Monday, October 26, 2015

*Speculation* Nintendo NX GPU is AMD Arctic Islands based?

Here is the info from wccftech and just how likely this could actually be based on a few things we know about Nintendo and how they have made consoles in the past:

This new family of 16nm GPUs is informally code named “Arctic Islands”. The new family will be based on AMD’s 3rd generation GCN (Graphics Core Next) architecture. Interestingly — despite the existence of three+ different iterations of GCN namely GCN 1.0 ( HD 7000 series ) , GCN 1.1 ( R9 290 series) , GCN 1.2 ( R9 285 and 380) and Fiji ( R9 Fury X, R9 Fury and R9 Nano) which is based on an updated GCN 1.2 design — according to AMD there have only been two generations of GCN so far. With the next major iteration coming next year in the form of the “Arctic Islands” family of graphics chips.

This graphics architecture will power everything from next generation mobile and desktop GPUs to APUs and semi-custom chips. There’s also strong evidence to suggest that Nindendo’s next generation gaming device “Nintendo NX” – which is coming next year – will be powered by an AMD semi-custom SOC. Making it a very likely candidate to feature AMD’s 3rd generation “Arctic Islands” GCN architecture.

The flagship GPU in the family has been numerously purported as “Greenland“. Which is set to be AMD’s most powerful and most advanced graphics chip to date. In addition to being built on an advanced FinFET manufacturing process – likely 16FF+ at TSMC, the same for Nvidia’s Pascal – the Arctic Islands family will feature second generation High Bandwidth Memory Technology and as mentioned earlier an updated 3rd generation GCN architecture. These -16nm, HBM2 and GCN3- are going to be the three main performance and power efficiency driving features. Greenland specifically is rumored to feature up to 18 billion transistors and 32GB of second generation HBM with 1TB/s of memory bandwidth. Making it the largest ever graphics engine conceived by the company, at approximately twice the transistor count of AMD’s current flagship code named “Fiji” powering the Fury series of Radeon graphics cards.


alan bane said...

likely now 14nm AMD aparantly have said all 2016 onwards will be 14nm not 16nm both dies they have signed up for global foundrys and the other one are for 14nm...

14nm new cpu new gpu new ram,customized will be very efficient....

alan bane said...

whats more exiting to me is the talk of high performance and the cpu...x86 by default is known for poor draw calls and multithread rendering in software and the hardware level

AMD have said for years x86 has this issue of not being like custom console tech were the cpu and gpu run in harmony together and hit high draw calls and other performance advantages

AMD have said this many times before x86 and pc in general is very poor clock for clock meg for meg,they usedto say if we could make pc run like console and removed windows pc's would be amazing

now mantel was AMD saying to the industry enough is enough we need to modenize are api and software closer to the metal ...but thisstill leaves the HARDWARE issue with how efficient pc is x86 needs to be better in how everything connects together

NX will not only be a small tight apu with a closeto the metal api and low latency i think AMD and nintendo will make (nintendo will insist) that the cpu performs at CONSLOLE level of effectvness not a pc

the talk of polygons and cpu at the highest end,suggests to me its very efficent and much more CONSOLE like than x86...

if you took near high end pc parts and made them work in a super tight efficent way you get that big boost in power that gamecube andold school hardware did vs pc...

if nintendo go for above ps4 spec with HUGE HBM ram bandwidth with far above ps4 effectivness we shall have rewl performance at a low power level WATTS-HEAT

nintendo will demand low latency and a tight api and a big cpu to gpu performance GAMECUBE WII AND WIIU HAVE A VERY HIGH DRAW CALL CPU TO GPU STRENGTH

doing this with cutting edge chips will be amazing this is what i get from the HIghest end statment

its about DRAW CALLS AND HOW THE CPUS REAL PERFORMANCE IS UNLEASHED instead of the usual pc bottlenecks that even ps4 has

ps4 has rubbish cpu and rubbsh draw calls

its like there saying there putting console back were it used to be when released its at the cutting edge of performance in polygons effects etc

like consoles always used to be

Anonymous said...

@alan bane - while it is nice to see someone that actually knows that draw calls (and other API calls) exist, and all that, I think you are mixing things up.

The problem with API calls comes from using a 3D API (Application Programming Interface, meaning, a bunch of function calls you can use while programming your own code) with a programming model that doesn't fit how the hardware (GPU) actually works. That's the case of the ancient OpenGL model (and thus DirectX, which is a OpenGL copycat). When the API implementation/driver gets the calls, it has to do a lot of work (translation, checks, lots of things) prior to convert it to actual GPU instructions, thus causing A LOT of driver overhead.

While it IS true this has been one of the main reasons why traditionally consoles has been able to achieve more with less (in terms of HW performance), it has NOTHING to do with x86 architecture. It's a matter of graphics programming model, and the API choice: ie software. For PC, developers have to use the generalistic APIs that all manufacturers support (DirectX, OpenGL)...

PS4 does NOT have this problem: it uses its own propietary low-level API to program graphics. Neither will PCs anymore (at least a 95% of the problem) once they start using the new Mantle-based DirectX 12 and Vulkan APIs, based on a graphics programming model that actually fits (maps to) how a current GPU works...

See? Nothing to do with x86. Your x86 will do more than fine once your games are using an adequate graphics programming API (and thanks to AMD for forcing this change to the industry - it was about damn time...)