Could/Should XB1S get a new, faster hardware revision? *spawn

MS will need a console to exist at a price point lower than Scorpio for at least the next 2 ~ 3. Currently, that means they'll continue manufacturing the X1S.
Yes. This is correct.
For almost no extra cost on BOM, MS could be at much less of a disadvantage, and actually have an advantage in some ways.
There are a lot of assumptions here for such a thing to happen.
Indeed!

Take X1S. Swap the already designed and already 16nm optimised "Jaguar+" cores in the X1X - that work just fine with X1 games - and use them. Take the DCC from X1X - which already works just fine with X1 games - and use it. Move over to DDR4 2400 if it's cheaper, save money and get a 10%+ BW increase on the most BW constrained part of the system (see iD's Wolf2 struggles).

So eliminate the DDR3 wall with DCC and mainstream DDR4, get additional CPU IPC, clock higher with yield improvements and Hovis stuffs if it's cheap enough .... and basically for the same manufacturing costs your substitute for the X1S that you have to make anyway is better at everything (fps, dynamic res, BC, loading times).

On the PC R&D for halo products trickles down into mainstream products. Xbox could do that too. It won't cost you silicon, and most of the R&D is already spent.
So I have an idea of where you are going with this. So I'll try to address what you're getting at. But you're basically saying, build a better XBO with a higher performance profile to replace the aging XBO for a cheaper price. That's the gist of it from what I can tell.

Lets check some constraints and poke and prod and see how well your idea may hold up (some things need clarification for me):
a) The market is still flooded and flooding with older OG XBO and S. Are developers going to profile 3 platforms now? Or will they only still only target 2 platforms?
b) When XB2 arrives what will happen to the XBO family of SKUs?
c) Will they immediately cancel product of 1S SKU? They'll need to call this something else right? How would consumers know which SKU you are buying if they are the same name?
d) We know today that 1X has to remove 1/2 its GPU power to run older XBO titles without patch. How would this work on this new SKU? How large of a GPU would it have to be to half it and be able to run compatibility?
e) with ESRAM still there, how much of a major advantage are you really going to get over the base model? real estate is still at a premium if esram is present.
f) If they aren't the same name, is this a new generation? They just launched a 1X assuring everyone it wasn't a new generation, so what's happening here?
g) Won't new owners of 1S be super pissed off by this?
h) How quickly would they be able to release such a model? We know it took nearly 4 year of development to release Scorpio.
i) Would they be able to release this product before next gen hits?

If you can answer these with strong points, perhaps the product is coming.

Put simply, MS have to keep making the X1S for another two or three years. Make it as competitive as possible, for as little cost as possible.
There are many companies that many thought were 'too big' to fail. Having a leadership team that knows when to cut losses, and move on is what keeps companies alive for the next round. When the round is lost, retreat is the best option, otherwise you are just throwing resources away for the hell of it.

If you are honest with yourself and ask what type of company MS is, and how it has endured so many changes, and so many failures and yet, still be up there. What do you really think MS would do in this situation, continue to fight the 2013 Gen? Or move the battleground to next gen?

Remember looking back at... Nokia... Windows vista, Windows 8, Zune, Windows CE, Windows ME, Live for Windows... OG Xbox... all the first party studio closures, game cancellations etc etc etc.
 
Last edited:
You should check out TV's, PC's, phones, tablets. The fact that revisions can be better and more competitive within their segment would blow your mind! :p
Well, I already pointed out the difference between consoles and other CE devices. They also separate themselves from the old model to tell us they're better. In consoles, we have revisions that aren't better. You buy a console one year and someone else buys the console another year, you get the same console. Any variation is minimal. As such, any new model not advertised as a new model will not be assumed 'new and improved' like a TV or phone, but assumed to be the same. Therefore the significant improvements would have to be communicated.

Not my main point, but that's not necessarily true. If brand attachment, software sales, Live/PSN subscriptions improved that could be worth it too.
It's not? You're not expecting better sales? So MS are supposed to increase their costs but not get any financial returns over and above what they're already get with XB1S??

The truth is that there are still people out there considering buying X1S as an alternative or in addition to PS4. That's a real thing that's really happening.

