• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Digital Foundry: If Xbox One X is $500 - How much will next-gen consoles cost?

Then PS5 will have 16-core Threadripper with a tiny GPU.

IIRC, wasn't it true that he wanted two Cells in PS3(sans a traditional GPU)before they decided it would be too difficult for devs to bother figuring such a scenario out?
 
Split memory adds complexity to the board and is likely more expensive than just reserving some from a single pool.

True but for Pro they did add 1GB slower RAM if i remember right .
Still of course all of this will depend on what type RAM they using , the set up and the price RAM will be at then
 
16GB GDDR6/HBM at 600+GB/s with a Zen+CPU has

16 * 64GB/s = 1024GB/s of bandwidth with GDDR6
16 * 56GB/s = 896GB/s of bandwidth with GDDR5x
16 * 28GB/s = 448GB/s of bandwidth with GDDR5

GDDR5x is most appealing in that case, but...Samsung doesn't produce it, Micron does so that makes me skeptical.
Of course you can stick with GDDR5 but you'd need more than 16Gb I reckon, XBOX is at 326 (12 * 27.16GB/s)
 
IIRC, wasn't it true that he wanted two Cells in PS3(sans a traditional GPU)before they decided it would be too difficult for devs to bother figuring such a scenario out?

you are not imaging things


Ken Kutaragi’s grand ambitions for his final system was that its Cell processor would be able to handle the game, system and graphics processing simultaneously.

The eventual realisation that this plan was unworkable meant the console was delayed by a year while Sony looked around for a graphics partner.

“For a while, [PS3 had] no GPU, it was going to run everything with Cell’s SPUs,” the insider told IGN.

“The ICE [Initiative For A Common Engine] team proved to Japan that it was just impossible. It would be ridiculous. Performance-wise, it would be a disaster. That’s why they finally added the GPU, closer to the end.”

With Sony brokering a deal with Nvidia to use its GPU, the system was delayed by about a year. Microsoft’s Xbox 360 was handed a year-long head start that Sony never managed to overcome. The late addition of the graphics card is also why the system’s price was significantly higher than its rivals.

The story has become a kind of folklore within industry circles, with claims that this was the key reason why the PlayStation 3 was significantly more expensive than initially intended.
 
The Xbox One X msrp would probably have been lower if it wasn't just an optional upgrade.

1. MS won't be manufacturing Xbox One X's at anywhere near the same scale as a true next-gen console.

2. Based on the fact 60% of PS4 Pro buyers in the UK already owned a PS4, XBO X sales will probably fare the same, and I imagine they won't result in big software/accessory sales boosts.
 
16 * 64GB/s = 1024GB/s of bandwidth with GDDR6
16 * 56GB/s = 896GB/s of bandwidth with GDDR5x
16 * 28GB/s = 448GB/s of bandwidth with GDDR5

GDDR5x is most appealing in that case, but...Samsung doesn't produce it, Micron does so that makes me skeptical.
Of course you can stick with GDDR5 but you'd need more than 16Gb I reckon, XBOX is at 326 (12 * 27.16GB/s)

What is funny is i remember a year go when people were talking about next gen .
The talk was all HBM things really do change fast .
 
you are not imaging things

Lol.

Sony was indeed smoking that good shit back then.

16 * 64GB/s = 1024GB/s of bandwidth with GDDR6
16 * 56GB/s = 896GB/s of bandwidth with GDDR5x
16 * 28GB/s = 448GB/s of bandwidth with GDDR5

GDDR5x is most appealing in that case, but...Samsung doesn't produce it, Micron does so that makes me skeptical.
Of course you can stick with GDDR5 but you'd need more than 16Gb I reckon, XBOX is at 326 (12 * 27.16GB/s)

I'll take the highest possible bandwidth that still has reasonable cost and power savings

If they have to strike some sort of balance, make sure its a good one. For me, good would be anything over 600GB/s, whatever it may be.
 
$399 or $449 might even be cheaper.


Xbox One X isn't for the mass market & MS can live with it never selling over 15 million but a new platform would need to be priced for the masses so it can build an big install base.
 
