Middle Generation Console Upgrade Discussion [Scorpio, 4Pro]

<double post somehow?>

 

8 GB are enough.

 

How's that not swtiching the model? The existing game industry model is a static console design for 5+ years which is then completely replaced. The new model that may replace that is forwards compatible consoles released at a faster cycle with different concurrent SKUs and different pricepoints and performance levels. It's as complete a paradigm shift as we can get!

 

It's never enough.

 

If the internet is to be believed, and the 1.6-1.7 Ghz 8-core Jaguar is a bottleneck for the current systems, then I have no idea how MS will get away with still using the same low power CPU and terribly BW constrained DDR3 for Xbox One-Two.

It's almost a given that Xbox One-Two will pack a beefy Zen CPU and possibly GDDR5x or HBM2. Especially if MS intends for it to be them kickstarting a new generation with a new Xbox console (of which I do believe 2017 is too soon).

I think a better question would be what kind of TDP/price point are they targeting, and will it be an APU or split CPU-GPU modules on an interposer?

A 6 TFLOPs GPU with new memory and CPU, on a relatively new process 14nm, and without the benefit of binning chips, will mean this will be a very expensive console.

 

Why Xbox Scorpion?

What a coincidence...

Isca 2014: Insight into the MICROSOFT XBOX ONE Technology


Session 1: Scorpio (Snoopy COherent Research Processor with Interconnect Ordering)

 

I guess I disagree since it's unified memory and so far MS has reserved 3GB for the OS. Recommended specs for VR from Occulus/Vive call for GPU's with 4GB of RAM. I think 5GB (even 6 if they free another) is really tight for the CPU and GPU if they will really want to support VR on it.

I think going to 12/16 would go a long way.

 

Seems they may be going that way again, selling at a loss.

 

The more I think about it, I think with 12GB it would have to be some sort of hybrid solution like DDR4+HBM. I don't know if mixing densities works well and I don't see them going to a 384-bit bus.

I think it would be one of these three options:

1. 16GB GDDR5x @ 320-384 GB/sec
2. 2 8GB HBM2 stacks @ 400-512 GB/sec
3. 4 GB DDR4 + 1 8GB HBM2 Stack @ 256 GB/sec

I terms of performance per watt 2 would be most efficient and probably most likely to allow a 6 Tflop GPU in a console sized box. This would probably be Vega based and since it should launch after Vega, maybe HBM manufacturing will be cost effective by then.

I could see 1 as well, it would basically be a high clocked Polaris based GPU (36CU@1.2-1.3 GHz), but you're probably looking at a 175-200W console.

 

I think this is the bigger deal, but it's being ignored because it's more unknown at this time and just isn't as "sexy" to discuss as hardware.

I see the 6Tf XB rumors to very possibly be leaks of their transition console. I imagine it could be rather costly, $500 - $600, but is the Xbox branded console that you can buy if your W10 PC doesn't meet the minimum requirement of Xbox as a service/software. I see it as an optional "Console" for those who either would have to buy an entirely new PC anyway or who simply don't like messing around with PC gaming and upgrades. In my instance, I would have to choose to spend $600 on this "console", or build a gaming PC myself which would probably cost $1000-$1500. (Not that I couldn't build one more cheaply, but as we all know, once you start researching your gaming PC build it's only a few dollars here and a few dollars more here and you are quickly rationalizing why your $600 budget has become $800 and for another $200 you can build something really great!

Anyway, to the original point, it's the W10 features and integration that are the big deal, not that MS appears to be leapfrogging over Sony's PS4.5 idea and just announcing/releasing their next gen console early.

(or, it'll be their failure to do so)

 

Yeah I can't see how a console with the equivalent of a 1070 (a bit less actually) in perf could cost less than 500$ even in late 2017. That would be certainly interesting for the market though and judging by the robust cooling solution found on the X1 i'd say they could go for higher wattage without the risk of h/w failures.

 

If it's aimed as an XB1 replacement or new system... then yes.

If it's aimed as an additional platform sku, a high-end sku, for a smaller sector, those diehard gamers (i.e. those who would purchase a $150 joypad), ... then no.

My only question is: How can MS combat a possible 60 million PS4 install base by the time Scorpio launches, with the potential of Sony releasing Neo+ or even Neo having the capability of being upgradable? A very uphill battle...

 

Very true...............but I still think we can safely call this whole thing a controlled leak as well. Just look at all of this stuff. I mean just LOOK at it. After months of silence it's like all the damns and levees gave way. All the rivers and streams swelled and became inundated all at once. Just about every source is barking LOUDLY in his own twitter yard.

I've got plenty of salt, but the timing and the way this all went down is far too artificial and not in keeping with the normal flow of information that had (or more appropriately hadn't) been coming from Microsoft.

 

Because from what we know at this point, MS and Sony are going in two different directions. Scorpio would provide an improved gaming experience over the XB1, whereas Sony has issued a mandate to all developers that the gaming experience isn't allowed to be superior on the Neo from the PS4 and that the Neo is just there for VR and 4K streaming. Also, the point would be for Scorpio to offer a distinctly obvious superiority over the Neo and be a true next gen console, not a half-generation iterative step that is leashed to its existing 40 million user base.

 

Depends on how you define superior gaming experience. Neo will allow better frame rates and higher graphics settings, but no exclusive gameplay experiences. .

 

Yep, and I asked that question earlier - "Why Now?" and nobody seemed to want to engage in that discussion. One of the articles (Polygon?) attempts to give some motivation - that Sony was launching Neo earlier than MS had previously believed, so they decided to leak scorpio in response. However, there's an entire timeline that should be interesting to examine here. For example, as soon as E3 2013, MS knew it was screwed. With a console gamers didn't want and one that wouldn't be able to go toe to toe with the tech in the PS4. They could have started design work on scorpio as early as then. Would Sony? When did Sony, with its huge lead this generation, decide to start work on Neo and to leverage the user base this generation with a mid-cycle refresh to promote 4K streaming and VR?

 

My mistake, you're correct.

 

It would be extremely strange and unusual for MSFT to advertise anything that is not to be release in short order at E3.
Whereas the leak are true or not, I'm pov is that for now there are no solid evidences, we won't learn anything anytime soon.
It is the silly season, mid-gen edition, AMD has made quite intriguing announcement, there might be definitely something coming from MSFT or SOny, it is just so easy to spout any crap out and get the clicks