Next Generation Hardware Speculation with a Technical Spin [post E3 2019, pre GDC 2020] [XBSX, PS5]

Status
Not open for further replies.
I speculate that all Next-Gen Console DevKits are NOT in their Final Form and won't be until Spring 2020. What they're using now is a mix of AMD Navi GPU with separate hardware for the RayTracing aspects. They wont be assembled in a single SOC until next year when RDNA 2 hits.

I think more like devs are using RTX 2000 series instead since it has the HW RT. At least on MS side.
 
With Scarlett's APU being around 390-400m2 does that mean they can fit more CUs but have to keep clocks down?

How does that compare to say PS5 with less CUs but more clocks and both being roughly around 10% of each other. Is that a wash or more CUs vs more clocks, which one is better?

I have to go back on my measurements a bit. I did a control measurement using the methods on a 1X picture and arrived at ~375mm. The Anaconda SOC is at the very least slightly larger than the 1X's. 365-380mm is my range.
 
Everybody should realize that Ray tracing isn't that important. Seriously. There are vastly more important metrics and architectural features that devs have to work with (just the jump from Jaguars to Zen 2 CPUs is 10x more important)...they can always run RT using compute for testing purposes (like DXR support on non Turing etc).
 
Last edited:
I speculate that all Next-Gen Console DevKits are NOT in their Final Form and won't be until Spring 2020. What they're using now is a mix of AMD Navi GPU with separate hardware for the RayTracing aspects. They wont be assembled in a single SOC until next year when RDNA 2 hits.
Wonder how important it is to send out early alpha dev boxes nowadays.
Example MS could just say target spec and DX12, and devs can develop towards that until actual dev boxes built with final silicon is available.
Will just be optimizing against it.
If a dev wants to make use of DXR then use Nvidia RTX cards, Nvidia also supports VRS.

Probably less leeway with PS5.

Everybody should realize that Ray tracing isn't that important. Seriously. There are vastly more important metrics and architectural features that devs have to work with (just the jump from Jgaurds to Zen 2 CPUs is 10x more important)...they can always run RT using compute for testing purposes (like DXR support on non Turing etc).
Running via compute, is that even supported on AMD via DXR even if it would make development painful and slower than necessary?
No one is saying CPU isn't important?
It would be zen 2 based speced.

It's a spec to develop against, or early dev box MS provides.
 
Last edited:
Looking at AMD's new roadmap

noygziviju331.png


they probably use 7nm+.

I think that unlikely. I doubt they’d have development silicon to show off yet.

I have to go back on my measurements a bit. I did a control measurement using the methods on a 1X picture and arrived at ~375mm. The Anaconda SOC is at the very least slightly larger than the 1X's. 365-380mm is my range.

What aspect ratio do you believe the chip to be?
 
What do you mean, show off to who?

To us in the video. Someone pointed out that you can read the GDDR6 chips in the video, and they actually have different densities in some (8Gb vs 16Gb). I think that makes it less likely a render.

Same as the 1X. I have the length at 24mm and width at 16mm.

That surprises me. The chip has a very rectangular aspect ratio from what I can see, and the shallow view angle should only mute that effect.
 
I think today's DevKits are more important for Developers to get used to the insanely-fast loading with the SSD and Memory subsystems. That might have a larger impact on game design than RayTracing.
Doubt that's the case for very early alpha dev boxes.

Also 90% of launch games will be cross gen, so won't take advantage of the ssd apart from loading times.
The alpha dev boxes can be built using pcie4 ssd.

I'm just saying that actual apus may not be available, until then like every other gen, alpha box be based on different hardware.
Difference here is that you would be developing against a stable DX and tool set compared to any other generation

To us in the video. Someone pointed out that you can read the GDDR6 chips in the video, and they actually have different densities in some (8Gb vs 16Gb). I think that makes it less likely a render.
Gotha ya.
I have to admit I assumed render.
But would it really be to early for 7nm+ if RDNA 2 based cards are due next year as is suspected?
 
Last edited:
Doubt that's the case for very early alpha dev boxes.

Also 90% of launch games will be cross gen, so won't take advantage of the ssd apart from loading times.
The alpha dev boxes can be built using pcie4 ssd.

I'm just saying that actual apus may not be available, until then like every other gen, alpha box be based on different hardware.
Difference here is that you would be developing against a stable DX and tool set compared to any other generation


Gotha ya.
I have to admit I assumed render.
But would it really be to early for 7nm+ if RDNA 2 based cards are due next year as is suspected?
Yes, it would be early. Mobile parts just hit HVM within the last few months. Zen 2 is not native 7nm+, so it would have to be ported. MS said they are using Navi, which is natively 7nm as well.
 
