Nintendo Updated Sales Figures for Switch / 3ds

[GoodBytes]

11 hours ago, Trixanity said:

That's why I said tweak the power on the existing design. It ain't a pretty way to do it and you'll lose some in the process but it'd probably still be superior to the X1. 

Well the X1 in the Switch, even when docked, is seriously downclocked chip. It goes at max speed 1GHz instead of the 1.9GHz.

I guess the heatsink/fan can't handle the extra heat without being very noisy due to the small form factor. I mean using the Switch it is clear that the fan doesn't come close to ever spinning at its full speed when docked.

 

11 hours ago, Trixanity said:

Or perhaps there is something in the supposed automotive design that makes it inherently unsuitable for the job that I simply can't see at a glance.

Another possible factor, is the other chip might be harder to manufacture and can't have the output that Nintendo needs.

 

11 hours ago, Trixanity said:

I also looked at product timelines and Nintendo may just have missed Parker while Nvidia is probably also giving the automotive industry dibs due to business strategy. Disappointing nevertheless. It simply isn't a good chip. Especially not for a game console with long lifespan.

Most likely.

[VanayadGaming]

I've always been tempted for a handheld console, BUT I don't see any good games on them

[Trixanity]

1 hour ago, Eroda said:

from what i heard the chip in the switch doesnt use anything except the 4 a57 cores the other a53 ones are unused so they could have set it up similar  but i doubt it would have come in at 299 price point nintendo wanted it probably would have been 399 and performance would have not been significantly higher in portable mode. what gets me is why the switch isnt just locked to 720p no AA or AF or HDR etc and use the dock to upscale to 1080P and add 6XAA/AF and HDR etc etc.

 

the swithc is just lacking it doesnt even match current gen consoles  which 3rd party devs are already wanting to leave behind. eventually itll just not be worth the effort to make games on the switch and it will just be nintendo making games again. so here is hoping that in a couple years nintendo makes a proper home console that is backwards compatible.

Correct. Nvidia gave them a special chip with A53 physically disabled. However from what I know the A53 cores don't work properly even on the original chip. It's supposed to handle scheduling on chip and be inaccessible by the OS so it's hard to tell. Some say the interconnect they made is broken. Who knows? 

 

It was very underpowered but pretty inefficient to begin with. The only thing it has going for it is decent GPU power for its size and design. As we all know A57 cores are bad and the 20nm process is bad. It's dragging the chip down. I haven't heard of it getting a die shrink and I don't think they have any intention of doing it. I don't think Nintendo is willing to pay for that.

 

I don't think there are many ideal substitutes so Nintendo will have to live with a console that's technically EoL at release and they went through with it so I suspect they're fine with a console that can chug along. 

 

I mean what are the alternatives? AMD had nothing in that power range and kinda still don't. I don't think Raven Ridge could do it in portable mode if we assume they'd make a mid-cycle refresh like their competitors.

 

Qualcomm? That'd be a Snapdragon 820. I don't think it'd have the raw GPU power but would otherwise have been a decent choice all things considered. There are no other candidate to make what Nintendo needs.

1 hour ago, GoodBytes said:

Well the X1 in the Switch, even when docked, is seriously downclocked chip. It goes at max speed 1GHz instead of the 1.9GHz.

I guess the heatsink/fan can't handle the extra heat without being very noisy due to the small form factor. I mean using the Switch it is clear that the fan doesn't come close to ever spinning at its full speed when docked.

 

Another possible factor, is the other chip might be harder to manufacture and can't have the output that Nintendo needs.

 

Most likely.

Nintendo was stuck between a rock and a hard place. That's for sure. The A57 cores and 20nm process are difficult to work with. They run really hot (Snapdragon 810 anyone?). That's why even Parker would have been nice if Nintendo held out for that. 16nm and some heavy lifting Denver cores with a Pascal GPU. It'd run a lot faster with less throttling. But I think it might be slightly bigger based on the side by side pictures I've seen of boards but not massively so (as is the case with the new Xavier SoC). So I don't think the price difference is as big as one would initially assume.

 

I wonder what a Snapdragon could do in that form factor. From what I can tell they're very efficient but still behind in performance (speaking of GPU). 

 

Anyway that's all hearsay. Fact is Nintendo is sitting on a console that'll struggle to keep up due the position the chip was in even at the launch of the chip (it was around 18 months old at the time of launch of the Switch).

[ExplosiveSloths]

Had a switch since day one, haven't used it but hey, it'll have games I want to play in the future.

 

Also addicted to mobile doom and sky-rim, also maybe some mario odyssey.