What is funny is i remember a year go when people were talking about next gen .
The talk was all HBM things really do change fast .

HBM1 and HBM2 are flops power consumption wise, and yield wise too. That is why HBM+ and HBM3 are in development in addition to GDDR6
 
I have not check in a while how are things going with HBM ?
Any chance it could make into a next gen console ?

They are making a cut down HBM design to significantly improve yields and voltage for lower tier consumer devices(aka L(ow)C(ost) HBM(LCHBM))

53536_02_low-cost-hbm-way-hit-mass-market-soon_full.png


HBM3 will remain for high end business and high end desktop related devices

9bb6b7df-6be9-45fe-a2d1-6e4580a309f1.jpg
 
I'll take the highest possible bandwidth that still has reasonable cost and power savings

Older chips have a higher voltage also 1.5v which would result in a higher power draw, the middle ground is GDDR5x at 1.35v.
More chips is also larger PCB unless they go with 2Gb chips then you got a clam-shell setup like base PS4 (16x512Mb, 8 on each side).

Cheapest is GDDR5 which probably plays a bigger role than the chip voltage, you'd get.
32 * 28GB/s = 896GB/s (869GB/s being the actual bandwidth), that's still a lot and that on a 1024-bit bus.
 
How do you define meaningful? What changes to games would such a CPU upgrade bring?

Its not just CPU, it would mean being able to use the full power of the console.

Scorpio could do amazing things if it had games developed from the ground up for it as opposed to running XB1 games in higher resolutions.

Forza 7 and every other game, illustrates this.

Its like how a 1080ti and i7 could destroy every game out there if there was a dev crazy enough to make a game similar to what Crysis was back then. But that would mean sacrificing any sort of meaningful player base. Even Star Citizen, the closest thing to a modern Crysis, has lower settings for lower level PC's to handle the game.

depends on whether or not they want to throw in shit like UHDBLURAY etc.

$400

XB1S kinda proved that the UHD BR drive doesn't cost all that much more than the BR drive itself if they really wanna include it.
 
$500USD is the maximum any of them will be moving forward, honestly... Well, except Sony's. Who knows there, based on how arrogant they are.
 
Next Gen will probably have the lowest spec hardware they can get away with, because the console market is price sensitive. No matter how good a 500-600$ box is, it just won't sell enough to make the console model work. So no bitching about next gen hardware when the time comes. Mid generation refreshes would be more powerful than any launch 600$ console anyways.
 
Next Gen will probably have the lowest spec hardware they can get away with, because the console market is price sensitive. No matter how good a 500-600$ box is, it just won't sell enough to make the console model work.
We never had a proper attempt to test this hypothesis. PS3 was too exotic and X1 wasted on Kinect.
 
The only difference there was the HDD it's not the same, but I get your point.
$499 is easier to swallow than $599 but in the case of the PS3 it was the same hardware, no improvement to the components whatsoever.
599USDollarspic1.jpg

CPU/GPU, yes. The $499 model also lacked some USB ports, memory card readers, and wireless. I felt compelled to say that, I don't think it changes the point. : p
 
IMO, much of this article was irrelevant to the factors that will likely influence PS5 pricing.

I can't fathom PS5 including XBX features like vapor cooling, and I don't see a 2018 release applying as mentioned at the conclusion of the video. Sony has no apparent to rush a new generation, particularly at a time before the next process will be available at mass production. For that, we're looking at 2019 or even 2020.

And 7nm will offer a generational leap at $400. Zen 2/3 with a 12-16 TF Navi-based GPU could be on the table for holiday 2020 launch.
 
Next Gen will probably have the lowest spec hardware they can get away with, because the console market is price sensitive. No matter how good a 500-600$ box is, it just won't sell enough to make the console model work. So no bitching about next gen hardware when the time comes. Mid generation refreshes would be more powerful than any launch 600$ console anyways.

Unless they both do what Ms already plans to which is keep the bottom console compatible with current gen hardware to allow for a bigger install base, and availability of a budget priced console and premium priced console.

