Xbox : What should MS do next? *spawn

Theoretically speaking, suppose that MS wants the slim to be the better alternative to base PS4 in every aspect. They decide that they have to match or exceed the performance.

They do the following.
Use a cut down version of the low end Polaris 11 as the GPU. Clock it at 1.2ghz.
Boost Cpu to 2.0ghz+.
From a technical perspective, sure, but from a business perspective the only way that this would make sense is to continue to have 2 platforms, so OG XBO would also be clocked to 1.2ghz to match the performance of the slim, esram is bound to the GPU clock, it doesn't have a separate clock so it's bandwidth will move accordingly.

But then you risk console longevity, console life time, what if everything starts breaking down then you have disaster on your hands -- the yellow ring of death comes back to haunt MS; unless all this time of course they've always intended for the XBO to be clocked higher - but that's a stupid dream to follow.
 
From a technical perspective, sure, but from a business perspective the only way that this would make sense is to continue to have 2 platforms, so OG XBO would also be clocked to 1.2ghz to match the performance of the slim, esram is bound to the GPU clock, it doesn't have a separate clock so it's bandwidth will move accordingly.

But then you risk console longevity, console life time, what if everything starts breaking down then you have disaster on your hands -- the yellow ring of death comes back to haunt MS; unless all this time of course they've always intended for the XBO to be clocked higher - but that's a stupid dream to follow.

I am completely confused why people keep referencing this supposed multiple platform issue. This is only 1 platform, Xbox. MS has been pretty clear since before the XBone launch that Xbox is an OS not a hardware configuration. Thus OG, improved slim, and Scorpio would all just be XBoxen. 1 unified platform. The only sticky wicket is really the esRAM, for which one supposes they had an answer prior to designing the ONE.
 
Right, my point was that sometimes you'll see higher res shadows on PS4, other times not - for the latter either it's a RAM consumption issue and/or PS4 isn't fill limited on that part. In the former case, if they're just doing 1/2 x 1/2 vs PS4, then it's 1/4 fillrate on there.

Of course, for 30fps games, you don't know whether it's a CPU issue (devs stuck on CPU particles/sorting etc) or if GPU-side is doing 30-50. idk

Case by case I guess?


LOL while this is totally off topic, I wanted to comment on personal bugaboo shadows and AA. Massive performance hit for minimal gain in quality for both. Of course, that is a bias from an old school gamer that spent most of the gaming life where such things didn't exist! This why I don't get people saying 4K 60FPS is out of reach for Scorpio, assuming ~6TF. Of course it is in reach, it is just a question of where you allocate the perf.
 
How do you get your 2$ figure?

You're talking about an area bigger than a Jaguar module, and that will in turn affect yields, require power, require cooling, require interconnects etc etc

How much do you think these chips are going to cost MS?


The marginal increase in die area is probably on the order of 10-15 mm^2 for 4 CU (giving 18 CU, 16 CU active for yields), which is ~5-7.5% of the die area of shrunken xbox APU assuming keeping the esRAM and the die shrink is 2X area shrink (which might be bigger). At 100% yield, each chip will cost < $25. apply %, gross up a bit for yields, viola $2. Remember, MS buys wafers not chips. AMD gets their money on royalties which likely wouldn't change.
 
did you not read what I wrote? You have to calculate that number first then apply a discount or as I worded it gross up to account for yields less than 100%
It's your assumptions of yields after that which are bizarre and even your unsourced cost per wafer for 14nm + package testing, or even the lack of consideration for what the increased power consumption (vs a straight shrink) would do to the overall product design and associated overhead costs.

So your extra $2 is simply a 92.5% yield? How do you figure that? Are you considering clock targets, TDP targets, or just the defect rate on the silicon itself?
 
Last edited:
Back
Top