So you want to keep the Sram , make the Apu bigger and then add in more costly DDR4.
I still believe MS wants to shrink as far as thy can and then ditch it for cheaper scorpios down the line instead of adding more to their pipe line for development
So you think adding $5 to COGS in exchange for beating OG PS4 and having solid 1080P60 machine that you can still sell at $199 is a poor trade-off?
Yes , because then MS will have the xbox one with 2Xm people , Xbox 1.5 with Xm people and then Scorpio with Xm people.
So which one does a developer target ? Not to mention that its not $5 as MS is going to have to provide support for the new hardware and support the xbox one , xbox 1.5 and scorpio
I understand from Sony's point of view they have the PS4 and PS4 Neo , The PS4 Neo will be used to stabilize the frame rate and for PS VR games. It makes sense to have it. But that's it , they have 2 models . One at 4XM units and one at Xm units.
But that's it just 2 levels to design.
I think if you futz with the xbox one your going to loose its only strengths which is its price and power consumption. The xbox one at 28nm was using 20-30 watts less than the ps4. The shrink could get them into more designs than simply the xbox one slim.
I think you are missing the point. OG + slim + Scorpio = xbox market. There won't be separate markets. Also, accounting nit, support wouldn't be in COGS.
But you will need to create product for all 3. Heck if you add in the rumored oculus support in scorpio you will have this in 2017
Xbox one , Xbox 1.5 , Scorpio , Scorpio VR . Games will still need to be QA and all the settings will need to be tested across all the platforms.
Your going from having to support one system to three different systems over night basicly. I don't see that as a wise move
By that logic it's still the same as Sony have
PS4 , PS4VR, NEO, NEOVR ?
I would also think even with a shrink that's new hardware, new chip design on finFet so also plausible to have specific bugs and performance quirks so the QA may be similar regardless of the actual performance.
I don't think it's ideal but if from here on in platforms will play previous generation games players are only going to get more invested. It's now or never to stem the flow of gamers and also the original hardware is going to fair even worse in comparison. If you have to recreate anyway because of finFet why not choose to create a neo or at least something similar unless you are aiming at a streaming box rather than a full console.
Either go ultra roku small and keep Xbox one performance or go regular sized slim console with neo.?
I thought they were on the same process. Samsung were making 8Gb LPDDR4 and DDR4 on 20nm according to the internets, and 12 Gb LPDDR4 seems to use it to, according to this:
http://www.samsung.com/semiconductor/about-us/news/24281
Sub 20nm is supposedly coming soon though, and it should take DDR4 speeds up to their promised 3200. I don't see why LPDDR4 won't use the same process too:
https://news.samsung.com/global/sam...ucing-industrys-first-10-nanometer-class-dram
As for performance, I don't really know. I've been naively looking at the peak calculated BW and leaving it at that. Think I read somewhere that latency for LPDDR4 4266 is about the same as DDR3 2133, but can't find anything to back that up now.
I think you're probably underestimating the cost and engineering effort. A new chip design could easily be in the hundreds of millions of dollars, ensuring BC is not going to be easy, and the clocks needed to comprehensively beat PS4 will mean more power, bigger heatsink, bigger chip, lower yields, more capable psu, bigger case, more shipping costs, storage etc than just a straight slim and so any advantage of "slimming up" would be reduced. And a slim X1 is where MS's best hopes lie this year, IMO.
If X1 were obsoleted by a similarly priced successor the costs of clearing inventory quickly could also be in the hundreds of millions.
On top of that, the Xbox team would need develop and maintain additional tools, developers would need to buy a new platform's worth of dev kits to sit alongside X1 and there would be the inevitable headaches associated with new and more platforms, all for a machine that would be superseded by Scorpio next year.
And it would still be much weaker than Neo. If you're going to go to the trouble of introducing a new platform, make it worth everyone's while.
one of the 'leaked' reasons behind the neo was that it was cheaper to go that route than what is considered a simple shrink would have cost. Obviously how true it is we don't know.
Or is that reason so unlikely that it has been ruled out already?
but if that is the case, it may also be true for xbox also.
Or is that reason so unlikely that it has been ruled out already?
but if that is the case, it may also be true for xbox also.
upnorthsox
Veteran
I don't think it's necessarily cheaper, more like if upgrading the apu costs you $50 but you're able to charge $100 more for it, you've now added $50 to your margins with the upgrade. Likewise, and to point of XBO too, MS and Sony have contracts to buy chips from AMD, contracts that AMD do not have to nullify unless they are getting benefit from it. So when you are devising scenerios for what chips will be in what box, you first need to ask, what benefit does AMD get from modifying its deal.
That's reasonable, and as you said that goes for MS also.
The only trouble with that is that it doesn't give them the opportunity to reduce the price of the base model, which is probably where the main sales would come from. Even if the profit from each sale is less.
May be able to ofset the cost of lowering the cost of the base model from the profits of the Neo?
Cost of a shrink compared to a change to tech that's already at that node may be very different than every one is assuming though.
If xo slim, is just a slim, MS is going to find itself in a very awkward position when the neo comes out and their seen as Sony having upto a year lead.
Dammed if they do dammed if they don't, haha
Baffin is supposedly somewhat bigger than the GPU component of X1. It goes against the idea of driving X1 cost down as much as possible, as does going for the fastest flavours of DDR4. It'd still be much slower than Neo.
A faster GPU would also demand doubling the ROPs, which in turn may require a significant re-engineering the interface to a now even smaller looking pool of esram.
You're taking about sinking lots of money and engineering into a marginal technical victory against a machine that's already won, while surrendering the low price market that MS could still actually gain some traction in.
A faster GPU would also demand doubling the ROPs, which in turn may require a significant re-engineering the interface to a now even smaller looking pool of esram.
You're taking about sinking lots of money and engineering into a marginal technical victory against a machine that's already won, while surrendering the low price market that MS could still actually gain some traction in.
One could argue that they'd be sticking to similar resolutions if they're sticking to 32MB ESRAM & similar main memory amounts.
More flops per pixel.One could argue that they'd be sticking to similar resolutions if they're sticking to 32MB ESRAM & similar main memory amounts.
Yeah. Unless that faster DDR4 can soak up more framebuffer as well as feeding a much faster GPU it'd be stuck at similar resolutions. So mostly 900p? And 3200 DDR4 isn't even mainstream yet, that'll probably come with the new processes being PRed right now and likely at a price premium for the first few years.
If you wanted to keep current resolutions and go to 60fps, you're going to need double the fill. Even if you just want to be as fast as PS4 at shadow rendering you're going to need double the fill.
So ramr's dream of "beating OG PS4 and having solid 1080P60 machine" for only an extra $5 doesn't seem so easily achievable to me.
The extra CUs would only be another $2 ish worth of silicon so ... yeah. You think $199 is not low price? That is what i was suggesting for the digital only version of the box.
How do you get your 2$ 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 costs of supporting a third device will be greater than the return on investment.
Sent from my iPhone using Tapatalk
And it's a third device that no-one is asking for, and that sits in a no man's land.
Similar threads
