Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]

Nintendo doesn't have to go high end and directly compete with the beasts. The suggestion of a quad core Zen APU has more to do with timing and off-the-shelf availability. The amount of SPs may come down to what quad-Zen APUs are spec'ed with, though there is no reason Nintendo couldn't ask for more if they want them. The big thing is that Nintendo would be going x86 and AMD GCN. There would also be AMD True audio. This continual downward spiral of trivializing 3rd party developers has to come to some kind of end if Nintendo wants to stay relevant to the big players, and adopting the now standard console ISAs will guarantee them much more business, even with just multiplatform games. Going "big x86 APU" also gives them a chance to be the gold standard in home game consoles for a few years, especially if they let's say require all games to run at 1080p native.

My suggestion for HBM has to do with Nintendo's love for eDRAM, lower power and for simplified board complexity.. I don't know what HBM's Latency is, but it would offer them lots of bandwidth and plenty of memory, even if just for VRAM, but without the heat and power needs involved with GDDR5. The interposer could function as an MCM as well for other components like a security processor, streaming suite, or even for just DDR3L main memory.

Sure these are somewhat flights of fancy, but the appearance of AMD Zen next year could be a sign of what's to come in the console world, and the reemergence of Nintendo as a real powerhouse in the console world.

 

Next year Zen is 8 core only. 4 core is 2017.

 

Nintendo always disappoints you more like. Yes the WiiU is weak on paper but look at what they get out of it with games such as Mario Kart 8 and Smash Bros U, games that also run at 60fps while the competition argue over what game can stay at 30fps the longest without dropping to 25fps.

The WiiU didn't fail because it had 160 shaders, it failed because it was priced out of their usual customer bases price range, advertised incredibly poorly and had a terrible first year filled with delays and no big name titles from their staple of IP. The whole idea of two screens was also incredibly hard to get across without using the console and even then it wasn't ideal because you have to keep moving your eyes between both screens killing immersion.

I expect NX to be another tiny, low wattage console with an ARM CPU, 6GB's of DDR3 RAM and a GPU on par with the one in Xbox One. This way third party developers will be able to down port the big name games to it while Nintendo can keep it around the $249 mark while still making a profit from day one.

The people expecting Nintendo to build a powerful console that is 2x or more as powerful as PS4 are living in fantasy land imo.

 

My point is whatever low expectations people trot out technically for nintendo, they usually far exceed in the opposite direction. For example given your prediction of one teraflop, given Nintendo's history one might expect something with 300 GFlops, 1GB RAM, 2 generations old ARM CPU.

Sooner or later I think they have to change course though, and it would seem past due to be this time IMO.

 

It's rather simple. If Nintendo wants to open the door for easy porting of current gen third party console titles, they will need to target a performance level that facilitates that.
Question is - is that actually one of their design goals with the NX?

 

Rumours suggest yes. History suggests .... erm ...

 

Well, the WiiU is certainly capable of accepting ports from the 360 and PS3. Even the very first launch batch was OK, and something like Bayonetta which came a bit later was reportedly very good. Which shows that being able to accept ports is no guarantee whatsoever that they will actually be made. Of course the barrier of entry could be lowered further by a more closely matching architecture, and more performance headroom.

The NX looks set to launch 3-4 years after the introduction of the PS4 and XBOne. If the WiiU had launched in 2010 instead, would it have changed its fate if there had been no PS4 or XBOne on the horizon? The situation for the NX is different, that much is clear.

However, I have seen no mention from anyone representing Nintendo that the failure of the WiiU in the marketplace is due to lack of third party AAA games. What they have identified as issues to adress has been for instance too high cost, inability to communicate the benefits of the controller, disconnect from the Wii audience,... but no blaming the lack of Bioshock infinite, or Deadpool, or GTA V for instance. And it does have the Need for Speed, Assassins Creeds, Deus X, Batman, Mass Effect 3, Watch Dogs... Honestly, I doubt any lack of performance for ports has much to do with the current WiiU situation. If Nintendo actually regards ports of multi platform titles to be vital the solution is simple enough, particularly when Iwata said this at the annual shareholders meeting:

If they start from 0, they aren't likely to have hardware backwards compatibility a design parameter which allows Nintendo to, for instance, use a design that is arbitrarily close to the PS4. If they want. However it is not clear to me from any statement or even rumour that this is a priority for them.

 

Here's my prediction for NX. It's a platform that will consist of multiple devices but with different power levels.

2016: Launch of handheld first
CPU: Quad ARM a 53 or Dual Core A57
GPU: 2 GCN CU's @ 500 MHz (128 Gigaglop), 8 MB SRAM on die
RAM: 3GB of RAM (1 GB for OS)
Process/Die: 28nm, die less than 100mm
Screen: 5" 480p
Flash storage: 4GB - SD exandable
Price: $179

2017: Home version
CPU: Quad A57 or hopefully A72
GPU: 6 CU's @ 800 MHz (~600 Gigaflops), 32 MB on die SRAM
RAM: 4 GB (1GB OS) (More ram for higher quality assets)
Process/Die: 28nm, die less than 200mm
Flash storage only: 64-128 GB (like WiiU, not an SSD)
Price: $199

OS is shared between the systems (hence the same OS RAM allocation) and so is the game interface, implying identical control scheme for both systems. No dual screen handheld, looks like WiiU game pad.

