Xbox Series S [XBSS] [Release November 10 2020]

depends on die size , i think MS can make 3 xss apus per xsx apu. So they shouldn't have a real issue meeting demand.
How quickly can Microsoft task TSMC to make more Series S than Series X (or vice-versa) for it to meaningfully mean adjusted console volumes manufactured, shipped and entering retail?
 
That was an interesting interview, but this bit stood out. The reason they kept support for the base XBO-S instead of the XBO-X.
  • XBO-X is too expensive to justify being at a price point that represents the bottom of the console stack.
    • This was expected.
  • They want to have the previous gen feature set disappear as quickly as possible without abandoning people that cannot immediately get a XBS-S/X. Continuing support for the XBO-X would have meant that the previous gen. limitations would potentially linger far longer into the current generation as it's a more powerful and capable machine.
    • Makes sense, I didn't think about it from this angle before.
Regards,
SB

So basically, planned obsolescence? But in a way the majority are actually cool with? Sounds good to me :cool:

Also, have to think that production on 16nm of the Jaguar cores will only get tougher as time goes on since the major foundries don't really have the space for it anymore, or will be phasing it out in relation to this class of CPUs most likely. Smaller foundries would probably still manufacture them but not in anywhere near enough volume to get good bulk pricing.

That makes the second point all the clearer.
 
How quickly can Microsoft task TSMC to make more Series S than Series X (or vice-versa) for it to meaningfully mean adjusted console volumes manufactured, shipped and entering retail?

I don't think you switch on a dime like that. MS has forcasts of what they are expecting to sell and have most likely purchased fab capacity based off that. They would have to go and purchase more capacity from TSMC.
 
I don't think you switch on a dime like that. MS has forcasts of what they are expecting to sell and have most likely purchased fab capacity based off that. They would have to go and purchase more capacity from TSMC.
You can't, it's likely around two months before any change will result manufacture output, maybe longer. Sony seem to be on a weekly-shipping schedule and Microsoft are on a slower cycle - based on re-stocking frequency during the pre-order period,
 
Whats the reason for SS struggling so bad on WD Legion? Worse than X One X

XOne 720P-900P
XSS 864P-1080P
X1X 1440P-1800P
XSX 1440P-2106P

Unfortunate it is saddled with slow loading but X1X is performing well in current showdowns


Edit: It's ray tracing. From the video author:


Pinned by ElAnalistaDeBits
ElAnalistaDeBits
2 hours ago
With Assassin's Creed Valhalla we had a good job overall in Series S | X, but with Watch Dogs we have the opposite result. There is still a long way to go to optimize Ray Tracing on consoles, as the result in Watch Dogs: Legion is a bit disappointing. In Series X we would have the equivalent of Medium / Low of RTX on PC, while in Series S it would be below the low mode. Specifically, Series S is seriously affected by this Ray Tracing, since it causes the game to go to a too low resolution. Series X has more detailed shadows and a better draw distance than the other versions. Without a doubt I am much happier with the results of AC: Valhalla and I think that giving the option to deactivate the RT would have been a good idea (especially in S Series).
 
You can't, it's likely around two months before any change will result manufacture output, maybe longer. Sony seem to be on a weekly-shipping schedule and Microsoft are on a slower cycle - based on re-stocking frequency during the pre-order period,

I doubt they will need to change manufacturing out put for one or the other. Even if the series s takes longer to sell out they will still sell them.

I'm not sure how MS will handle the new systems coming in. I think there will be another drop around thanksgiving from MS as employees were told to expect the employee store restocked around then
 
I'm still curious if MS will nail down the production allocation balance that best suits demand for both models, knowing these two systems are sharing production budgets for manufacturing, assembly, shipping & handling etc. But so far it seems they have a pretty good handling on that.

It was my biggest worry for a while actually because I got over the whole "it'll hold next-gen back" nonsense rather quickly (and with WD:Legion results on Series S I don't think that concern really flies anymore), but matching production budgets to both systems in line with expected levels of demand, that could always be a challenge.


I'd worry more if they can sell them. Series S was available to preorder for months in some Eu countries like Germany and even UK where Xbox demand is a bit softer. Whereas XSX sold out quickly everywhere. And from what we know allocations went more heavily to XSX.
 
I'd worry more if they can sell them. Series S was available to preorder for months in some Eu countries like Germany and even UK where Xbox demand is a bit softer. Whereas XSX sold out quickly everywhere. And from what we know allocations went more heavily to XSX.

Considering it seems like we are going back into lock downs in europe and the usa i think everything is going to sell and with more lock downs it means more unemployment and more people unsure about future finances which means less spending on entertainment.

A $300 console can become very popular esp if MS uses it as bundles for other device deals with other companies. Buy a new Samsung tv get a free xss and the like.

