Hmmm One "bonus" to 6 discreet cores is it allows 1 core to emulate 1 thread, which may be necessary if they go OOOe but have a lower frequency. DDR3 would be a poor GPU memory, so maybe there will be a split configuration: 2GB DDR3 for the CPU and 2GB of memory for the GPU + interposer?
Anyways, I await a silicon footprint, especially for the GPU. That will tell us a lot--if MS is aiming for ~ performance / budget or if they are cutting corners. As it is over 50% of the market are on the HD consoles, it will be interesting if 2 players move to the low end and vacate that 50% to the competition.
Ps- says the longtime Nintendo fan/owner (NES, SNES, N64 import from Japan, GCN day 1, + mobiles... I even wanted a Virtual Boy!) who saw the Wii and jumped ship. I am ready to jump again. Whats that? Jump In?
I were to post something like it yesterday and pushed the "forget about it button". I start to wonder about the odd of an underwhelming (not Wii level though) system. We have pretty divergent statement coming from all parts:
SoC versus discrete CPU and GPU(s).
ARM vs IBM.
The only thing that somehow matches is time line Charlie spoke of Q1 2012 for tapeout (still he said a SoC) and that could match a CES announcement. MS would fine tune its PR accordingly to how well the chip is doing.
What I find bothering is that MS stated that selling the XBox as a set top box was important to them. I expect them to do it with the 360, it sounded a lot like this actually. I wondered about the odd of an ARM SoC to be there to ensure low power operation for the set top box and offer compatibility with their other embedded platforms. Still it's an extra cost while in any case MSFT is likely to launch something within 2 years. I also believe that to maximize Kinect success they need to provide the device to a broader audience, and they need to do it fast before competition (not from consoles manufacturers alone) catch-up. The fact that Nintendo is to launch fall 2012 is another factor, still I would say a lesser one but MSFT might have informations about the system we don't have.
So I wonder if MS pushed back its "set-top box" plan and anticipated its new system launch plan.
Every body speak about the MSFT leaky nature but I dare to disagree when they want they can keep a secret they do, project Natal is a proof of it. So who knows?
MSFT may be able to launch something fall 2012, most will say it's a tight schedule, I'll agree but not impossible especially if MSFT chose an well known "ironed out" process. Say 45nm for the CPU and 40nm for the GPU.
It's a bit disheartening when you consider the requirement for a system that act as a set top box from scratch: lesser form factor, lesser noise, standard HDD, low power consumption (at least during non game activities). Add the cost of Kinect and a controller as well as the bad economic climate that dictates a reasonable launch price (I would say 400 the whole thing) and one could start to expect a Xbox 540 instead of Xbox 720 if you see what I mean. The thing would still beat the WiiU clean but less an improvement than I were expecting after 7 years (and I expected les than most already.
I would expect pretty tiny chips (tinier than Xenon and Xenos at launch).
For the CPU (assuming power ISA and using Xenon as a basis) something akin a jump from Cortex A8 to A9 instead of A8 to A15. It has to be tiny and fit 6 cores + caches.
From the GPU in raw power something between 5670 and 5770 updated in term of feature could do the trick. Discarding architectural improvement something offering x3 or x4 the raw throughput of Xenos (so below 1TFLOPS).
The ARM/PowerPC makes things a bit more confused. I feel like MSFT could be cutting corner. THey may have consider ARM but moving time line may have push them to stick to stick to POWERPC for compatibilty with the 360 and fast ports to the new system (it may have been different if they decide(d) to push the 360 two more years as a set top box and jump to a clean system afterward).
As for your PS, I don't know I was expecting nothing crazy (in form factor, power consumption) but really well though out, a clean start could really help achieve something good. I would bet on something ARM based (cpu and gpu). ARM seems pretty open as a company, Micro Applied has been working on a ARMv8 implementation for a while whereas the thing (ARMv8) has only been made public some weeks ago. I expect them to allow the same with their GPU technology if one were interested. May be one could access to the thing likely to come after Mali-T658 and grow based on it. Mali are deferred renderer which is great for a SoC as it alleviates many bandwidth concerns.
I don't know may be Sony will go with ARM + PowerVR, I don't know they are really spread thin at the moment with plenty opposing forces they have no grasp on. And I still don't see direction in the way they act.
May be someone like Samsung could enter the game for a reasonable price depend on the kind of performances they have to match/exceed.
