Nintendo Switch Tech Speculation discussion

I know many nintendo fans want a portable xbox1. Is it technically possible with 5~10W power dissipation in 16nm finfet process?

 

u w0t.

Define "portable". XO Slim is still about 75W at load.

A 1050Ti (2*3SMs) strapped to 8xA72 (octopus, get it!? ) would have been neat for a straight up home console though.

 

I don't think so, even if you try to compete with xbox1 fp32 throughput with fp16. But IMO it should be possible to do significantly better than the rumored X1 at ~300MHz. My guess is Nintendo wants something at around the speed of 216-2017ish phones, so that some of their games can be released on other mobile platforms without major changes to their assets.

 

A portable console which can run xbox1 games at 720p or 540p, with a screen > 5 inch. This may be what fans want.

 

No. Not this year, not next year.

 

Yea, Xbox One S is using the Finfet 16nm process and still pulls 50-75 watts, so for a product that is going to pull 5-15 watts, yep, not happening. I remember a guy over at Anandtech who was said to be a developer, apparently vetted by the forums, and he was saying right after the Switch reveal that developers would likely target 504P, the minimum resolution Nintendo allows. At the time I was thinking yea right, but now that we know clock speeds, I am thinking this lines up pretty good. Any developer who does try to port modern AAA games to Switch will likely have to reduce the resolution to this extent to get the performance they want, and bump up to 720p in docked mode. Who knows, perhaps we will see checkerboard rendering techniques used extensively on Switch. We saw with Fast Racing Neo what a 720p game using checkerboard rendering looks like. On the big screen the image quality was obviously not a true 720p image, but on a 6 inch screen it was fine.

 

Probably not doable.

Mind you, if you downclock the GPU, you're also affecting the front end of the rendering pipe i.e. geometry setup. So games targeting the main console wouldn't just be a simple pixel throughput issue.

For Switch, presumably devs are targeting mobile from the get-go, so increasing clocks is less of a problem.

 

There is no fan in that thing, just passive cooling.

 

really ? that's impressive if true

I see your reasoning but bulldozer family to zen family is a huge sea change that you wouldn't get from zen to zen.

 

The Tegra's are just too power hungry at reference clocks for mobile, but they make for great PR at Nvidia's events. I wonder if going the Vita route with reference ARM and PowerVR would have got them more perf/W.

Vita successor vs. Switch would have been something to see...if only Vita wasn't a market failure.

 

GCN2 geometry throughput is pretty low compared to Nvidia Maxwell/Pascal. Clocking it low shouldn't be a problem in this sense. Also Xbox One has only 16 ROPs and Maxwell tiled rasterizer is pretty efficient.

The perf problems are most likely seen in pixel and especially in compute shaders. GCN is pretty good in compute and async compute is widely used in console games to bring GCN occupancy higher than PC.

 

nV handling the software is probably more valuable to Nintendo than...

... gold pressed latinum or something.

 

NV's software department has been off their game lately though, at least when it comes to drivers, they're been utter crap version after version last months, introducing new major bugs and fixes to them not working

 

Right, just talking about targeting the full speed vs low clock mode. It doesn't scale exactly with just pixel throughput if you're assuming a certain setup rate, so targeting mobile then going up is less problematic than the other way around, which is what it would be for an Xbox portable that would have to be clocked low in order to hit mobile power consumption.

e.g.

Games targeting 853MHz suddenly running on 300MHz setup rate. Changing resolution only goes so far, so "Xbox portable" games wouldn't scale down so easily with a naive resolution reduction. (Context was creating xbox handheld, off topic)

Vs

Games targeting mobile from start then running at 2.5x clock. Setup rate doesn't matter since it's based around the low end already.


Edit: ugh,

Person meant Nintendo handheld with power of Xbox One, not a handheld version of Xbox One.

Anyways.

 

Maybe all the best were working on the Switch

 

That tends to happen when you aggressively push release schedule beyond ridiculous levels, it's like they are in a bragging contest.