Whats the reason for SS struggling so bad on WD Legion? Worse than X One X

XOne 720P-900P
XSS 864P-1080P
X1X 1440P-1800P
XSX 1440P-2106P

Unfortunate it is saddled with slow loading but X1X is performing well in current showdowns


Edit: It's ray tracing. From the video author:


Pinned by ElAnalistaDeBits

Well the XSS has Ray tracing so that could be the issue. Could also just not be as optimized. I believe XSS was a profile in dev kits and not the actual hardware. Hopefully a patch comes that prings it up to a locked 1080p
 
I think 2 days after launch might be a tiny bit too early to make that call.

Well let me ask you a question. With the outlook of future node processes how low in price do you think all 3 consoles will drop over their life span ?

Remember ps4 launched at $400 and sells for $300 right now 7 years after launch. Xbox one launched at $500 with kinect and now without an optical drive sells for $250 before the digital one was discontinued and the optical version $300. Ps4 pro launched in 2016 and is still $400 and xbox one x never officially dropped under $500 outside of one time sales bringing it to $400 I believe.

So if we do the normal $100 price drop the Xbox series X will drop to $400 over its life , PS5 will drop to $300/$400 and Xbox series S will drop to $200.

Now look at what AMD with navi - navi 2 same node
RX6800_11.jpg




So what will a 2024 5nm navi 3 or 4 bring in terms of performance ? What ram types and speeds will be available ? How much will Ray tracing improve each generation of navi ? Will infinity cache prove itself and get better ?


So if you have a choice between selling a 5nm 2020 designed console for $100 less than launch or just phasing it out and making a 2024 design that vastly out performs it ?

That to me is the future and we saw it starting to take shape with last gen.


edit * to be clear i do think the xss and xsx will hit 5nm before going away , i think they will wait till navi 4 most likely in 2024 to replace the consoles out right , with 5nm first being introduced in 2022/3 for the console chips. I also think the xss may stay in another form factor
 
Last edited:
edit * to be clear i do think the xss and xsx will hit 5nm before going away , i think they will wait till navi 4 most likely in 2024 to replace the consoles out right , with 5nm first being introduced in 2022/3 for the console chips. I also think the xss may stay in another form factor

IMO, this entirely depends on the cost associated with getting 7 nm Anaconda/Lockhart onto 5 nm (it may not be a straight shrink which would lead to it being almost as costly as designing a new chip) and the cost associated with designing a new chip.

If MS and AMD streamline their partnership, then the cost delta between those could be small enough that MS may decide that it's a better choice for various reasons to just design a new chip. While multi-platform titles likely wouldn't take full advantage other than better resolution/speed/effect quality, 1st party titles would have the option to push the hardware which could give it a more noticeable competitive edge compared to just moving Anaconda/Lockhart to 5 nm.

Regards,
SB
 
Well let me ask you a question. With the outlook of future node processes how low in price do you think all 3 consoles will drop over their life span ?

edit * to be clear i do think the xss and xsx will hit 5nm before going away , i think they will wait till navi 4 most likely in 2024 to replace the consoles out right , with 5nm first being introduced in 2022/3 for the console chips. I also think the xss may stay in another form factor

I did wonder if they would be doing yearly upgrades to the chips they are using for cloud, going from RDNA2 and Zen2 to RDNA3 and zen3 etc, not for a performance increase but for the improved performance per watt, would also increase their experience with RDNA. It Would make it much easier to do a mid gen refresh on short order too, would probably only need 18 months lead time
 
IMO, this entirely depends on the cost associated with getting 7 nm Anaconda/Lockhart onto 5 nm (it may not be a straight shrink which would lead to it being almost as costly as designing a new chip) and the cost associated with designing a new chip.

If MS and AMD streamline their partnership, then the cost delta between those could be small enough that MS may decide that it's a better choice for various reasons to just design a new chip. While multi-platform titles likely wouldn't take full advantage other than better resolution/speed/effect quality, 1st party titles would have the option to push the hardware which could give it a more noticeable competitive edge compared to just moving Anaconda/Lockhart to 5 nm.

Regards,
SB

The goal of XSS is to get as small of a form factor as possible. They may have a 4tflop navi 3 or 4 based version of the XSS. The goal as I've said before is to get down to fire stick / tv size for their console. They want something that will play games at 1080p and then also be used for xcloud. A navi 3 version or ideally a navi 4 on 5nm could accomplish this while at the same time performing better than the xss. like i said if infinity cache works well for amd this could be a way to drive costs down even further with newer ram and an even smaller bus.

