Next Gen GPU architecture GCN, RDNA, Navi 10/20 (PS5 Navi Hybrid, Xbox Navi Pure) *spawn*

Hypothetical launch dates. I'm going to move this tangent into another thread as it hasn't led to technical discussion really.

 

Yea. And if we’re discussing hypothetical launch dates, they can still
Launch as early as April as per their investor report, but I would not be surprised if the truth of a oct 2019 launch date was accurate.

That would have been substantial.

 

As far as naming of architectures/generations is concerned, it’s simply an etiquette.
NVidia gives each revision a separate name (Kepler-Maxwell-Pascal-Volta-Turing) as does Intel (Nehalem-Sandy Bridge-Ivy Bridge-Haswell-Broadwell-et cetera) whether the changes are significant or not.
AMD has gone with version numbers in both GPUs (GCN1-2-3-4-5) and CPUs (Zen1-2-3-).

I think forum traffic generally indicates that AMD is making a marketing error, and would be better off giving all their revisions fanciful names.

 

They should name the APU family for consoles Super Blast. Then all the new consoles can have Super Blast Processing.

 

Except Nintendo...

 

Last Processing

 

Last gen processing.

 

Assuming the rumor its true pure RDNA will always be better, will have a more modern feature set and be more efficient.
In the last 20 years of GPU history i can't come up with a single example where the new arch isn't better than the old one.
In this case it not just any arch change, its a drastic one after nearly a decade of GCN

The other reason is Navi 10 (48CUs) doesn't make any sense for a 350mm2 console APU on 7nm where you could fit 80CUs. You'll get much better perf/watt with more CUs clocked lower than less CUs clocked higher
The smallest is see them going is with 56CUs enabled.

Sony wouldn't make the Wiiu mistake of compromising next gen performance for pure hw emulation, if it comes to it they are much more likely to invest a couple million on a software layer that assists

There can be many reasons why MS might go with Vega for the nextbox that doesn't make it likely or even realistic.

I haven't seen any indicating Sony might go with a hybrid design, quite the opposite rumors were claiming Navi was designed for Sony

Corny said they believe in clear cut gens, their goal is to deliver true next gen experiences and for that every bit of performance matters, architecture and all the new features and ways to more efficiently render games play a big role in it. Especially importan in consoles that have strict TDP budgets new arch efficiencies and features matter where developers can specifically target those.

Why would they go the next decade of gaming with an arch that will become obsolete within a year? It could be as big a screw up as going with RSX just when unified shaders are available. Im sure Sony is very conscious of that.
I don't know whether SNEK will me more powerful than PS5 or not that's up in the air, what im certain off is Sony will go with the latest arch possible including future features on their GPUs. They've show this with PS4 and PS4 Pro

I just doesn't seem likely for Sony to rush next gen and gimp their console in the process, they currently hold the high ground. MS has more reason to rush to next gen in the hopes in recapturing market share.

 

That's Mr Cerny for us, thank you very much. 'Corny' is reserved for the inner circle: mainly Shu, Shawn, Kaz and Lisa.

 

Well that seems to be missing the point of goals. A larger GPU on an older architecture can outperform a smaller GPU on a newer architecture. There are many scenarios here that wouldn't necessarily lead down the path that every APU combination needs to be exactly the same.

You'll also need to increase the pipeline everywhere to accommodate for the increase in width or you'll bottleneck your pipeline and waste all that performance. Clockspeed speeds the entire pipeline together making everything faster, bottlenecks move together, and your silicon budget is saved by not needing to be wider everywhere.

You should look at PS3. Every product has its place, without understanding the goals of the company and their product, any company can and will compromise for a product to meet its goals. MS did it with Xbox One, and Nintendo does it for their console.

Well that's actually why this thread spawned to begin with. It's just speculation that fit other speculation. If you take the rumours claiming that Navi were designed for Sony, then obviously the pure variant of Navi couldn't be the one they are referring to. It was never going to be ready until 2020, and Sony missing out on E3 for 2019 among the rumours that Sony aimed for 2019 but had to delay to 2020. They released their 4Pro in 2016.

Then if we follow the rumours that Navi 10 is indeed some hybrid GCN/Navi device and it releases this year; the same year that Sony could have been aiming for launch, then the Navi 2020 model was always out to begin with.

And looking back that the way 4Pro was designed, which is largely 2 PS4s taped together with some Vega features, that was what they needed to accoplish ot sustain their BC with PS4, and boost mode is still not applicable to everything.

The rumours about needing more engineers would make sense if they are attempting to rush for a 2019 release as well. And if the design is 'exclusive', then that might also make sense since MS might only have access to the pure variant of Navi which was slated for release in 2020, and they plotted along releasing X1X in 2017.

No one is saying any of this is real. The reason this thread is spawned is because we've largely ignored/in disbelief of all the rumours about Sony and Navi that has come and gone prior to their reveal announcement article. When this rumour came out, we looked back and see if it all sort of fit together. And surprisingly, it's not half that terrible. Doesn't mean it's true. It's just not half that terrible of a story. With Sony delaying their console to 2020, who knows what it is now, its likely not what we're discussing.

I think we're all confident that Sony can make a great machine and they have the competence to do so.
But.
a) Eventually you need to release hardware, next year something better is always coming along
b) If you planned for 2019 at a specific price point to come out 1 year ahead of your competitor, then you need to choose the technology for it, and if it's not ready than w/e
c) As many write, power is not everything, exclusives are, at least for Sony's folks this is what matters. So the emphasis on power doesn't seem all that important as long as you have a fairly competent machine.
d) Positioning of your product matters. maybe it's designed for this or that. Maybe it was never designed with BC in mind. Maybe it was designed for VR in mind. You won't know until Sony lifts the curtain off the hardware. For now everyone just assumes it just needs to play games very good, which is a safe assumption, but we haven't seen this to be true every generation.

