Will AMD have some sort of neural network solution by then mitigating the need for such hefty CPUs ? A.I. as a cloud service for games, maybe a collaboration between Google and Sony with MS finding their own way ?? We shall see but yeah I want more CPU too.
Post Xbox One Two Scorpio, what should Sony do next? *spawn* (oh, and Nintendo?)
- Thread starter mpg1
- Start date
Will AMD have some sort of neural network solution by then mitigating the need for such hefty CPUs ? A.I. as a cloud service for games, maybe a collaboration between Google and Sony with MS finding their own way ?? We shall see but yeah I want more CPU too.
trainplane
Newcomer
My guess is XBox One also had thermal headroom because their fan and box was giant sized as an overreaction to rings of death.
Sony seemed to have developed the PS4 to take what it was designed for. They are better engineers in that sense. If they had tried to increase the clocks at that time, an in-game explosions might have turned into a real life explosion. We like realism in games, but not like that.
One would assume lighting and 4k resolution will be a solved problem by the next True Next Generation of consoles so that will leave wireless VR and A.I. as the up in the air reasons for a new console with maybe new controller schemes and wireless streaming to portable devices should someone WANT to try to follow in Nintendo's wake. I do want more debris with more permanence from a logic perspective but that is just me. There will be a reckoning around violence and viscera especially with VR I would assume as well when photorealism hits a certain point.
I think there's a decent chance Sony could have bumped clocks by one or two multipliers, if it had the need and advanced notice to. Orbis wasn't pushing the envelope for Jaguar or GCN, and there was slack in the cooler and power delivery.
As the overall leader in raw throughput, Sony had less incentive to change. The broader launch in terms of markets and a higher-volume price point would mean having more chips would be encouraged, and making Orbis a little faster wouldn't change the demand.
Possibly, the timing for when the platforms actually did commit to a final spec was different, and Sony committed first.
trainplane
Newcomer
We could do so much more to advance AI and player choice in games. Followers in Fallout that stop running into you and actually feel like companions, more world affecting player choices (like the thing you could do to Megaton in Fallout 3, an amazing moment in gaming to me--what if we could do it to any city), NPCs that react to and remember your actions, use processing power to make these worlds more living and breathing, not just more realistic looking. In fact, the more realistic these worlds look, the more stark the limitations stand out.
The problem though is with this multi-generational approach, you're only as good as your last generation and by the time Playstation Geo roles out, they still have to design games for the PS Neo with the Jaguar potato chip. I really hope Scorpio doesn't use a Jaguar potato chip.
I understand that it would cost an enormous amount of money to ditch the Neo AND start over. What about just ditching it for now? That could certainly cost less and, presumably, you could announce a full PS5 at E3 2017 for launch in 2018 or early 2019. That would put a damper on Scorpio. What do devs have invested in Neo at this point? Enough they would be pissed if it was cancelled, or no big deal?
I think Neo is going to outsell Scorpio. I think it'll be priced lower, and I think it'll be a "good enough" entry point for 4k content and provide a premium boost for games. PS4 games already look pretty fantastic, and more than twice the gpu power is going to make them look fantastic. Artwork will shine through, and that's all it needs. PS4 has a solid player base, and a good reputation. That'll do a lot for it, despite any potential power gap it may end up having with Scorpio.
Isn't Neo the generational mid-life kicker, and that it is slotted at between 3 and 3.5 years after launch? Would the PS5 come in within 1-2 years of that midpoint?
I suppose if there were multiple teams, something might have been in progress for the PS5 while Neo was being looked into, but how much would that have been besides earlier stages of planning, design, or engineering?
Some were hoping for a late 2016 debut, which would put Neo close to physically existing in final or near-final form.
If Neo turned out not to be at a hardware disadvantage, the PS5 being far enough along to slot close after it means the PS5 would be 1 or more years out of date relative to what it could be if it were genuinely a successor to the PS4--if it truly is to be another distinct generation.
If the development processes were following a regular cadence, dropping Neo doesn't make the PS5's development effort start any earlier because Sony would need to have known this would happen before it happened. Dropping Neo now just means Sony has nothing or it has to switch to some other initiative that has ongoing work done--but what would that be?
I suppose if there were multiple teams, something might have been in progress for the PS5 while Neo was being looked into, but how much would that have been besides earlier stages of planning, design, or engineering?
Some were hoping for a late 2016 debut, which would put Neo close to physically existing in final or near-final form.
If Neo turned out not to be at a hardware disadvantage, the PS5 being far enough along to slot close after it means the PS5 would be 1 or more years out of date relative to what it could be if it were genuinely a successor to the PS4--if it truly is to be another distinct generation.
If the development processes were following a regular cadence, dropping Neo doesn't make the PS5's development effort start any earlier because Sony would need to have known this would happen before it happened. Dropping Neo now just means Sony has nothing or it has to switch to some other initiative that has ongoing work done--but what would that be?