 

If I look at the the system as a handheld I like the performance bracket Nintendo aimed at. Nintendo does not seem interesting in sinking incredible amounts of money to develop modern AAA games. In that context PS360 level of performance is a sweet spot, modern hardware power is tapped more easily and extra RAM will do wonder.

Now for the hardware per self, I think it's going to be a plain Tegra X1, it is just nonsensical to go through the pain and money to design something else that would be that close. As for the pixel fill figure I guess it is simple matter of pixel/cycle instead of Gpixel/s /human mistake.

Honestly I dislike the whole thing. The performance level for the handheld is great but it is useless as home console, the improvements over the Wii U are to be marginal. The system is not ready for 4k, even for simple game for a system releasing in 2017 it is bad.
Nintendo is imho doing a mistake, the concept is pretty good: same games on TV and handheld, cartridge based. Now execution is awful they should have design two separate systems.
The Switch is going to be mostly as a mobile device but will have to deal with lots of overheads which are likely to affect negatively its price hence its reach in the market (it needs a cooler, it is complicated compared to handhelds like the 2DS, the PSP, the PS Vita and many android based console wannabee).
On the other hand view as a home console the deal is sour, especially for the fan that bought a Wii U.

As I see things Nintendo needed consistent CPU performance and a x4 ratio between the handheld performances and its home counter part.
I think Nintendo should have targeted qHD (540*960) for the handheld and 1080p for the home console. The home console could have stream games to the handheld with x4 supersampling, the main purpose would have been battery saving.
The home console would have muscle to push 1080p graphics for simpler games and using a clever upscaling trick to output in 4k. As for the quality think of playing PS360 games which are available on PC... on a low end gaming pc so with so extra bells and whistles on and an overall smoother experience.

As for the IP by self, I'm not super fond on Nintendo choices either. The CPU is a little dated and not the best matched for a low power device. I also think that Nintendo could have pushed the number of cores furthers (6 or 8).
For the GPUs, Nvidia does great things but they are also among the sharpest business persons around. Having them designing something custom at a sane price is a pretty awesome achievement but 2 SOCs...nah...

PowerVR does great mobile GPUs that also back a bunch but ultimately my belief is that Nintendo should have done with a pure ARM, hence dealing with a single partners (as they are doing with Nvidia) AND a partner that offer a complete range of up-to-date solutions on multiple process.
The last rendition of 28nm are great and widely available, ARM last IPs are available on such process: Nintendo should have produced two SOCs.

 

I suspect a big part of the negative response to the tech specs of the Switch comes down to poor messaging on Nintendo's part. Everything about the console suggests a unit designed as a 3DS replacement that would also work on a TV but the messaging has become that it's a 'hybrid' device that is both a home console and a handheld. As a home console it's fatally weak if you expect it to fight it out with PS4/XB1 but as handheld it's significantly more powerful than it's nearest rival (Vita) and has a much simpler TV experience.

If they had been clearer from the start that their design goal was a handheld that can also be plugged into a tv rather than a console that can also be brought out of the home we'd see far less disappointment.

I still think this is an exciting handheld which I'll probably use docked more often than not but I can throw in a bag for flights. As for IQ Nintendo have proven past masters at tailoring their art style to the technical limitations of their hardware so that they deliver a great looking image so I'm not that worried there.

 

I think they got a deal from Nvidia + special talk from Nvidia awesome sale persons.
I think that the Tegra X1 falls a little short as home console and is over specced for a handheld BUT it can be used for both, comes with great tool an software, Nvidia reputation, etc.

On the other hand it is free money to Nvidia the SOCs is available, they just have to place orders...no work serious work (sale persons aside) involved.

No it comes from a poor concept, no hardware the size of a handheld can double as a mobile device and stationary device. Laptops can, laptops are bigger then there is the matter of costs.

 

microsoft and google also has been chasing this concept for ages and none of them have succeeded. Microsoft do seems to have an edge on this tho (with their W10continuum and x86onARM thingy).

athough in this case, nintendo have distinc t advatage: focused as video game device