mutantsushi said:
JustBeingReal said:
TBH NX could have a base dock, with that GPU inside and a smaller AMD APU in the tablet
Hell they could just stick with cheaper 28nm tech for the console part and then use 14nm for the handheld to house more power efficient tech for on the go.
|
The concept of having separate GPUs or even CPUs, each to be used in different mode: mobile/console, is insanely wasteful. It means a large chunk of hardware budget is power that is not being used in any given mode. Theoretically one could access both, but there will be performance penalties vs. unified solution, and development will be much more difficult. The obvious solution is simply to have unified, advanced GPU/CPU which can down-clock and de-activate cores that aren't needed, e.g. in mobile mode which will target a lower resolution frame buffer, while console mode will output 1080p with fully activated GPU/CPU. That just results in two quality settings to tweak, but using same architecture, vs. totally distinct architectures rather hard to develop for. (either totally distinct GPU/CPUs, or heterogenous architecture combining 2 CPUs and 2 GPUs)
When I read the EG story, I read them speculating that Nintendo will use Tegra 2, and have a product quite distinct from Tegra 1/ Shield. Not sure why that possibility is discounted completely in this discussion.
|
It's no more wasteful than having a console sitting at home that you can't play when you're out of the house, actually being able to use your console's tech when you're out of the house is actually reducing the waste of having said console, because you can play games on it now, anywhere you have access to an internet connection.
The hardware would always be accessible provided you have an internet connection and the game you're playing requires it, not all games do, but why limit your platform's capabilities when the tech to include it is actually really cheap now?
The tech to match a modern console, but in a handheld/tablet form factor doesn't exist yet, but the tech to offload non-latency sensitive tasks to an external processing unit does, it's called the internet.
BTW if you're deactivating parts of the CPU and GPU in the tablet because you can't power it for long on the move then you're wasting it more than if you were offloading tasks to a console unit at home through the internet. Same deal for reducing clockspeeds, you're wasting potential compute time, may as well just put the processing performance offsite, where it's available when needed to share workloads with the tablet device.