 

[GoodBytes]

29 minutes ago, Trixanity said:

Qualcomm? That'd be a Snapdragon 820. I don't think it'd have the raw GPU power but would otherwise have been a decent choice all things considered. There are no other candidate to make what Nintendo needs.

It's not just power. It only support OpenES, not the full OpenGL. And the GPU is a full (architecture wise) GeForce, so no hacks or tricks to do to get anything working on it. Plus all the dev tool form Nvidia are accessible. So it is very dev friendly, which is a huge selling point, especially after the WiiU.

 

Essentially, the only thing that devs needs to be concern about is ARM CPU strength and weaknesses in doing certain tasks over x86 (which are all super well documented thanks to smartphones and non-Windows tablet market), and huge success of the architecture. The rest is all about making highly optimized shaders as every drop of performance counts on the system, and do lower level of details on polygons models. I mean there is more to be concerned about such as AI as the CPU isn't as fast either, and the system has less RAM, and so better streaming texture based on position algorithms are needed (and this might mean an engine change or rework to help the Switch limitations)... But those aren't tall orders if a game is in the works with Switch in mind. It's not a "we need to make a special version from scratch for the console", which was a huge issue with the Wii and WiiU. The only upside the WiiU had was that it was an XBox 360 just with a faster GPU and more RAM... But the GPU was funky, GPU was split rendering 2 screens, preventing on taking advantage of the full power of the GPU unless they do like Nintendo's very own Donkey Kong Country Tropical Freeze, where they output nothing on the tablet screen, and the poor sales of the console didn't help push anything. Now the Switch has the market share to start justifying third party support seriously. So in the next 1-2 years we should start seeing better ports on the system, especially if past version of the game on the system sold pretty well.

 

[linustouchtips]

switch better than wii u confirmed but my goodness the software sales

[Trixanity]

2 hours ago, GoodBytes said:

It's not just power. It only support OpenES, not the full OpenGL. And the GPU is a full (architecture wise) GeForce, so no hacks or tricks to do to get anything working on it. Plus all the dev tool form Nvidia are accessible. So it is very dev friendly, which is a huge selling point, especially after the WiiU.

 

Essentially, the only thing that devs needs to be concern about is ARM CPU strength and weaknesses in doing certain tasks over x86 (which are all super well documented thanks to smartphones and non-Windows tablet market), and huge success of the architecture.

It supports Vulkan though. I don't think standard OpenGL would fly on a console. In fact, I assume Nintendo don't use standard APIs in the same way previous gen consoles used non-standard APIs. I know Nvidia uses their full architectures on SoCs as I mentioned it previously. X1 and Parker uses Maxwell and Pascal respectively with Xavier using Volta. The biggest selling point of Nvidia is definitely their dev resources. They pretty much brute force their way into game development through the sheer amount of aid they can lend. It definitely helps their cause.

 

As familiar as ARM is, the A57 cores are pretty bad. They aren't really faster than many of the Krait cores of 2013 and the like. The CPU simply isn't very fast and it isn't clocked very high due to process node and poorly behaving core design. It'll bottleneck the system in the same way that the Jaguar cores do on PS4 Pro and Xbox One X. Those need to be clocked pretty high to feed their respective GPUs and try to get around that fact through tricks used in the custom implementation to augment the relatively weak cores. Granted, the Switch throttles pretty heavily due to the poor thermal management which is no fault of the device design (at least not necessarily). They'd need to employ some beefy cooling to avoid throttling and that's not even mentioning power throttling.

 

Other than that, I agree that devs can achieve miracles through perseverance and creativity. I'm not saying the system is doomed as such but what I am saying is that there just isn't a lot of room to work with. Arguably we've been there before with a system on its last leg where out of nowhere a dev team can pull even more out of it. I just don't think it's wise to launch a system that's dated on release. We kinda saw that already with the current gen of consoles (PS4 and Xbox One) where we need a mid-cycle refresh to keep up. The previous gen was in some respects ahead of their time or at the very least the cutting edge coupled with a highly controlled developer environment and tools to extract everything possible from them. Again, the argument could be made that it's a Nintendo console and it doesn't matter but then we're in the exact same boat as always with third party developers abandoning the console because they can't make what they want and there is no reason to go into the footsteps of first party games. No one can compete with Mario and Zelda on their own turf so that would be suicide and pointless.

[Derangel]

16 hours ago, Trixanity said:

That's why I said tweak the power on the existing design. It ain't a pretty way to do it and you'll lose some in the process but it'd probably still be superior to the X1. 

 

The Xavier SoC (as an example) is in the 30W range. For maximum battery life you may want to cut back on that.