For Sony to hold the high ground and release a console 1 year ahead of their competitor would secure them the next. Even if it weren't more powerful. We see that today, X1X hasn't put a dent in PS4 sales. Obtaining that critical mass is what continues to drive the success for PS4 and a well timed console launch accompanied with TLOU2, DS, Ghosts and others (CyberPunk 2077) during the launch window is unheard of while still holding a full 12 months ahead of your competition is somehow considered bad strategy; because you have a hybrid architecture over something pure? I'm a bit surprised to hear that. They can release a mid-gen that is the full architecture if they wanted to 3 years after.

 

boipucci said "Sony wouldn't make the Wiiu mistake of compromising next gen performance for pure hw emulation". Sony didn't do that, they literally included chunks of PS2 hardware in PS3 (later engineered out) just as they did with PS1 hardware in PS2. This compromised the product in terms of cost but not performance.

Nintendo definitely do this but I'd challenge you to provide evidence for Microsoft making Xbox One decisions compromised "performance". Xbox One's compatibility relies entirely on software rather than hardware, which is why you can't just play any game, only those games Microsoft have made an effort to support.

 

Maybe he's referring to the following: Microsoft had a $500 Bill of Materials for a lower performance system compared to Sony's offering because of decisions with Kinect.

 

Not referring to BC specifically. Just on the point that optimizing for absolute maximum performance is not always the goal for consoles and that they would be willing to make some compromises to reach other goals.
But I was making mention to RSX and it’s pairing with Cell for instance.
Vs pairing with the most advanced GPUs

 

I can’t see how Sony was aiming for 2019 It’s common sense, why would Sony release the ps5 so soon after the ps4 pro ? They don’t need to jump in before the next Xbox . They’re the market leaders by a long way. If Navi is a hybrid set for the ps5,( meaning Navi 1.5) it’s possible that Navi 20 is only suited to large PC GPUs.
And l don’t believe the ps5 Navi is gimped at all to be B/C. It just has to be efficiency in all areas given the constraints that they have to work with, it’s not all sauce and no hotdog.

 

4Pro was not a reset button.

Momentum & PS4 overall sales trending towards the 100M mark (at the time when planning & decisions were to be made).

edit: i.e. consider the atypical high rate of sales for 6 years meaning shorter time to market saturation and the need for such decisions to be made years in advance.

 

Arguably decisions on cost affect decisions on performance. So it isn't so black and white either way. If they did not include BC in either PS2 or PS3, would they have invested that more into improving the performance characteristics of those consoles?

Of course, there is no way to say definitively either way although I'm sure people could come up with reasons they would have gone one way or the other.

Regards,
SB

 

People are too caught up in thinking Sony and MS give a damn what each other do. They have may effect each other in some ways, but that doesn't mean you base all the decision making around beating your opponent.

If we're being real, winning for Sony , MS and Nintendo, is making massive return on investment. They'll work together or individually to secure that income. Competition is only about securing more of it. If you aren't able to secure solid return on investment you'll leave the stage.

If MS didn't exist, Sony still needs to drive the next generation. The driving force for new generations and new hardware is to get the spending cycle to recur again. If Sony is eyeing profits we are near the tail end of the generation and so releasing a new device will cause more spending to happen.

Calculations for ROI is fairly straight forward. You look at the total TCV (total contract value) of a customer for Sony. Which can be calculated by the average revenue from the attach rate of games and services for say 6 years time in the Sony ecosystem. Once you're hitting the 6th year, your TCV is completed, and your projections for profits and revenue drop. So what better than to start another generation and re-calculate your project profits for another 6 years of TCV and start subtracting off any subsidies or costs required to gain a customer.

You take this concept and you look at how Sony positioned itself, a launch window filled with some of the most desired PS4 exclusives and being backwards compatible with their older library; they can effectively reduce their cost of acquisition to gain new customers by riding off the hype for the tail end of their generation. This will result to a roaring start on securing their next generation of profits during a time in which they can be 12 months ahead of their competition; meaning they have more ample time to gain new customers than they otherwise have to compete against MS for - since they are 12 months behind in hardware and their games still won't be ready yet.

Perhaps that strategy changed, I'm not sure. But to say this is an invalid or poor strategy seems far too focused on power being the main delivery agent for being the biggest driver of securing a base and therefore profits.

 

My question is, if PS5 was to launch in 2019, what particular first-party software would have been ready on justifying a $499 purchase when Pro is still selling quite well at $399 MSRP? I mean what, a bunch of cross-generational titles that are anchored to OG-PS4 tech? For what purpose? A more native 4K PS4 gaming experience? And it really doesn't make any sense when PS4 (even without any official price cuts) hardware sales are still quite high, and many of their first-party teams are still in the process of releasing PS4 software this year and the next.

If anything got delayed would have been Sony's attempt on releasing another streaming system (which are outlined in many patents dealing with a new streaming platform) possibly this year. But their infrastructure on supporting such a new cloud-gaming system entailing such features as cloud assisted AI, ML, physics, and so-on, would have probably swamped PSN services, which already services some 90M users. Hence, their investors meeting outlining their push on better cloud gaming services solutions. And the more likely reasoning for delaying such a system was in-lieu of a possible partnership with Microsoft towards cloud gaming. Just saying...

Edit: I hate auto-correct...

 

Product launches are complex and aren't necessarily down to a single factor.

 

Yes. But the current reasonings are based on a rumor or conjecture, not actual proof. Heck, I find the Scorpio Engine II (Vega) to be more plausible than this.