Baseless Next Generation Rumors with no Technical Merits [pre E3 2019] *spawn*

[iroboto]

As for RCC, is that some sort of abbreviation for Raytracing? Like Raytracing Compute Core? Raytracing Central Core?

The A4 document says RCC = Ray Tracing Chiplet

 

[BRiT]

The A4 document says RCC = Ray Tracing Chiplet

Oh, is that what that said?

 

[iroboto]

Oh, is that what that said?

Yea. Ray tracing custom chiplet to be precise.

The measurements in flops is awkward to say the least.

 

[iroboto]

It's actually my college dormitory resident assistant. Sorry I'll be more clear in the future.

I’m having a tough day sorry mate.
I’m easy to troll today please spare me.

What’s going on with all the thread on patents and things (context wise) ? I went through some twitter posts that you posted with blue— (trying to figure out context) and not all of it is lining up with things i actually know.

I usually don’t invest too much into patents, sometimes they are just that. Many of my MS buds have patent cubes on their desks, not all of them make it into production.

 

[Jay]

Been away couple days and come back to this, got some catching up to do.
Interesting, although couple things I don't like one that hasn't been mentioned is
12gb Lockhart
24gb Anaconda

Why the huge difference? Really saying need around 10gb for better textures?
Such a huge difference would make scaling pretty hard

 

[anexanhume]

I’m having a tough day sorry mate.
I’m easy to troll today please spare me.

What’s going on with all the thread on patents and things (context wise) ? I went through some twitter posts that you posted with blue— (trying to figure out context) and not all of it is lining up with things i actually know.

I usually don’t invest too much into patents, sometimes they are just that. Many of my MS buds have patent cubes on their desks, not all of them make it into production.

Sorry, I was only teasing because BRiT had already given the correct answer.

 

[turkey]

Been away couple days and come back to this, got some catching up to do.
Interesting, although couple things I don't like one that hasn't been mentioned is
12gb Lockhart
24gb Anaconda

Why the huge difference? Really saying need around 10gb for better textures?
Such a huge difference would make scaling pretty hard

It seems to come from the idea Anaconda is double Lockhart, now it could be they need the extra chips for bandwidth but I think it's probably a mistake in that whoever wrote it thought 4k was twice 1080p.

As it is, half the flops for a quarter of the work seems rather generous.

 

[BRiT]

Yeah, but the major flaw in all those is the base amount of memory is too far apart where games would have to be too drastically changed for the different memory models.

 

[Jay]

yea, I get that part, but the way they're talking about Azure isn't so much in the same context that they tried with XBO. That was weak attempt at trying to use the cloud as this augment for a lack of graphical performance.

The azure discussion seems based more around setting the soc up for datacenter usage, ie they want to be able to stream xbox games to clients, but when it's not being used they want the idle compute for their azure business.

I would believe this scenario much more than cloud augmentation of games because it decreases their risk and cost in data centre builds.
TLDR; it's not something they really should be selling, or marketing.

still referencing this quote
“The thing that’s interesting for us as we roll forward, is we’re actually designing our next-gen silicon in such a way that it works great for playing games in the cloud, and also works very well for machine learning and other non-entertainment workloads. As a company like Microsoft, we can dual-purpose the silicon that we’re putting in.

We have a consumer use for that silicon, and we have enterprise use for those blades as well. It all in our space around driving down the cost to serve. Your cost to serve is made up by two things, how much was the hardware, and how much time does that hardware monetize.

So if we can monetize that hardware over more cycles in the 24 hours through game streaming and other things that need CPU and GPU in the cloud, we will drive down the cost to serve in our services. So the design as we move forward is done hand-in-hand with the Azure silicon team, and I think that creates a real competitive advantage.”

interview: https://twinfinite.net/2018/12/phil-spencer-buy-ea-next-gen/

This is part of why I think it may not be navi. I think internally this is a bigger factor for MS then competing with Sony.

Stadia chose to run with Vega. And that's a pure cloud based streaming service. And so as per my commentary on the other thread, implying MS has something different from Navi doesn't imply they're getting a higher performing console; Sony chose Navi perhaps because it was best for console, but Google chose Vega, perhaps that was best for pure server builds.

Perhaps MS is somewhere in-between or something entirely custom because the product they want doesn't quite exist.
@Jay

Sorry, not been about.
Yea, I think that the console is maybe the most important part but not the only part they are considering.
I think something like streamed remote desktop that they have just released recently is a good example. Remote desktop from cloud would perform a lot better running on Scarlett xcloud as that is tailored for latency for example, so would give a lot better experience.

The gpu will be custom but I can't see anyway possible it's base isn't navi or what follows navi. Based on original time lines post navi was due 2020 I believe.
I'm not saying I don't think it's navi, just throwing out the possibility it might not be, and it wouldn't be such a crazy outcome.

 

[Jay]

As it is, half the flops for a quarter of the work seems rather generous.