 

And honestly, looking at the Parker chip (which would be the one that Nintendo could have opted for) makes me scratch my head. It wouldn't have been perfect but it actually fits decently with what Nintendo needed. It can be configured as low as 7.5W which is definitely in line with a battery powered device. Devs would probably have been annoyed with the separate clusters of A57 and Denver cores though.

 

All I'm seeing here is Nvidia being inflexible and not having a lot to offer at the time and Nintendo asking for the cheapest chip possible to keep their margins.

 

Or perhaps there is something in the supposed automotive design that makes it inherently unsuitable for the job that I simply can't see at a glance.

 

Edit: from what I can tell Parker isn't significantly bigger than X1. However Xavier is three times bigger if I'm reading this correctly. Probably due to double the CUDA cores and double the high performance cores with corresponding cache sizes. So that pretty much throws it out the window. Nintendo isn't going to pay for that.

 

I also looked at product timelines and Nintendo may just have missed Parker while Nvidia is probably also giving the automotive industry dibs due to business strategy. Disappointing nevertheless. It simply isn't a good chip. Especially not for a game console with long lifespan.

Nintendo wouldn't even have been able to launch until this year, maybe, if they waited on the X2 to be ready and to have enough supply. Their supply issues might have even been worse then they already were since the X1 was not what held them back. On top of that there is zero chance they would have been able to hit their price target if they went with a more expensive chip. Nintendo was not in a position that allowed them to wait another 12-24 months to launch a new system.

[Trixanity]

1 hour ago, Derangel said:

Nintendo wouldn't even have been able to launch until this year, maybe, if they waited on the X2 to be ready and to have enough supply. Their supply issues might have even been worse then they already were since the X1 was not what held them back. On top of that there is zero chance they would have been able to hit their price target if they went with a more expensive chip. Nintendo was not in a position that allowed them to wait another 12-24 months to launch a new system.

Source for it being more expensive? I can't see much of a size difference so it would come down to the foundry asking for more. Supply, sure. That could be true. The wait you imply would only be true in the case of supply issues.

[Derangel]

1 minute ago, Trixanity said:

Source for it being more expensive? I can't see much of a size difference so it would come down to the foundry asking for more. Supply, sure. That could be true. The wait you imply would only be true in the case of supply issues.

Simple logic? Newer chip, likely with less readily available supply= more expensive. The wait I imply would be true regardless. The X2 was not even rolling out to companies by the time the Switch launched last year. Consoles need to have specs finalized anywhere from 6 months to a year before launch. Since Nintendo clearly didn't want to launch in the middle of the busy game season that would have pushed the launch until March of this year. Which would also mean delaying Zelda, and other titles, another year and having to stick with the Wii U for another fiscal year which would have been disastrous. As nice as it would have been, there was just no way that Nintendo could have waited on the X2 to be ready for them.

[GoodBytes]

21 minutes ago, Trixanity said:

Source for it being more expensive? I can't see much of a size difference so it would come down to the foundry asking for more. Supply, sure. That could be true. The wait you imply would only be true in the case of supply issues.

It does.  Sorry, you'll need to believe me on this. I mean... it's expensive in the sense that you'll have no problem buying yourself, if you could... but it is too pricey for selling a console at 300$.

[Yoinkerman]

72 million 3ds holy crap

[Eroda]

i cant see ports of the big titles running on the switch but if the swithc sells enough maybe it will get didicated 3rd party games that they will then port to X1 and PS4, i was going to post this on reddit shrotly but 

 

 

 

this is the type of game needing to be on switch something witht he flavour of battlefield  BC2/BF3 but more simple arena gameplay basically rush with low vehicle count and player count but more TF2 Combat i think this would be a winner  in todays market

[GoodBytes]

4 minutes ago, Eroda said:

i cant see ports of the big titles running on the switch but if the swithc sells enough maybe it will get didicated 3rd party games that they will then port to X1 and PS4, i was going to post this on reddit shrotly but 

 

They can.. they perfectly can. This is not a 3DS where massive changes are needed.

They just need to optimize the game engine texture loader for lower memory bandwidth of the system (which they did in the past for older consoles, so it is tech that can be ported in the newer game engines if it has not been done already). Then the rest is cutting out polygons, textures resolutions, and reduce object on the screen... like remove 3D models of people in heavy crowed and switch to the good old 2D sprites trick, and things like that. All stuff where the game port needs to be done during the game development, and not after at the last minute. The issue with this, is us, the consumer who will need to wait for in the works games.

[Eroda]

it gets to the point where the game looks and might even play like nothing that the original title plays like at which point i might as well play the original title