The challenges, rewards, and realities of a two tier console launch

The topic of this thread is "The challenges, rewards, and realities of a two tier console launch", a mid-generation launch like PS4 Pro, Xbox One X, and Switch Lite are outside the scope of the conversation. After a platform has existed for a few years, it's common for hardware refreshes to come out with added or subtracted features features to meet consumer demand and market conditions. That's not what we are talking about here. This thread is about 2 consoles launched at the same time marketed as the same generation with fairly substantial differences in performance.

I know I made a cheeky remark about Stadia. I don't hate the idea of game streaming, I just don't like Google's implementation. It might get better, though. But I can definitely see the value of Stadia and other game streaming services. Especially regarding MMOs and other games that don't require super low latency. And if they can get latency low enough, streaming would be the ultimate anti-cheat solution for other online games because there would be no user access to game files. I think the next generation of game streaming should include some form of a hybrid solution, though. Like all of the realtime 3d elements are rendered by a server farm but HUD elements and text boxes are handled locally. That way you can read text if you are playing games streamed to your phone.

 

As far as we know at the moment both platforms will allow the same games just for different scaled visual targets and affordable to different customer bases. One who wants the best visual results and has the money to spent for the console, online traffic and the power bill. The other does not. MS surely assumes they can broaden their market this way which hopefully creates a larger market for the new baseline sooner than just with the visual high-end model.

That system has worked for the 1 and X greatly and the X model was well received by its customers. So I seriously can't see what you all arguing about here. To its customers MS had no problems dealing with 2 different visual target platforms. Defining this to be magically "different" when released at the same time(we don't even know that) looks ridiculous to me.

 

There was some suggestion earlier this year and before E3 (via reported rumours but I can't find 'em now) that Anaconda would be 4x faster than current gen, Lockhart 2x. Possible to read too much into that, but the 4X did turn out to be true, and be for CPU. Anaconda has 8 cores, so half that would fit with a single four core CCX. I assume AMD are working on a quad core Zen 2 4xxx series APU for PC, though 8 would be even better (think we'll more likely see that with Zen 3 though).

Hopefully both versions of Scarlett have the full 8 cores, but as games are going to need to run on quad core CPUs for next few years anyway, and you'd save a not inconsequential amount of die area and power (for a small, cheap, and cheap to power/cool system) maybe you could make the case. Again, I wouldn't wish to see this. It would also mean you could run two "Lockhart" instances on a single Anaconda though, increasing hosting density for cloud streaming.

I agree totally about higher than 1440p steaming being a waste for many cases. Being able to host more than one game instance on a socket could be a big win for MS, I think. There may be other ways to achieve this than a quad core Lockhart, of course (e.g. an additional CPU chiplet).

Edit: I'm a big fan of the idea of the same CPU configuration and the same GPU featureset across both models, and think that's more likely and more desirable. I'm just mulling over a couple of ideas I find interesting.

 

Has not been confirmed as far as i know.

 

Even if it hasn't, it's going to be eight cores. It won't be gimped to 4 cores as it isn't going to have more than 8 cores.

 

Why not?
Zen2 cores can get really small, and Microsoft could choose to cutdown on e.g. L3 size to get more CPU througput, especially if they get access to wide and fast GDDR6, or if they #GASP# use two stacks of 3.6Gbps HBM2E for 920GB/s and do away with L3 altogether.

 

No. On consoles GPU is the most important part, more than ever (GPGPU, soon VRS, RT etc.). Look what they have done with tablet CPU in the current gen. I'd say we are already lucky to have 8 Zen 2 cores (and 16 threads !) in those future boxes.

HBM2 lol. This isn't the baseless next-gen thread.

 

Best balance for a console regards price and performance, where CPU power has always been the weakest option. What advantage would 12 cores bring when the lower performance console has less and the cross-platform options all have less? 8 cores Zen 2 as a baseline target (not upper end as on PC) is a whole lot of CPU power for games.