It is generous for 1080p, but I can see couple reasons why would do this

1. Call it 1080p as that's what people understand and easier to talk about. But it's a range of 1080p-1440p with it being closer to 1440p and just scaled to appropriate output resolution. Will give good results on both 1080p & 4k displays.
2. When Anaconda is dropping res to gain performance say 1800p, Lockhart won't need to drop below 1080p so will hold up in the future pretty well.
3. Disabling more than 50% doesn't improve yields/bining, so may as well leave it over powered.

Yeah, but the major flaw in all those is the base amount of memory is too far apart where games would have to be too drastically changed for the different memory models.

Yea, I can understand that improving bandwidth. But the golf in memory size would have fundamental impact on what can be done in game design.

 

[BRiT]

Yea, I can understand that improving bandwidth. But the golf in memory size would have fundamental impact on what can be done in game design.

I could understand a 4GB difference for Game memory because of resolution differences, but not 50%. Though things could get trickey as there's only so many combinations of memory configuration to get around the same bandwidth and capacity when dealing with a single memory pool. Maybe they just drop off a couple memory channels and 4GB? I don't know if that would have too drastic of an impact on general memory. This is where things could be simplier if they used separate physical memory pools for RAM and VRAM. I dont think the tradeoffs for that are worth it though.

 

[Jay]

CPU chiplet congratulation.
Reasons of cooling would seem unlikely.

What would sound reasonable to me though is the chip allowed for a 2 chiplet configuration, then the version for cloud would have 2 chiplets for 16c32t, consoles 1 chiplet for 8c16t.

Guess having possibility of 2 chiplets could allow MS to say just give use any combination that adds to 8c16t for cheapest you can. So can take AMD's lower end chips eg ones that all cores aren't functional. Although this sounds just as way out there

 

[Deleted member 91098]

[...]

Yea, I can understand that improving bandwidth. But the golf in memory size would have fundamental impact on what can be done in game design.

I guess it has to do with the claimed usage of Anaconda as the base hardware for xCloud, and that it can stream two 1080p Lockhart streams, which would mean they need 24 GB memory since the claimed Lockhart would have 12 GB. The xCloud hardware on the spec sheet would also have16c/32t (so double of what the consoles would have).

The specs however were already shot down by someone who seems to be accepted/semi-verified as a xbox insider. To quote resetera:

Is this legit btw ? I m quite disappointed to be honest .from HDD still being there in combination with SSD and those gpu TF and cpu clocks.i m lowering back my ps5 gpu power to 10 TF .I was hoping it would be closer to 12 TF but if this is anacando then 10 is also optimistic for ps5.

Absolutely not.

The insider however said this was well:

"Cloud Native" is a thing.

Which could mean similiar sharing like the faker proposed could still happen.

 

[Jay]

guess it has to do with the claimed usage of Anaconda as the base hardware for xCloud, and that it can stream two 1080p Lockhart streams, which would mean they need 24 GB memory since the claimed Lockhart would have 12 GB. The xCloud hardware on the spec sheet would also have16c/32t (so double of what the consoles would have).

The specs however were already shot down by someone who seems to be accepted/semi-verified as a xbox insider. To quote resetera:

I didn't think this was true, I just find it interesting to use things like this as a basis of discussion. Until we get proper leaks anyway.

The point is scaling specs from a common design for different usage. Anaconda having twice the amount of memory than Lockhart wouldn't make any sense to me, cloud would.
The cost, the fact that it would make game design usage hard for the consoles, etc.

Which could mean similiar sharing like the faker proposed could still happen.

cloud native was mentioned by MS I believe.

Oh literally just remembered in what context.
In the current xcloud set up, devs can make a game cloud native. From what I remember it means it's specifically takes into account that it's running in the cloud, etc

 

[bgroovy]

What would sound reasonable to me though is the chip allowed for a 2 chiplet configuration, then the version for cloud would have 2 chiplets for 16c32t, consoles 1 chiplet for 8c16t.

Yeah, that might make sense. But having 8 cores spread across two chiplets is silly, especially when you already can see performance hits if the two quad core elements in a single chiplet have to communicate.

 

[ultragpu]

I guess it has to do with the claimed usage of Anaconda as the base hardware for xCloud, and that it can stream two 1080p Lockhart streams, which would mean they need 24 GB memory since the claimed Lockhart would have 12 GB. The xCloud hardware on the spec sheet would also have16c/32t (so double of what the consoles would have).

The specs however were already shot down by someone who seems to be accepted/semi-verified as a xbox insider. To quote resetera:



The insider however said this was well:



Which could mean similiar sharing like the faker proposed could still happen.

Haha, the length some attention seeker would go by hand writing all 8 A4 pages of them just to what, fool the forum goers? Not too long now, E3 shall do a cleansing of all those rumors.