Dropping Neo now does have 1 benefit. How many people are nervous about this whole concept of mid-life upgrades? No one knows how it is really going to work. No one knows how the general public will react. It may turn out to be no big deal, but you could have a Sega CD situation on your hands as well. If Scorpio is going to outpower you and there is nothing you can realistically do about it, you could dump Neo, state you never formally announced it and decided against it for any number of reasons. You are then free to rain on your competitors parade 1 year from now. Assuming work on the PS5 has been ongoing. It's not like Sony are above misleading people about the release date either. At least they were not in the past. By dropping it an NOT replacing it but going with an announce next year of a true PS5 in Fall 2018 you get the benefit of outpacing MS, assuming you can build something more powerful for less 1 year later, which may not be true, and you don't incur the cost or risk of releasing the Neo.
Now that is just one unlikely scenario. I'm not saying it will happen or that it should happen. This is just a what if. If Neo is basically built around PSVR and the whole shebang shows well at demo kiosks in stores and costs less than a Rift and VR capable PC then you are probably in good shape. A lot, to me, depends on launch window and price. Especially price. I still find it hard to believe the Neo will launch in a couple of months and we have no official announcement and jack for leaks. Manufacturing would have to be underway very soon. Dumping Neo is just an idea that would let Sony one up MS without incurring the cost and risk associated with quickly replacing it or going ahead with production. The high-end/ hardcore, money to burn consumer may turn out to care a lot more about power and holds off to buy the Scorpio. That may be true either way if you already have one box and both say forwards compatibility is a must, if you are going to buy a new one you may as well get the one you do not have yet. I know that is probably not the case. Many people are locked in and have a weird loyalty to one group or the other, but it is not out of the realm of possibility.
trainplane
Newcomer
I have to say, as much opportunity as cloud computing can bring, it also scares me, especially for single player focused experiences. What if it's down, what if there is an XBox Live paywall behind it--after all someone has to pay for the A.I. servers, what happens in 10-15 years, how does that translate to porting the game to other systems? What if the A.I. servers turn into Skynet? What will we do then?
For single player games, I think having cloud should be a good option, but not core to the experience, at least not until it's a proven solution to the ancillary experiences.
They could cancel, but they're already on record as to Neo's existence and planned release.
With a product that is at least 1 but likely >2 years away at that point.
That's part of what I believe needs some kind of further justification.
Being able to rain on Scorpio's release means Sony was planning on raining on Neo's parade after 1 year on the market.
Why would they have planned to do that?
The risk of having Neo not work out against the competition is replaced by the guarantee that the PS5 that was apparently right on its heels will negate the investment into Neo after <1 year.
If there was something Sony is supposed to squeeze out in 1 additional year past Neo's timeline, it would need to be some kind of significantly advanced Neo plan B, or hopefully there's a PCIe or other form of interconnect and AMD can plug in a discrete GPU (or another Neo chip in SMP *cue dramatic music*).
Sony would have something like a year to figure out how to get multi-GPU to work and walk back its unified memory marketing point.
I've been struggling to figure out where to put this next bit of thinking, since it partly covers Microsoft's plans, but here it goes:
Let's suppose Neo is using Polaris, and Scorpio will use Vega as everyone seems to postulating. Let's not get into the Jaguar/Zen/Zen lite speculation for now.
Imagine the upside for AMD in this case by getting the console market to inject R&D funds into not just one generation like the current ones, but two. Semi custom clients pay for R&D and NRE for hardware that has shared features that are subsidized by reuse.
Not only that, these are "premium" consoles that apparently do not intend to replace the base console chips. Likely higher prices for AMD, and likely less volume on these higher-margin chips such that the slim margins of semi custom don't necessarily drag overall corporate margins down as much as they tend to. The current chips continue to be produced and produce revenue.
And then apparently Sony rushes a PS5 out before Neo could cost-reduce too much--or it's cancelled and AMD pockets the up-front payment since it's not their fault Sony overran its own console project.
Bravo on getting those clients to pay all that money.
However, there is a speculative way to have Neo and the PS5 closer in time, but it requires that Sony was going to be more aggressive about its cadence back when 20nm was going to be used.
AMD took a charge when it indicated it would move any chips in development to FinFET. Let's say there were console shrinks in that set.
If the Xbox One Slim happens to be FinFET, it might be a porting of a 20nm shrink. I'm not clear on whether that product has changed nodes.
Scorpio could be where it is in timing and in peak performance because it is the result of Microsoft not starting from a 20nm basis and purposefully targeting a FinFET node with the technology that would come with it. Maybe Microsoft was going for a somewhat slower cadence, or was willing to go back to the drawing board. That might mean being able to use a Vega GPU, or in this scenario it might be that it is the one using Polaris on a 384-bit bus that AMD has opted for in place of a Polaris 12 in the discrete GPU realm.
Let's pretend Neo could have come out in 20nm, and that Sony didn't want to start all over again, and the year since AMD announced the 20nm skip involved a port of a 20nm Neo. In that case, Sony started out with a faster cadence, and the post-Neo architecture was genuinely further along in its development before Neo took a significant porting delay.
That scenario might mean among other things that Neo might not be Polaris, but rather the tech that could have been at 20nm. The void where that GPU might have been is outlined on the 28nm side by Tonga and Fiji.