Other important feature of NX. Crossbuy, buy the game once and play anywhere on any NX device with synchronized cloud saves.

2018-19: Premium handheld using a shrunk home console SOC in a larger tablet form (8" 720+ screen)

 

AMD pretty much ditched ARM designs, no? Surely 20nm ought to be fine for a low power target & volume/low cost by next year.

 

http://www.amd.com/en-us/products/server/opteron-a-series

 

Fair enough. Wasn't thinking they'd spend time on it other than servers though.

 

Kabini looks a lot like that and is above 100mm2. Another concern is power efficiency and both the CPU and GPU are at a significant disadvantage against all their competitors, performances are going to be bad relatively to the silicon invested and the power dissipated. It will not reach 500MHz without cooling. Anandtech stated that Kabini tablets part default to 225MHz to remain within TDP limit.
The set-up for the screen sounds right. The OS is heavy.
I think the big issue is the price, looking at tablets in that price range does the product any good. I expect that by the time that ship Amazon will have newer Kindle 6 on sale for 99$ and a kid version too. I also suspect that in 2016 the basic for cheap phones and tablets alike is going to be 2GB of RAM.
Though whatever the operating frequencies are and ultimately the performance, it would be a massive jump from the 3DS new or not.
I would not buy that hardware at that price though.

 

An interesting question is whether Sony would still exist as a console manufacturer if it had only been able to ship with 4GB RAM in the PS4. For Nintendo, if they want 3rd party games, then that needs to be 8GB RAM (EA/ubisoft/activision will not want to create/QA a separate set of assets for the NX).

 

Yes they would. Developers would have to make different, or more drastic, compromises. Almost all software development is a case of deciding which compromises you're going to make.

 

Anyone posted this yet ? -

True, I never in a million years expected a 176GFLOP GPU from a console released in late 2012 and just a year before consoles with 1.3 and 1.8TFLOP GPU's. It really is remarkable what Nintendo can achieve though while using older tech from Mario Galaxy on a 12FLOP GPU and 88MB's of RAM to MK8 on WiiU. The mind truly boggles when you think what they could achieve visually with even an XB1 level GPU with more CPU power and a lot more RAM.

I do think NX will be a turning point for Nintendo hardware wise though and whilst I don't expect them to release a console more powerful than PS4 it isn't out of the question for them to release a GPU in the middle of XB1 and PS4 power wise. The fact that Miyamoto is out of the way for the consoles design and also that the American arm of the company seem to be designing the chipset gives me a lot of extra hope.

 

Would the witcher 3 devs have compromised? Fallout 4? Mass effect 4? Would UC4 have dropped the endless driving sequence? What would horizons be if the open world was so restricted? similarly, reducing the RAM for the OS substantially - how would that have worked out? [cutting the video record thing/screenshots/background downloads?] Given the problems elsewhere within Sony, if the PS4 had struggled, I do wonder what the 2015 console landscape would have looked like... (for all the complaints about the XB1, I think the PS4 was incredibly fortunate). But, it's kindof an interesting question - as there's no way to know.

Anyway, for "this-gen" I would say the defining characteristic is RAM - and developers have done wonders with it.

For next-gen - I'm struggling to see any technological improvement that's likely to have a similar impact. I guess "a bit faster hardware, and a bit prettier 1080P"?

 

Yes. Yes. Yes.

Why would they need too? Uncharted 2 had the endless train sequence with an tiny fraction of the available RAM of the 4Gb your propose. There are a metric ton of ways you can reduce your RAM footprint and you may be surprised how large data tables becomes for inconsequential things like objections in the environment. Havok eats RAM for breakfast especially when you're allowing a large number of objections to react not just to the protagonist, but against other objects as well as we've seen in the demo.

Read any post by Laa-Yosh and VFX_Veteran and they'll pinpoint a ton of compromises where image quality is compromised, beit model complexity, animation and lighting and general shadowing and lighting of environments which is still a huge con and limited. Hopefully the latter will make a huge leap forward. It's always getting better but always seem shy of what folks would ideally like.

 

Nintendo will not be purchasing off-the-shelf chips. They will want their own design, and if the Zen core is ready, AMD can spin it into any configurations they want.

Not that I think Nintendo will necessarily go for Zen.

 

I love speculation regarding new hardware, especially Nintendo's
I still think NX won't be a single device, but a whole family, starting with a handheld with a 4,5-5,0" screen (720p max.) and around WiiU performance (4 CPU cores, 150-200GFLOPS GPU, 2GB RAM). This should be doable in 2016 for ~150€. Games will be cross-releases with WiiU games (Zelda will be a launch title), they will also share account system.
The home console will come a year later (2017) with 8 CPU cores (or threads, depends on the architecture), 1024 GCN shader units and 8GB of HBM2.
Both will use flash cards like the 3DS which will allow for a smaller, cheaper packaging of the devices.

 

IBM have produced the first 7nm chips on SiGe.

Ars have a nice article about it. Still challenges with cost of lithography to be met for it to launch before 2018.

Still, made me fantasize about 20Gflop+ consoles being viable in the next cycle.

http://arstechnica.co.uk/gadgets/2015/07/ibm-unveils-industrys-first-7nm-chip-moving-beyond-silicon/