Yes, it would be early. Mobile parts just hit HVM within the last few months. Zen 2 is not native 7nm+, so it would have to be ported. MS said they are using Navi, which is natively 7nm as well.
RDNA 2 would need to be ported to 7nm.
Can see more reason for zen 2 to be ported to 7nm+ than RDNA to be ported to 7nm.

I'm assuming RDNA 2 is also called navi. So either Sony or MS could be RDNA or RDNA 2 based.

I've been leaning towards Scarlett being RDNA 2 based, due to VRS, RTRT, etc.

Given me more things to consider. More reasons why could be RDNA based with RDNA 2 features back ported...
Stupid Cross-Generational Games! ;)
I'm with you there :LOL:
 
Looking at PS4 hardware development in interview with Mark Cerny
https://www.digitaltrends.com/gaming/meet-the-guy-who-engineered-the-playstation-4/

"Some of the tech in the PlayStation 4 didn’t exist even a few years ago, at least not on a level that made it practical for consumers. When you first started designing the hardware, how did you plan around tech that may not have been available yet?

We had six years to make the hardware, and it only takes about four years to do the actual engineering, so we had two years to figure out what we wanted to make the PlayStation 4. And we looked at a huge variety of technologies, including some that were just coming into possibility. In the end we decided that we could hit sort of a sweet spot by working with AMD’s PC technology, the CPU and GPU, and then aggressively enhancing the GPU for that long-term growth."


"Did the hardware development determine the release date, or did the release date determine the development cycle?

We’d been targeting Holiday 2013 for many years, but with hardware, you never quite know if you’re going to be able to reach the date. If you screw up something in your main custom chip, you could easily find that it takes six months to fix. And then you end up out of your targeted launch year."

"So you finalized the hardware two years ago?

We finalized the spec of the hardware. The process of creating the hardware is about four years. Two years into that it’s locked enough that you can start talking about all this other stuff that’s going to surround it. It’s not like your work actually finishes at that time, or that was any particular milestone. But it was sort of an “ok guys, we know what the hardware is now.”

If we assume that this development process also applies to PS5 and maybe Scarlett, will this enable us to narrow down the possibilities for next gen console hardware in the light of AMD's market availability for RDNA 1, RDNA 2 and Ray Tracing architecture?

I like to be surprised rather than disappointed when we finally get the juicy specs :)
 
Looking at PS4 hardware development in interview with Mark Cerny
https://www.digitaltrends.com/gaming/meet-the-guy-who-engineered-the-playstation-4/

"Some of the tech in the PlayStation 4 didn’t exist even a few years ago, at least not on a level that made it practical for consumers. When you first started designing the hardware, how did you plan around tech that may not have been available yet?

We had six years to make the hardware, and it only takes about four years to do the actual engineering, so we had two years to figure out what we wanted to make the PlayStation 4. And we looked at a huge variety of technologies, including some that were just coming into possibility. In the end we decided that we could hit sort of a sweet spot by working with AMD’s PC technology, the CPU and GPU, and then aggressively enhancing the GPU for that long-term growth."


"Did the hardware development determine the release date, or did the release date determine the development cycle?

We’d been targeting Holiday 2013 for many years, but with hardware, you never quite know if you’re going to be able to reach the date. If you screw up something in your main custom chip, you could easily find that it takes six months to fix. And then you end up out of your targeted launch year."

"So you finalized the hardware two years ago?

We finalized the spec of the hardware. The process of creating the hardware is about four years. Two years into that it’s locked enough that you can start talking about all this other stuff that’s going to surround it. It’s not like your work actually finishes at that time, or that was any particular milestone. But it was sort of an “ok guys, we know what the hardware is now.”

If we assume that this development process also applies to PS5 and maybe Scarlett, will this enable us to narrow down the possibilities for next gen console hardware in the light of AMD's market availability for RDNA 1, RDNA 2 and Ray Tracing architecture?

I like to be surprised rather than disappointed when we finally get the juicy specs :)

The first quote is really key. They brought things like more ACE, ID buffer, and RPM forward in the PS4 and Pro GPUs. I wonder what else besides RT they might bring forward from RDNA 2.0+? VRS, Micropolygon, etc.?
 
I don’t understand how an entirely different image at a different perspective gives you a basis measurement. Can you explain?

I did some basic math using the perspective lines. I reached at ~373mm for the 1X SOC. The Anconda SOC is nearly the same length as the 1X's, but it's a tiny bit wider.
 
Status
Not open for further replies.
Back
Top