That might mean that Neo has delta compression, but not other architectural features or adjustments to increase clock speed that Polaris shows signs of having. Given that Neo is clocking not much above where the current console GPUs clock...
Power-wise 36 CUs might have been difficult to justify, although it might have been scoped out with the assumption that 20nm wasn't going to be a large disappointment for this power band, and that some of the tweaks done for GCN3 would have helped. As an aggressive premium target, Sony might have also been willing to just burn as much power as the PS3 once did and also align Neo more with PSVR.
This scenario might also introduce some downside for AMD. Possibly some of this extra work had to be done to make up for the failure to get to 20nm, and AMD might have taken a hit in the number of chips it can put out due to reshuffling its limited engineering resources.
For full clarity: this is all just my making things up off the top of my head for the purposes of some of these scenarios.
I don't buy Phil's words about all Scorpio games going to run on the X1. Considering the capabilities of Scorpio & the eagerness of MS to butcher and onslaught the Xbox One, they will soon leave the no one left behind thing. The current console is always going to be a Remora, both technically and commercially.
let's hope your right, because I don't like the idea 8years from now games having to still work on the xo.
how long do people want expect their console to still play every game?
by the time Scorpio comes out xo would've been out close to 4years, I personally don't see many games that would be Scorpio only in the first 2years, but lets just say 1year.
At that time xo would be 5years old, and still playing 95% of all the games.
let's say 2years in 50% Scorpio only, xo would be 6 years old and still getting good support. So on.
most people complained last generation lasted too long, about all the remasters, about the the cross gen development.
Remember the reason for most of those are install base which is the main reason to continue to support both the ps4 and xo, until neo and Scorpio has a very large install base.
some people are complaining that there won't be Scorpio only games, because then your not doing things that you fundamentally can't on the xo.
as they say can't please everyone I guess.
Mid-life crisis? Would explain a lot, both consoles wanting to be more powerful, more valued, more impressive. Expect Scorpio to look like a Harley...
How would the market respond though? I think it'd be negative. There needs to be 3+ years until new hardware, nless it's a parallel console. That or a progression, not a new gen. PS4.5K.
Sadly, I am sure there is a focus group out there that tested that concept and pronounced it sane. I was more referring to people losing trust/ faith in the manufacturers that their up front investment is no longer going to be such a nice long term value proposition.
Yea but 8 gigs has come crashing down the pipe. There are going to be RX 470/480s with them costing under $250 maybe under $200 for an 470 8 gig and considering they seem to be around the 390 series in performance or better they can actually take advantage of the ram amounts. I am sure NVidia's 1060 will also feature 6 gigs of ram. Both lines of cards should be very popular . So adoption of 6+ gig cards will start to sky rocket. In the high end I can easily see 8gigs to 16 gigs becoming very popular over the next year.
Scorpio wont even launch for another year
Then you have system ram which right now is mostly 16 gigs but if you look at steam surveys 32 gigs is starting to gain steam... hehe
But what is the foot print of a game ? Some games are already in the 50 gig range. Quantum break on my pc is almost a 100 gigs I think with the videos . Games will also grow over the live span of the console. So do they put 128 gigs in which could be 2-3 games at launch but will quickly fall to 1 game or maybe even a single game wont fit in the cache ?A SSD which covers the usual footprint of 1-2 games should release a lot pressure in real gaming situation.
How often do you switch between games? I usually play them one after another. Then you reload levels/areas a lot. The hit/miss ratio should be extremely high. I can't really see how such simple uncritical cache scenario would even run into endurance problems
Techreport had a SSD endurance test running for a long time which burried a lot of my preconceptions if you just look at the numbers.
All this could be offered optional.
I switch often. On my pc I switch between my vr games and my non vr games. On my console I have about 4 games that I switch between right now. Garden Warfare 2 , limbo , titanfall and Ki . Limbo is almost done but i'm sure cuphead will replace it or we happy few.
Techreport is doing gods work with the SSD endurance tests. But remember those drives have a lot of nand dedicated to back up also and expensive controllers meant to mitigate writing to often to the same cells.
If your basicly going to put an ssd inside the console , why not just make it your dedicated storage and put in enough of it ? Why put 128-256 gig ssd and then a traditional 2.5 inch drive ? how much more would it cost to just put in the 1TB ssd. The price difference for a consumer between a 256 gig drive in the same line as a 1TB is about $100 at this point.
I still wonder if MS went to AMD and in the contract negotiated that they would have the most powerful custom soc for a console in 2017-2018 or something like that . That way MS knew they wouldn't have a problem with their marketing.
Also I'm sure as I've said before that if sony changes clocks or configuration before neo launches which sounds like the fall. MS has a year to react. I am sure they will adjust disabled portions of dies and clock speeds and maybe even memory amounts to keep their perfered performance difference.
I hope they go with at least 24 gigs of ram as I've said before. I think 12 gigs would be a mistake.
Yes but Neo with ~5GB(for games) @218 GB/s vs Scorpio with ~8GB(for games) @320GB/s will be all that matters in terms of consoles. Microsoft will have the advantage over Sony.
Similar threads
