintendo could open up 2 or 3, and they perform ~53% of A57 cores clock for clock iirc, so you'd get an extra core or two between all of that.
 

FyreWulff

Member
True. I wonder how they will get gamecube controller support for melee though. I can't see that not happening.

I think they made the the GameCube adapter USB on purpose so they'd only need to make a software shim for it for any future platforms. If Switch has USB ports for the dock then that's probably how you'll use GC controllers on it.
 

AlStrong

Member
I believe the Tegra X1 was 121mm^2 on 20nm. I don't know exactly how large an additional SM would make the chip though.

If they did add SMs, it'd be more likely for nVidia to add them in groups of two for layout symmetry and then disable one for yields (wafer defects, though I'm not sure that'd be a huge concern for a mature process). Given the general expected specs it's... kind of a non-insignificant thing to add, but who knows. The chip size ought to be small enough that they'd already produce a pretty high number of chips anyway (if on 20nm).

For the other two consoles, adding 2 extra CUs was relatively trivial next to their 350mm^2+ die sizes.
 

thefro

Member
The sky was falling because Switch fell ~40gflops outside our low estimates. It was a bit ridiculous TBH, the big surprise was the CPU clocks, but if there is issues there, I think Nintendo will upclock that part, we are talking maybe .5watt for 1.2ghz clocks, which might sound like a lot, but you'd be gaining quite a lot of performance, there might also be 4 A53 cores locked away from developers, Nintendo could open up 2 or 3, and they perform ~53% of A57 cores clock for clock iirc, so you'd get an extra core or two between all of that.

I can't imagine they're going to let the CPU be a huge bottleneck for this thing for no good reason, particularly since they should have the ability to upclock it just with an OS update.

I'd suspect there's either something we don't know which mitigates that or they ran into some major issues with something at the last minute.

Takeda even said they screwed up with the Wii U by making the processor too weak.
 
If they did add SMs, it'd be more likely for nVidia to add them in groups of two for layout symmetry (not unlike the disabled blocks in other consoles) and then disable one. Given the overall size it's... kind of a non-insignificant thing to add, but who knows.

Wouldn't that greatly increase the cost yet offer very little in the way of benefits? Are you basically saying that, since SMs are typically added to the die in groups of two, adding a single one would require extending the die size such that it could receive two?

I wonder how much power 4 SMs would be drawing at these clock speeds... could explain the fan I guess?

Nah that's far too hopeful of an outcome.

I can't imagine they're going to let the CPU be a huge bottleneck for this thing for no good reason, particularly since they should have the ability to upclock it just with an OS update.

I'd suspect there's either something we don't know which mitigates that or they ran into some major issues with something at the last minute.

Takeda even said they screwed up with the Wii U by making the processor too weak.

I think that was actually Miyamoto (or Miyamoto said the same thing additionally).
 

AlStrong

Member
Wouldn't that greatly increase the cost yet offer very little in the way of benefits? Are you basically saying that, since SMs are typically added to the die in groups of two, adding a single one would require extending the die size such that it could receive two?

They have to consider the layout of the chip. If you look at the Cell floor plan, it's all symmetric. If they add a third SM, the layout might end up with a chunk of dead space in order to make it rectangular anyway.

Ultimately, we don't know the chip layout, and as an SOC it's possible they could readily find something to fill the gap or they just space things out with some dead space (unused silicon area).

I'm not sure it'd be worth adding more when the chip is probably small to begin with. Too many unknowns.
 
So imagine that you spend most of your free time on a mobile phone on Facebook and playing Pokemon GO. Why does it matter to you that you can't play some button mashing game, even if you would enjoy it? Your time is already filled up. You could use that $200-$300 dollars you would spend on a Switch on other things...like for instance a better mobile phone which is a necessity while the Switch is not.
Well, this is really arguing towards the death of all dedicated gaming machines. I mean, I don't have a PS4One because I can get close enough experiences on the same machine I'm typing into right now. All-touch mobile experiences I don't find a satisfying enough alternative, though.
 
They have to consider the layout of the chip. If you look at the Cell floor plan, it's all symmetric. If they add a third SM, the layout might end up with a chunk of dead space in order to make it rectangular anyway.