 

[RDGoodla]

If this is true it seems like a hybrid drive with a huge SSD cache (512 GB) in xbox2, which cannot support zero loading time. GPU is not some secret sauce. How can next xbox be more powerful than competitors? Cloud CPU computing again (just like "4x CPU for xbox1")?
https://www.mcvuk.com/microsoft-cloud-makes-xbox-one-four-times-more-powerful/

 

[Jay]

GPU is not some secret sauce. How can next xbox be more powerful than competitors?

You've not said what your basing this on?
Example: specs of gpu, type and amount of memory, accelerators, customisations?

Or are you saying based on ps5 having 1TB SS, it will be more powerful? (even though we don't know what Scarlett is doing)

 

[Deleted member 91098]

I didn't think this was true, I just find it interesting to use things like this as a basis of discussion. Until we get proper leaks anyway.

The point is scaling specs from a common design for different usage. Anaconda having twice the amount of memory than Lockhart wouldn't make any sense to me, cloud would.
The cost, the fact that it would make game design usage hard for the consoles, etc.
cloud native was mentioned by MS I believe.

Maybe it's a result of the required bandwidth like @turkey aluded to? I've looked at the current consoles for that and unlike discrete GPUs they all stay above 50 GB/s per TFLOP.

xbox one   ( 68 GB/s | 1.310 TFLOPs | 51.91 GB/s per TFLOP | doesn't inlcude ESRAM bandwidth)
ps4        (176 GB/s | 1.843 TFLOPs | 95.65 GB/s per TFLOP)
ps4 pro    (218 GB/s | 4.130 TFLOPs | 51.90 GB/s per TFLOP | + delta color compression)
xbox one x (326 GB/s | 6.001 TFLOPs | 54.33 GB/s per TFLOP | + delta color compression)

The fake would fit into that:

Lockhart               (336 GB/s |  5.587 TFLOPs | 60.14 GB/s per TFLOP)
Lockhart + RCC         (336 GB/s |  6.587 TFLOPs | 51.01 GB/s per TFLOP)
Anaconda               (672 GB/s | 11.174 TFLOPs | 60.14 GB/s per TFLOP)
Anaconda + RCC         (672 GB/s | 13.174 TFLOPs | 51.01 GB/s per TFLOP)
------------------------------------------------------------------------
Anaconda 352 bit       (616 GB/s | 11.174 TFLOPs | 55.13 GB/s per TFLOP)
Anaconda 352 bit + RCC (616 GB/s | 13.174 TFLOPs | 46.76 GB/s per TFLOP)
------------------------------------------------------------------------
Anaconda 320 bit       (560 GB/s | 11.174 TFLOPs | 50.12 GB/s per TFLOP)
Anaconda 320 bit + RCC (560 GB/s | 13.174 TFLOPs | 42.51 GB/s per TFLOP)
------------------------------------------------------------------------
Anaconda 288 bit       (505 GB/s | 11.174 TFLOPs | 45.19 GB/s per TFLOP)
Anaconda 288 bit + RCC (505 GB/s | 13.174 TFLOPs | 38.33 GB/s per TFLOP)
------------------------------------------------------------------------
Anaconda 256 bit       (448 GB/s | 11.174 TFLOPs | 40.09 GB/s per TFLOP)
Anaconda 256 bit + RCC (448 GB/s | 13.174 TFLOPs | 34.01 GB/s per TFLOP)

I guess the more bandwidth the merrier, but is it actually some hard limit for consoles with a shared memory pool to be over 50 GB/s per TFLOP? Maybe due to the fact that CPU bandwidth seem to reduce GPU bandwidth disproportionately like Sony says?

Slide 13: http://rdwest.playstation.com/wp-content/uploads/2014/11/ParisGC2013Final.pdf

Oh literally just remembered in what context.
In the current xcloud set up, devs can make a game cloud native. From what I remember it means it's specifically takes into account that it's running in the cloud, etc

That makes sense. Though I have to admit that I had hoped for more meaning behind that "insider" sentence.

Haha, the length some attention seeker would go by hand writing all 8 A4 pages of them just to what, fool the forum goers? Not too long now, E3 shall do a cleansing of all those rumors.

Apparently it's 11 pages written text plus 5 pages data/drawings, so they had to make 16 A4 pages. Assuming their middleman tells the truth and really sat down with the faker then that would mean the faker had to train and memorize this stuff in order to answer questions...

But I'm not surprised anymore after the past leaks where people created 3d printed nintendo controllers (if my memory doesn't deceive me I think there was some 3d printed controller for Durango before the PS4 launch as well) and used 3d rendering to fool people.

 

[Jay]

Maybe it's a result of the required bandwidth like @turkey aluded to? I've looked at the current consoles for that and unlike discrete GPUs they all stay above 50 GB/s per TFLOP.

It did allude to it also being for bandwidth, but not the way I'd go personally.
I would go for same width bandwidth for both, and just less and slower memory modules in Lockhart.
If someone says that the width is to expensive and complicated, for the bandwidth required, then I'd still think there was a better way than just doubling it.