I did wonder if they would be doing yearly upgrades to the chips they are using for cloud, going from RDNA2 and Zen2 to RDNA3 and zen3 etc, not for a performance increase but for the improved performance per watt, would also increase their experience with RDNA. It Would make it much easier to do a mid gen refresh on short order too, would probably only need 18 months lead time
Not yearly. I think they will upate every 4 years in cadence with a console release. Xcloud is going xbox series x in 2021 so i expect an refresh of xbox series x with new navi and zen in 2024 and 2025 for xcloud
 
The goal of XSS is to get as small of a form factor as possible. They may have a 4tflop navi 3 or 4 based version of the XSS. The goal as I've said before is to get down to fire stick / tv size for their console. They want something that will play games at 1080p and then also be used for xcloud. A navi 3 version or ideally a navi 4 on 5nm could accomplish this while at the same time performing better than the xss. like i said if infinity cache works well for amd this could be a way to drive costs down even further with newer ram and an even smaller bus.

The problem is that 5 nm is unlikely to get there for them. 5 nm Lockhart is likely to cost roughly the same as 7 nm Lockhart. Memory also isn't progressing in cost nor size nearly as fast as they would need to even think about Lockhart in a fire stick form factor. Hell, I'm doubtful they could even get 5 nm Lockhart + everything else it needs into an Apple TV sized box, although it's potentially possible if they go with some expensive cooling and an external power brick. Part of the reason the XBS-S is as large as it is, is due to the relatively simplistic yet large SOC cooling solution that they use. But that means it's also cheaper. For XBS-S, cost was far more important that trying to get it to be as small as they could make it.

The end result would be a smaller Lockhart based machine at 5 nm that would potentially cost a fair bit more than the XBS-S.

Regards,
SB
 
IMO, this entirely depends on the cost associated with getting 7 nm Anaconda/Lockhart onto 5 nm (it may not be a straight shrink which would lead to it being almost as costly as designing a new chip) and the cost associated with designing a new chip.
Nor should we expect SSD cost/gb to drop as quickly as has happened with platter HDDs previous generations. I think we're looking at a much slower pace of significant cost-reducing revisions.
 
Nor should we expect SSD cost/gb to drop as quickly as has happened with platter HDDs previous generations. I think we're looking at a much slower pace of significant cost-reducing revisions.

SSD's are the one area that is dropping fairly fast relatively speaking....it's just very high to begin with.
 
SSD's are the one area that is dropping fairly fast relatively speaking....it's just very high to begin with.
HDDs started at a lower point, and dropped faster to an even lower point.
 
The problem is that 5 nm is unlikely to get there for them. 5 nm Lockhart is likely to cost roughly the same as 7 nm Lockhart. Memory also isn't progressing in cost nor size nearly as fast as they would need to even think about Lockhart in a fire stick form factor. Hell, I'm doubtful they could even get 5 nm Lockhart + everything else it needs into an Apple TV sized box, although it's potentially possible if they go with some expensive cooling and an external power brick. Part of the reason the XBS-S is as large as it is, is due to the relatively simplistic yet large SOC cooling solution that they use. But that means it's also cheaper. For XBS-S, cost was far more important that trying to get it to be as small as they could make it.

The end result would be a smaller Lockhart based machine at 5 nm that would potentially cost a fair bit more than the XBS-S.

Regards,
SB


Its all steps on the road to a fire stick like device. I don't think they expect it over night.

You can read DF's interview with them where they talk about the improvements they added to XSS that will make it perform above its "4tflop" power thats inside of zen and navi vs jaguar and gcn. Like i said improvements in navi 3/4 and zen 4/5 can allow a similar 4tflop console to perform better than the current xss. If they are targeting the same 4tflops or so of power but with a newer architecture they should be able to use less power and produce a smaller console.

Smaller APU that uses less power and heat to perform the same or better , infinity cache will allow them to use slower ram with a smaller bus and doing all this will allow for a smaller power supply and smaller cooling which leads a smaller sized console.
 
Its all steps on the road to a fire stick like device. I don't think they expect it over night.


Smaller APU that uses less power and heat to perform the same or better , infinity cache will allow them to use slower ram with a smaller bus and doing all this will allow for a smaller power supply and smaller cooling which leads a smaller sized console.

I didn't think there was an Infinity cache on the scarlet SOCs? I know the digital foundry guys where told there is 76 Mb of cache on the chip but its been radio silent on that front since. Did something change just before the chip design was finalised?

EDIT: or do you mean in the future?
 
I didn't think there was an Infinity cache on the scarlet SOCs? I know the digital foundry guys where told there is 76 Mb of cache on the chip but its been radio silent on that front since. Did something change just before the chip design was finalised?

EDIT: or do you mean in the future?
in the future as a way to improve performance . 128MB of cache on 5nm will be extremely small and if they are targeting 1440p again they might not need 128MB perhaps a 64 or 96MB cache would be enough
 
Back
Top