I really prefer this move because launch consoles take years to get a large enough install base for the sales large dev houses need.
 
I'm hoping $500 at launch to pack the most power.

Cost should come down to $400 far quicker than PS3 days, given use of more traditional architecture.
 
It's like they suddenly figured out consoles tend to release when they can be made for certain price points.

Like, you could have technically released an N64 in 1990, but it would have cost about 120,000$ per person soooo
 
This is what's bothersome and not the APU itself: RAM prices on the rise
Here's a funny calculation that somehow ends up being the same, XBOX jump from PS4P is 1.8TF in year yeah,so let's say 2018+2019 the same equation is kept in mind, that would give us 9.6TF (6+3.6) in a PS5.
Now this funny part, Vega Eclipse GPU is or should be 10.7TF for $399, if you take the same GPU and downclock it which should be the case because it runs at 1500Mhz, initially I calculated a 12TF PS5, that would put us at a downclock speed of 1340Mhz.
56 CU * 64 shaders = 3584 (the exact amount Vega Eclipse has) *2 1340Mhz = also 9.6TF

A lot of this is just speculation and assumption, far from facts but I can't help but laugh at the coincidence.
More so when I think that from a price point of view and specs that Vega Eclipse will be the most feasible GPU for PS5.
 
I think part of the strategy for xbox is the Xbox One X will reduce in price in the coming 2 To 3 yrs and become the defacto new gen box for MS. They bring out a newer box in say 3 To 4 yrs that will be $500ish while the X at that point will be around $300 or lower and become for MS what the S version is today. At some point in the coming 3 yrs MS will unshakle devs and let them make X only games and discontinue the S model from production a yr maybe after that.
 
Didn't some MS reps say that they were marketing XB1x as a "premium" product? I don't think that consoles will cost more than 400. It wouldn't make sense given how fast sony was able to move ps4s at that price and how quickly the switch is selling as well.
 
I don't think that consoles will cost more than 400. It wouldn't make sense given how fast sony was able to move ps4s at that price and how quickly the switch is selling as well.

$400/$500 the price difference is like paying $1.6 dollar more per month on a 5 years investment.
400/60 = $6.66 dollars per month worth of entertainment, I'm pretty sure a day's food cost more than that or 1 visit to the cinema or even buying 1 BD movie.
Of course you can argue that the games and subscription adds up but that's with everything, but the initial investment is peanuts really.

#cannot_compute_human_logic
 
What's more amazing to me is the Switch price tag. I thought about getting one until I saw how much it costs. It's almost as much as a PS4 Pro goddammit, and that's without a proper controller. I mean, wtf?
 
This is what's bothersome and not the APU itself: RAM prices on the rise
Here's a funny calculation that somehow ends up being the same, XBOX jump from PS4P is 1.8TF in year yeah,so let's say 2018+2019 the same equation is kept in mind, that would give us 9.6TF (6+3.6) in a PS5.
Now this funny part, Vega Eclipse GPU is or should be 10.7TF for $399, if you take the same GPU and downclock it which should be the case because it runs at 1500Mhz, initially I calculated a 12TF PS5, that would put us at a downclock speed of 1340Mhz.
56 CU * 64 shaders = 3584 (the exact amount Vega Eclipse has) *2 1340Mhz = also 9.6TF

A lot of this is just speculation and assumption, far from facts but I can't help but laugh at the coincidence.
More so when I think that from a price point of view and specs that Vega Eclipse will be the most feasible GPU for PS5.
9.6 tf for PS5 would be disastrous. It wouldn't even be a bigger jump from 1X as 1X is from XBO, Sony would have serious issues to show a generational leap. I think they need to be around 14-16tf or better unless they go for a Switch clone.
 
9.6 tf for PS5 would be disastrous.

It would be stronger than XB1+PS4+XBOX combined, not to mention the big CPU bump, amount of memory at it's disposal and everything else.
They could also customize the GPU by adding 2CU to it like MS did with XBOX, that'd give you 56+2 * 64 = 3712 *2 1350Mhz (10Mhz more) ---> 10TF.
So many ways to keep the console at $399, just don't fuck with the amount of RAM, hell you can even downclock the CPU a bit (which I highly doubt).
 
