By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Nintendo - AMD Hinting that they will get the win for Nintendo's next handheld?

nomad on 13 July 2014
phaedruss said:
nomad said:

That's why it's still going to run hot. Tegra K1 TDP is supposedly 11W at max load, that's hot. It was at 28nm node. Let's assume by 2016-17, when 16nm finFET is ready, power consumption is reduced by ~50%, that would mean about 5.5W. 11W would be no problem for a actively cooled tab like Surface and 5W is not a problem for a iPad/SG tab size devices, but will still be a problem for handhelds like Vita/3DS.


Yea Tegra K1 would be too expensive anyways, but even shrinking down the current wii u soc would be something since it's 45 nm. I was reading an AMD presentation and it said that shrinking from 40 to 28 nm reduces power consumption to a third of what it previously was IIRC. Given that and a slight downclock you could get it pretty low I think.

The problem is diminising returns with silicon technology, the smaller they go the less they gain (relative to the previous node drop).



Around the Network
phaedruss on 13 July 2014
nomad said:
phaedruss said:
nomad said:

That's why it's still going to run hot. Tegra K1 TDP is supposedly 11W at max load, that's hot. It was at 28nm node. Let's assume by 2016-17, when 16nm finFET is ready, power consumption is reduced by ~50%, that would mean about 5.5W. 11W would be no problem for a actively cooled tab like Surface and 5W is not a problem for a iPad/SG tab size devices, but will still be a problem for handhelds like Vita/3DS.


Yea Tegra K1 would be too expensive anyways, but even shrinking down the current wii u soc would be something since it's 45 nm. I was reading an AMD presentation and it said that shrinking from 40 to 28 nm reduces power consumption to a third of what it previously was IIRC. Given that and a slight downclock you could get it pretty low I think.

The problem is diminising returns with silicon technology, the smaller they go the less they gain (relative to the previous node drop).

Sure, but there are still plenty of gains that can be made.



nomad on 13 July 2014
phaedruss said:

Sure, but there are still plenty of gains that can be made.

I don't think it's enough for a WiiU BC handheld to be feasible by 2016-17.



phaedruss on 13 July 2014
nomad said:
phaedruss said:

Sure, but there are still plenty of gains that can be made.

I don't think it's enough for a WiiU BC handheld to be feasible by 2016-17.


At a lower resolution I think so :)



fatslob-:O on 14 July 2014

If I were Nintendo I would actually work with Intel to design my next console rather than AMD. Intel has the advantage in process node which is important for getting a lower power consumption in small form factor devices like handhelds.

The new Nintendo handheld likely won't be anywhere near the WII U's raw power so it's preferable that they stick with a resolution of roughly around 480x270 to get the best image quality and it is a slight improvement over the 3DS's resolution.

As for backwards compatibility I wouldn't bother as almost no one other than Nintendo fans will miss being able to play WII U games. I would go with the x86 architecture for the handheld as IBM just keeps neglecting to improve the micro architecture. Mind you, I'm not kidding when I say that the WII U is using a very old micro architecture ever since the gamecube days. It's that exact same line of thought that got where AMD is today due to the fact that they thought that they could keep rehashing the K6 architecture.

I can see it now. A dual core clocked at 1 Ghz with a GT2 IGP clocked at 300mhz manufactured at 14nm would be the ideal combination. As for the memory configurations, a capacity of 2GB and a single channel DDR4 controller will perform admirably with a bandwidth of 25.6 GB/s will be more than enough to feed the entire SoC.



Around the Network
Captain_Tom on 14 July 2014
phaedruss said:

I saw this on NeoGAF and though I'd post it here as well to see what people think.

http://blogs.barrons.com/techtraderdaily/2014/06/10/amds-papermaster-moshkelani-talk-wearables-infrastructure-server-chips/

The quotes that are making people think this, "Another area of interest for semi-custom is handheld gaming, believe it or not. “Everyone thinks it is dead, but the [Nintendo] 3DS is still selling.” And this, "“Has to be at least $100 million annual revenue for us to go for it,” says Moshkelani. That’s a minimum, he explains, not a target."

A couple of points here.

1. How powerful could it be using AMD? Could they get it to around Wii U level? Yes, easily within a year or two.

2. What about backwards compatibility? It was my personal understanding that both the next handheld and console would be backwards compatible with the Wii U since they are using the Wii U as a starting point for their unified architecture.

Completily different architecture, but they could always just tape a 3DS to it like they usually do.





Captain_Tom on 14 July 2014
nomad said:
phaedruss said:

Sure, but there are still plenty of gains that can be made.

I don't think it's enough for a WiiU BC handheld to be feasible by 2016-17.

LOL it already practically is.  Look the Puma+ cores they are working on.  One of them will be as strong as a Wii U and consume 5w of power.



Captain_Tom on 14 July 2014
fatslob-:O said:
If I were Nintendo I would actually work with Intel to design my next console rather than AMD. Intel has the advantage in process node which is important for getting a lower power consumption in small form factor devices like handhelds.

The new Nintendo handheld likely won't be anywhere near the WII U's raw power so it's preferable that they stick with a resolution of roughly around 480x270 to get the best image quality and it is a slight improvement over the 3DS's resolution.

As for backwards compatibility I wouldn't bother as almost no one other than Nintendo fans will miss being able to play WII U games. I would go with the x86 architecture for the handheld as IBM just keeps neglecting to improve the micro architecture. Mind you, I'm not kidding when I say that the WII U is using a very old micro architecture ever since the gamecube days. It's that exact same line of thought that got where AMD is today due to the fact that they thought that they could keep rehashing the K6 architecture.

I can see it now. A dual core clocked at 1 Ghz with a GT2 IGP clocked at 300mhz manufactured at 14nm would be the ideal combination. As for the memory configurations, a capacity of 2GB and a single channel DDR4 controller will perform admirably with a bandwidth of 25.6 GB/s will be more than enough to feed the entire SoC.

My god I am glad you aren't making the decisions.  The Atom CPU's are utter garbage, and a screen of that resolution would be laughably bad.



Soundwave on 14 July 2014

It doesn't have to match the Wii U in "raw power", it just has to be able to approximate it so Nintendo can port some of their Wii U catalog/engines to it IMO (which you will have to pay for again, don't expect freebies here).

The first thing they can do is slash the resolution of Wii U-style games to 480p or 540p, that reduces the processing power required.

I think perhaps an ARM CPU core + perhaps an AMD GPU, if they can make one power efficient enough, might be the way Nintendo goes.

Perhaps the next Nintendo handheld will also be larger than people think ... it could be more along the lines of the size of an iPad mini IMO, which provides more room for heat dissipation from a hot running chip. 



arthurchan35 on 14 July 2014
Smear-Gel said:
What exactly is AMD?


Is this really something you dont know?



My Bravely Default CE is for sale, check it out: http://www.ebay.com/itm/Bravely-Default-Collectors-Edition-Nintendo-3DS-2014-/221490881912?pt=Video_Games_Games&hash=item3391e2a578