Giving them the most compelling machine for their money would have at least some value.
For a business case, if the cost to attract these customers is more than you'll get from them, it's a bad decision. How many people are considering XB1 or PS4 now? What are their reasons for holding off an XB1? What's the cost of fixing those concerns?
 
Lets check some constraints and poke and prod and see how well your idea may hold up (some things need clarification for me):
I'll field these for function. ;)

a) The market is still flooded and flooding with older OG XBO and S. Are developers going to profile 3 platforms now? Or will they only still only target 2 platforms?
These new machines is supposed to replace XB1S. It'll run the same code, just at higher framerates. It won't be targeted as a discrete target.
b) When XB2 arrives what will happen to the XBO family of SKUs?
No different to whatever happens without XB1SE
c) Will they immediately cancel product of 1S SKU? They'll need to call this something else right? How would consumers know which SKU you are buying if they are the same name?
XB1SE. Well, that's my argument with function too. I think there's a bit of a communications problems that'll add to costs, telling people the new model is so much better than the old model that the new model outperforms the alternative platform.
f) If they aren't the same name, is this a new generation? They just launched a 1X assuring everyone it wasn't a new generation, so what's happening here?
It's XB1SE with a turbo charger fitted. Still the same generation and platform.
g) Won't new owners of 1S be super pissed off by this?
Probably, but that's life. What they gonna do?

The other points I don't have answers to. ;)
 
Would a more powerful mid-range sku at a mid-range price reach an untapped market? Or can they make a more powerful sku for the same cost as the S? Or take a loss?

I think it's a mistake to assume the PS4 success over the XB1 is primarily about console power. Whenever console A have more success than console B, the reflex is to propose a new sku with more power, or overclocking etc... I think that is not possible without raising the price, all else being equal.

I remember after the PS4 launch success, Yoshida said they didn't understand why the PS4 had more penetration than they predicted internally, and that it's a very bad thing for a company, regardless of the success. It was a dangerous situation where the next move would be based on the wrong assumptions about the market. If you don't know the precise reasons something worked, you can't do it again consciously. Years later he was asked about that statement and he said something like "yes we eventually figured it out, but I'm not telling". From the timeline, that interview was much more important than it looked on the surface, because it was about the PS4 Pro which was in development. My take on this is that they chose price, early launch, make sure everone understand it's the same generation, and they did very very little publicity so it didn't canibalize the ps4 slim.

MS don't need a new mid-range product, they need to drop the price on both the X and the S. Spending hundreds of millions to develop an intermediary SoC would be a weird move. The S is flying off the shelves at $189.

More power for more money never worked for consoles. At some point they have to take a loss to grow the market. They chose not to.
 
Last edited:
More power for more money never worked for consoles. At some point they have to take a loss to grow the market. They chose not to.
Reducing the price on something people don't want is, not going to solve problems :)

They needed to first make the product desirable again, then they can work on price points. I think the relaunch of Xbox with the 1X helped bring Xbox back into positive light. A lot of the features on 1X are also on 1S but no one paid attention to those features until MS had launched Scorpio. Then people realized that they could get nearly the same feature set but for significantly cheaper.

A big marketing win for MS if true, they really needed to turn around their brand.
 
If anything I see Microsoft releasing a cheaper X1X once they can get a 7nm chip and maybe phase out X1S all together. Wouldn't be shocked if Sony did the same tbh..
 
Unless the reason they don't want it is the price...
one day ;)
I think from a business standpoint you want to be careful about subsidizing hardware for growing your market, if no one intends to buy software or use your services then it's a pure loss.

From my perspective, if I were MS, I would not want people buying the 1S as an MS Exclusive + UHD player machine. There aren't enough exclusives to really make a profit, and you don't profit from UHD. If no one buys 3rd party games on your system, or doesn't sign up for game pass or Live Gold, there's very little to recoup your cost of acquisition. Which is the long run isn't going to be much.

I think after this years black Friday, this will be their first big run with 'Game Pass' being sold to the masses, if this is successful it's possible that we could see discounts on hardware in the future.
 
XB1S added HDR + 4K output + 4K drive. That's already good enough for many consumers. Clocks were slightly improved to make it as easy as possible for developers to patch in HDR support (extra tonemap pass). Simple and easy upgrade, all games work perfectly out of the box, with some new features that consumers are looking for in modern equipment they connect to their UHD TVs.