9.6 tf for PS5 would be disastrous. It wouldn't even be a bigger jump from 1X as 1X is from XBO, Sony would have serious issues to show a generational leap. I think they need to be around 14-16tf or better unless they go for a Switch clone.

Why? Thats around a x4 jump from the base PS4, its good enough to run native 4K and its more powerful then the Xbox One X. Thats a typical generation jump for the 80% of PS4 owners who have a base PS4 or a slim.

Besides the CPU and Ram bumps are the bigger deal and the type of people who own a pro will mostly understand that.
 
9.6 tf for PS5 would be disastrous. It wouldn't even be a bigger jump from 1X as 1X is from XBO, Sony would have serious issues to show a generational leap. I think they need to be around 14-16tf or better unless they go for a Switch clone.

Your thinking in terms of a jump from Pro and XB1X. I keep saying, this thinking is completely flawed.

Just in terms of basic flop count from 1.84 to 9.6tf (which isnt indicative of the power differential), the flop count would be over 5x from PS4 to PS5. The jump in flop count from 360 to XB1 is about 5x itself.

Pro and Scorpio's only purpose is to run PS4 and XB1 baseline games at higher resolutions. They are not utilizing anything close to what many of the components in the machines(especially Scorpios) could in a vacuum.

PS5 is not an iterative unit, it is the new baseline. And so what it will bring to gaming is something completely different, even if it was only a few flops higher than Scorpio.
 
It would be stronger than XB1+PS4+XBOX combined, not to mention the big CPU bump, amount of memory at it's disposal and everything else.
They could also customize the GPU by adding 2CU to it like MS did with XBOX, that'd give you 56+2 * 64 = 3712 *2 1350Mhz (10Mhz more) ---> 10TF.
So many ways to keep the console at $399, just don't fuck with the amount of RAM, hell you can even downclock the CPU a bit (which I highly doubt).


To give you a sense of scale, a native 4k 30fps game that exists on Xbox One X would not be able to be pushed to 60 frames per second on this hypothetical <10tf PS5. It's true that you could say "well, we'll just target reconstructed-4k instead of native", but the PS4 Pro is only slightly under half of that power, and that's producing PS4 quality games at reconstructed 4k for the most part. Sony would be unwilling to drop the target resolution of it's new console below the previous mid-gen refresh, so you're basically saying "oh ok, let's ship a new console. It's twice as powerful as the last one!" That's not what anybody expects from a "generational improvement".

The advancement of GPU technology is slower than in the past, die shrinks are much slower, and the demands of 4k - even reconstructed 4k, make the prospect of shipping a new console with such a modest power increase highly unlikely.

I could see a 2020-2021 console launch targeting between 16-20 Tflops, depending on how future chips pan out. Even 16 would feel underwhelming in some ways, but Sony and Microsoft are both slaves to the industry at large and can only ship with the technology available, not the technology they might want. A few years back, I thought it was lunacy for them to aim for another 8 year generation, but now it feels inevitable thanks to the mid-gen refreshes and the push for 4k.
 
I see where you're getting at, even if 16TF seem like a lot at the moment.
AMD isn't the best partner in the sense of progression, at least not GPU wise because VEGA is way behind schedule.
Which is a thing of concern for the next generation of consoles, if they don't pick the pace up they'll probably be chasing after the fact with Navi as well.
 
To give you a sense of scale, a native 4k 30fps game that exists on Xbox One X would not be able to be pushed to 60 frames per second on this hypothetical <10tf PS5. It's true that you could say "well, we'll just target reconstructed-4k instead of native", but the PS4 Pro is only slightly under half of that power, and that's producing PS4 quality games at reconstructed 4k for the most part. Sony would be unwilling to drop the target resolution of it's new console below the previous mid-gen refresh, so you're basically saying "oh ok, let's ship a new console. It's twice as powerful as the last one!" That's not what anybody expects from a "generational improvement".