 

I'd advise to read more carefully on the quoted posted.
The hypothesis was not to take away GPU die area to give way to more CPU die area.
L3 on Zen 2 is taking more than half of the chiplet's die area. If they have a fast enough system memory they could do away with L3, making the CPU cores and their L2 occupy around 5mm^2 each. 4 more cores would mean 20mm^2.

The "baseless rumors" thread is to comment on baseless rumors, not to discuss hypotheses.
HBM isn't the cookie monster that some made it out to be. The super exaggerated FUD-driven suggested pricing for HBM adoption has mostly been debunked.

Better performance on a CPU-based hardware RT implementation, for example.
Or going with lower clocks with more execution units for higher power efficiency, leaving out more power to the GPU.


Just to clarify: I dont think Scarlett will have more nor less than 8 Zen2 cores (though like Sony, it might use ARM application processors but that's a separate thing).
In fact, I think there's a large chance AMD will just be selling the same 76mm^2 Zen 2 CCD we know from Rome and Matisse to Sony and/or Microsoft, and pair it with different I/O + GPU dies.

My only point was that Scarlett's CPU core count AFAIK can't be deduced with 100% certainty, yet.

 

Pretty sure MS has confirmed GDDR6 already.
Don't think Sony has said what type of memory? Expecting GDDR6 also though.

 

Yup. @1:29

 

I think because of Lockhart on Scarlett games will looks almost the same as on x one x just with native 4k and sometimes 60fps. We have seen that already whit new Halo Infinite trailer that looks definitively like game for current gen. What worse I predict big popularity of Lockhart if the price will be around 300$ so most of multiplatform games will be similar looking to current gen. Only hope with few exclusives for PS5.

 

More likely then not that both consoles will have just GDDR6 memory.

Agree, this gen didn't have much exclusives, and PS5 might even have less then that.

 

I remember home computers went 2 tier
(eg: spectrum 16k/48k)
great if you had the top tier machine, not so much if you didnt

 

Personally, I can see Lockhart as the backbone of MS next gen streaming service. Compressing Anaconda’s 4K frames is probably a waste on anything not streamed locally. However, the Anaconda sitting in the entertaining center can handle that case.

If your 4K game looks marginally better than a 1080p title then why not just go with a native 1440p resolution. A smaller frame that needs less aggressive compression.

Furthermore, the cost saving on the networked hardware puts less strain on subs to recoup the investment. And if in the case of mobile devices the difference in quality is imperceptible then Anaconda powered servers may represent a ton of wasted investment.

 

If current gen mult-plats are burden by a base console of 1.3 Tflop then why should next gen games continue to look like current titles when the baseline will be greater than triple the performance of the current lowest performance console?

 

Releasing a premium console halfway into a gen may encourage a ton of sales from existing owners, but what tells us those sales are worth cutting in half a premium console’s availability during a gen?

A gamer who buys a $500 console at launch and another $500 premium console a few years later, doesn’t seem like a gamer who would turn their nose up to $600-$700 premium console at launch(given the right specs).

We live in a world where companies readily market and sell console accessories that rival consoles in pricing. I have a hard time believing that high end gamers won’t be willing to buy high priced consoles.

 

I would say current base is ps4 1.84tf, 4tf (even navi) is just too small jump (smallest in console history nextgen vs last generation). Yes we have cpu jump, yes we have ssd but that mean game will have more npc's, stable frames and short loading time but graphic? Small improvement, but probably some jaw dropping on ps5 exclusives.

 

Premium console at launch vs premium after 2-3 years. I dont think that that 100-200€ extra would give the same performance boost as 2-3 years of tech developing.

I rather take 500€ box on launch and truly more powerful after 3 years

 

Some of us think that way... But applied to a next generation. Spend the money and after a time buy a new, more powerful thing when the next generation comes.

I don't like mid-gen enhancements. It feels unnatural to me. The very fact of somehow segmenting the users of the same console, which is not the same at the end of the day.