Refresh with around 40% perf increase (to reach parity with PS4) featuring GDDR5 (no ESRAM) would have required much higher development cost. New die, new memory controller, new motherboard, etc, etc. Current software is developed around ESRAM. Scorpio backwards compatibility mode simply maps 32 MB ESRAM area to GDDR5. This isn't a problem since Scorpio has 4.5x performance advantage over Xbox One. But with a console equal to PS4 perf, the extra memory copies from GDDR5->GDDR5 and the memory contention would both hurt the performance (**). In order to get best performance, developers would need to patch their games. And how many developers are patching games simply because of 40% faster hardware? PS4 Pro is over 2x faster and still not all games are patched.

(**) Memory contention is a real issue when both GPU and CPU access the same memory. Look at the graph in this page: https://wccftech.com/sony-ps4-effective-bandwidth-140-gbs-disproportionate-cpu-gpu-scaling/. Please ignore all the console war bullshit text in that article, just look at the graph. This isn't a problem that only affects consoles. It affects every system with shared memory between CPU and GPU. Large LLC obviously helps (Intel's way to reduce the problem). Dedicated video memory (even a tiny one at 32 MB) still has some advantages.
 
(**) Memory contention is a real issue when both GPU and CPU access the same memory. Look at the graph in this page: https://wccftech.com/sony-ps4-effective-bandwidth-140-gbs-disproportionate-cpu-gpu-scaling/. Please ignore all the console war bullshit text in that article, just look at the graph. This isn't a problem that only affects consoles. It affects every system with shared memory between CPU and GPU. Large LLC obviously helps (Intel's way to reduce the problem). Dedicated video memory (even a tiny one at 32 MB) still has some advantages.
Hey Sebbbi, question about this one, perhaps one that I've had a long time trying to break down and figure out.
I've seen _a lot_ of people say that resolution won't affect CPU load. Such that if your CPU can sustain 60fps at 1080p it can also sustain 60fps at 4K (provided your GPU was large enough).
I've read this everywhere an without fail, this is the common thought process on this one, it's recently come up again as well with the recent Digital Foundry comparisons between 4Pro and 1X, so it's got me thinking that we're all wrong? (Too many fanboys are trying to benchmark the hardware, when we should be bench marking how the software utilizes the hardware. Typical console war BS. )

But I reckon now that this is a general statement and is more centered around the behaviour of PC than they are to consoles. This particular topic arises when I think Assassin's Creed Unity was released at 900p on both consoles and they claimed that CPU was the reasoning for the resolution. And everyone had just screamed BS because if CPU is the bottleneck that resolution could increase upwards with the GPU as required. I think this is true for situations that's if the CPU doesn't have enough cycles, it becomes a bottleneck, with having no effect on GPU load.

But then, this particular topic of memory contention always comes to mind for me. Is it plausible that if the CPU load was particularly heavy, and heavy in accessing memory, could memory contention be so drastic that there is little to no available bandwidth left for the GPU to perform higher resolution?

Suppose a scenario for PS4 that memory contention was so great that it would drop to below 100 GB/s maybe 70 GB/s for the GPU, then your available bandwidth to ROPS and CUs is so low that to meet a 30 fps target, resolution reduction is your only chance to meet the frame time?

Really, bringing it back, but the move to mid generation refresh, to support higher resolution, a great deal more bandwidth is required because of contention? And that is probably a larger limiting factor now than ever as we are attempting to push 4x more pixels (whether reconstructed or not) ?

Hell, if we extrapolate further, the move to Forward + /tiled/volume tiled/ etc rendering is ideal then because the hit on bandwidth is less for higher resolution for consoles, and ideal in this case versus deferred?
 
Last edited:
@iroboto Shifty already did a pretty good job of answering many of the points, but I'll try and do what else I can! ;)

There are a lot of assumptions here for such a thing to happen.

True, but I think there's a fair chance that the core ones are correct: basically same die area, same number of memory chips, lower memory cost at some point, same or lower motherboard complexity [I'd remove HDMI in], no additional dev work, best IQ within its price segment, highest frame rates within its price segment, faster load times, best possible app experience, etc.