The advancement of GPU technology is slower than in the past, die shrinks are much slower, and the demands of 4k - even reconstructed 4k, make the prospect of shipping a new console with such a modest power increase highly unlikely.

I could see a 2020-2021 console launch targeting between 16-20 Tflops, depending on how future chips pan out. Even 16 would feel underwhelming in some ways, but Sony and Microsoft are both slaves to the industry at large and can only ship with the technology available, not the technology they might want. A few years back, I thought it was lunacy for them to aim for another 8 year generation, but now it feels inevitable thanks to the mid-gen refreshes and the push for 4k.

I can tell you this in no way would a 16Tflops console feel underwhelming around 2020 .
The jump from PS4 to Pro is less than 2.5 times GPU wise .
That would also be a bigger jump from PS3 to PS4 GPU wise also .
 
Going by the beefiest mainstream console of each generation:


Xbox: 20 GFLOPS
360: 240 GFLOPS (+ x12)
PS4: 1.8 TFLOPS( + x9.. ish)


General trend seems to indicate that following the modern tech trend we should expect the next generation to aim somewhere around 12-14 TFLOPS if it maintains the relative amount of jump. The XBX is currently an x3 ish jump when trend is indicating the next one should be closer to 6x.

They might get lucky if HDMI gets stuck at 60hz/4K and they can sort of catch up. If 144hz and 4K starts becoming standard, that 14 TFLOPS is gonna feel cramped.
 
$399 / £349 is the sweet spot, so they would be crazy to push for something much higher when next gen arrives.

PS5 in 2019 / 2020 will give Sony time to make something more powerful but cheaper than the Xbox One X, as component prices come down.

PS5 - 3.0GHz Custom Ryzen CPU, 8 - 10 Tflop GPU, 16GB RAM, 2TB HDD.

Not a massive leap forward but big enough to make a difference, handle 4k gaming well and keep cost down.

Possible scenario, but if that's the price point then MS might be able to see the XOX for 199$ and IF they're able to fill it with games, might end in a good position having a slightly less powerful console but with a userbase built in three years.

Really, will be all about games again where we know that Sony does a good effort, while MS goes up and down like a rollercoaster.
 
It seems like GPU's are making big jumps all of a sudden, you're seeing 25-30TF top of the line cards scheduled for 2019.
If that all pans out as nVidia/AMD expects around that time, then a 16TF+ in a console in 2019/2020 might seem like peanuts.
 
16 * 64GB/s = 1024GB/s of bandwidth with GDDR6
16 * 56GB/s = 896GB/s of bandwidth with GDDR5x
16 * 28GB/s = 448GB/s of bandwidth with GDDR5

GDDR5x is most appealing in that case, but...Samsung doesn't produce it, Micron does so that makes me skeptical.
Of course you can stick with GDDR5 but you'd need more than 16Gb I reckon, XBOX is at 326 (12 * 27.16GB/s)

You talk about a 512 bit interface here.
makes the SOC just bigger, more expensive and power hungry
 
I can tell you this in no way would a 16Tflops console feel underwhelming around 2020 .
The jump from PS4 to Pro is less than 2.5 times GPU wise .
That would also be a bigger jump from PS3 to PS4 GPU wise also .

16TF would be 3.8x increase from PS4 Pro, or 2.67x increase from Scorpio. If you only compare to base model PS4 and Xbox One it's a great increase... but we already have the half steps providing a large resolution boost and the new systems aren't going to reverse that trend, if anything people will expect a native 4k presentation more frequently than they do on PS4 Pro, vacuuming up more power.

I would be very happy with 20tf in 2020-2021, but am skeptical they could reach that in a $399 SoC. It will ultimately come down to Navi, it's successor architecture, and the 7nm node as to how good they can ultimately be at an affordable price. Maybe in a few years we'll be going "haha how dumb we all thought they'd never reach 20 and they gave us 25!" or even my pessimistic 16tf could end up being optimistic if future architectures are underwhelming or get delayed significantly.
 
Top Bottom