Ultimately, we don't know the chip layout, and as an SOC it's possible they could readily find something to fill the gap or they just space things out with some dead space (unused silicon area).

I'm not sure it'd be worth adding more when the chip is probably small to begin with. Too many unknowns.

You'd think dead space would still be cheaper (and more energy efficient) than fabbing a whole extra module that wouldn't be used though, right? At the very least from a use of materials standpoint. You're absolutely right though that there are too many unkowns here to make these types of determinations.
 

Somnid

Member
To be fair, the phone has a radio, TONNES more storage, probably better screen and glass. Relatively more compact I imagine with force touch screens and the whatnot.

The difference is mostly the screen and camera. And when you get past the $400 Chinese value phones it's all margin on Apple/Samsung/Google's $600+ models. 32GB is standard, they don't have tonnes more and when they do the markup is extreme ($100+ for 32GB).
 

Hoo-doo

Banned
You guys just can't resist rounding the numbers upwards, upwards and upwards.

Go back and read the 'Nintendo Switch: Powered by Custom Nvidia Tegra Chip (Official)' thread again.
It's almost eerie to see people just take something and extrapolate and extrapolate upwards until it finally satisfies some vague subconscious demand for more processing power. It's really not worth the disappointment in the end.

An affordable and fully portable Wii-U with a semi-decent battery life is freaking great news. Why can't people be satisfied with that?
 

EDarkness

Member
You guys just can't resist rounding the numbers upwards, upwards and upwards.

Go back and read the 'Nintendo Switch: Powered by Custom Nvidia Tegra Chip (Official)' thread again.
It's almost eerie to see people just take something and extrapolate and extrapolate upwards until it finally satisfies some vague subconscious demand for more processing power. It's really not worth the disappointment in the end.

An affordable and fully portable Wii-U with a semi-decent battery life is freaking great news. Why can't people be satisfied with that?

Because that's not all the system is. It's a hybrid, and the console side is totally gimped. You're fine with a portable Wii U. Obviously a lot of othe people aren't.
 

Schnozberry

Member
If they did add SMs, it'd be more likely for nVidia to add them in groups of two for layout symmetry and then disable one for yields (wafer defects, though I'm not sure that'd be a huge concern for a mature process). Given the general expected specs it's... kind of a non-insignificant thing to add, but who knows. The chip size ought to be small enough that they'd already produce a pretty high number of chips anyway (if on 20nm).

For the other two consoles, adding 2 extra CUs was relatively trivial next to their 350mm^2+ die sizes.

For reference:

Tegra X1 (20nm + 256 Cuda Cores): 11mm x 11mm = 121mm^2

Apple A9X (16nm + Dual Core CPU + 12 Core GPU) = 147mm^2

Some guesstimates for adding additional memory controllers and CUDA Cores:

11 mm x (11 mm + 3 mm) = 154 mm^2

Or perhaps

11 mm x (11 mm + 4 mm) = 165 mm^2

Not really outside the realm of possibility. But it would potentially alter yields.
 

Mr Swine

Banned
Isn't there a slight possibility that Nintendo has locked away faster GPU speeds and that they will unlock it at a later date to get out more of Switch? Kinda like Sony did with PSP?
 
You guys just can't resist rounding the numbers upwards, upwards and upwards.

Go back and read the 'Nintendo Switch: Powered by Custom Nvidia Tegra Chip (Official)' thread again.
It's almost eerie to see people just take something and extrapolate and extrapolate upwards until it finally satisfies some vague subconscious demand for more processing power. It's really not worth the disappointment in the end.

An affordable and fully portable Wii-U with a semi-decent battery life is freaking great news. Why can't people be satisfied with that?

While I'd certainly be happy with a portable Wii U and 5-8 hours of battery life, I do think most people who have been paying attention are right to be surprised by these clock speeds. Especially on the CPU side.

That said, this is something that happens when we endlessly discuss specs in- more or less- an echo chamber of hope and positivity (which incidentally would be a fantastic name for a book) and judging by people's comparisons with the WUSTs (I wasn't here for those) it's probably best to just expect lower than the most seemingly reasonable expectations.

And then if somehow they go higher we can be pleasantly surprised.
 
Status
Not open for further replies.
Top Bottom