So I have an idea of where you are going with this. So I'll try to address what you're getting at. But you're basically saying, build a better XBO with a higher performance profile to replace the aging XBO for a cheaper price. That's the gist of it from what I can tell.

Yeah. MS will need a cheap console until at least a year or two into a 7nm Scorpio shrink, so try to tweak the cost / performance parameters to make the experience of buying one as attractive as possible.

a) The market is still flooded and flooding with older OG XBO and S. Are developers going to profile 3 platforms now? Or will they only still only target 2 platforms?

As Shifty said, it's just the two.

b) When XB2 arrives what will happen to the XBO family of SKUs?

As Shifty said, exactly the same thing that's going to happen anyway.

c) Will they immediately cancel product of 1S SKU? They'll need to call this something else right? How would consumers know which SKU you are buying if they are the same name?

I'd have been planning for a switchover. Handle it in the same way as X1 to X1S - shift manufacturing and sell through the old SKU discounted. Naming and branding would be something indicating a fun, swift, value based alternative to the bruiser of Scorpio. X1Lite, or something to that effect.

d) We know today that 1X has to remove 1/2 its GPU power to run older XBO titles without patch. How would this work on this new SKU? How large of a GPU would it have to be to half it and be able to run compatibility?

You base it on the X1 SoC. Same arrangement, including esram. The SoC works just like it does now, only faster and with modules replaced with enhanced ones where they've been developed (enhanced CPU, enhanced ROPs). No units go to waste. If yields are good enough maybe activate the two redundant CUs. It's still basically the X1.

e) with ESRAM still there, how much of a major advantage are you really going to get over the base model? real estate is still at a premium if esram is present.

You have to plan around esram even if you no longer have it and are just emulating it. So I'm saying keep it, and stick with cheap main memory. Cheapest and fastest route to the most suitable product.

It would still offer huge advantages over the base model. esram bandwidth scales with GPU frequency. With on-chip transfers, this thing will fly. Off chip, you have delta colour compression and DDR4 2666 or the like to eliminate the DDR3 wall.

DDR3 2133 to DDR4 2666 would give a 25% increase in main memory bandwidth, and DCC (up to 40% effective) would increase this further. For games that are hampered by DDR3 bandwidth (e.g. Wolfenstein 2) such a large increase in effective bandwidth would allow even greater filtrate increases than the clock speed increases would indicate e.g. your SoC is 25% faster, but your fillrate is now 35% higher.

Large increases in clockrate and bandwidth (especially main memory) would allow forced 16x ansio - something Scorpio has already proven works exceptionally well.

With stable frame rates and very high IQ, an enhanced X1 would look amazing on 1080p sets. Check out the Digital Foundry comments about X1 games running unpatched on Scorpio. Halo 5! :runaway:

f) If they aren't the same name, is this a new generation? They just launched a 1X assuring everyone it wasn't a new generation, so what's happening here?

It's definitely not a new generation. It's the X1 you know and ... perhaps like .... it's just a new revision.

g) Won't new owners of 1S be super pissed off by this?

Maybe some? But having worked in retail I know some people are always pissed off. ¯\_(ツ)_/¯

Its better because it's newer. But it's still an X1, and every sale secures further development for the base X1 and X1S. And it works with all their games and peripherals and accounts so they can upgrade if they want.

h) How quickly would they be able to release such a model? We know it took nearly 4 year of development to release Scorpio.

Don't know. The months (years?) of profiling for Scoprio, and their experience on 16nm means they should already have all the data they need for speccing and starting to simulate the device (I'm sure AMD have lots of data on DDR4 buses). It'd now be a case of modifications to some blocks of the chip, motherboard and power supply.

I'm not proposing anything new not already developed for X1S or X1S (bar DDR4). Re-purposed IP, same software, same BC, same IQ tweaks, same manufacturing process ....

Two years? Starting from last summer? :eek:

i) Would they be able to release this product before next gen hits?

Hopefully! But when next gen hits, you'll still need that < $200 device. And by god, I hope it'll have stable frame rates and aniso filtering .... :mad:

If you can answer these with strong points, perhaps the product is coming.

I hope it is!

There are many companies that many thought were 'too big' to fail. Having a leadership team that knows when to cut losses, and move on is what keeps companies alive for the next round. When the round is lost, retreat is the best option, otherwise you are just throwing resources away for the hell of it.

If you are honest with yourself and ask what type of company MS is, and how it has endured so many changes, and so many failures and yet, still be up there. What do you really think MS would do in this situation, continue to fight the 2013 Gen? Or move the battleground to next gen?

I think MS need to attack the entire concept of generations. One game library, on a constantly evolving platform that always exists at high and low hardware price points.

The can still piggyback off the hype of a competitors next gen by offering something as capable. Then they can release something better two or three years later.
 
Last edited by a moderator:
It's not? You're not expecting better sales? So MS are supposed to increase their costs but not get any financial returns over and above what they're already get with XB1S??

Miscommunication! :(

I am expecting better sales (not than PS4, but than they would have otherwise had), but also - even though it's not my main point - anything that encourages software sales and the use of services on the Xbox platform (as opposed to competitor platforms) is a benefit.

It's true I don't know enough to offer a conclusive cost / benefit analysis though. Hell I wouldn't even know where to begin ....

If anything I see Microsoft releasing a cheaper X1X once they can get a 7nm chip and maybe phase out X1S all together. Wouldn't be shocked if Sony did the same tbh..

I think MS will release a cheaper, shrunk, Scoprio. But it may be some time till that can replace the X1S at the necessary price point, and in the mean time I think there is a case for a noticeably better and more attractive X1S proposition.
 
DDR3 2133 to DDR4 2666 would give a 25% increase in main memory bandwidth, and DCC (up to 40% effective) would increase this further. For games that are hampered by DDR3 bandwidth (e.g. Wolfenstein 2) such a large increase in effective bandwidth would allow even greater filtrate increases than the clock speed increases would indicate e.g. your SoC is 25% faster, but your fillrate is now 35% higher.
I think reading your answer, there's probably something in the works here already. There will probably come a time where the economics of DDR3 will be at a loss compared to DDR4. And so there's definitely something here that will probably likely to happen for MS to continue to reduce cost on the unit.

I'm not entirely sure about clocking higher, or inserting better features into the unit, but I think both gpu clock speed, and a change in memory are likely to be found true if the economics are right.
 
On one hand I think it's a dumb idea...because you would be adding a 3rd target for developers which would just be ridiculous imo. On the other hand maybe with the BC work done on Scorpio for Xbox One games maybe you could just release it as a *boost mode only* console. So developers wouldn't have to do anything...

Still I don't really see the market. I think if you're Microsoft you just try to work to get the cost of X1X down.
 
I think reading your answer, there's probably something in the works here already. There will probably come a time where the economics of DDR3 will be at a loss compared to DDR4. And so there's definitely something here that will probably likely to happen for MS to continue to reduce cost on the unit.

I'm not entirely sure about clocking higher, or inserting better features into the unit, but I think both gpu clock speed, and a change in memory are likely to be found true if the economics are right.

Yeah, economics permitting. I wouldn't propose a tweaked CPU and DCC if MS hadn't already developed them and deployed them with no compatibility concerns. Likewise, I'd only propose marginal frequency gains if it weren't for proof of techniques already developed and deployed on a mass scale allowing them to reach significantly boosted clocks. Even allowing for a lower cost unit with less emphasis on bleeding edge performance than X1X, a 2.1 gHz CPU and a 1.1 gHz small scale GPU doesn't appear fanciful (1.3 gHz boost for mobile Ryzen SoCs, 1.18 for the much larger X1X GPU).

Likewise, by 2018 DDR4 2666 will be eminently mainstream. Ryzen processors already support 2666 as standard, without overclocking or unofficial (but totally hinted at) multiplier support. Don't think there's even a premium for DDR4 2400 over 2133 any more....

I think despite being mostly "900p", a cheap system with really stable frame rates and top of the line texture filtering (which makes a big difference to what you can actually see) and fast load times could gain fans, or at least lose less and do so more slowly.
 
... maybe you could just release it as a *boost mode only* console. So developers wouldn't have to do anything...
Precisely. The target is XB1 - not XB1S and not XB1SE. These other consoles would just run the same code a little bit faster.
 
Back
Top