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

[Silent_Buddha]

So a user who has been vetted for inside information has revealed that PS5 was due to launch in 2019 (around this time) but they were held up by BC issues.
According to the poster, "It is my belief that this is one of (but not the only) reasons Sony delayed PlayStation 5 to holiday 2020."
https://www.resetera.com/threads/ne...ly-fan-noise-and-hot-air.129193/post-25291605

IIRC I think @DSoup made this exact prediction. 2019 delayed to 2020 for BC. I suspect being 2017, ray tracing must have been a point as well, DXR demos were revealed by end of 2018. And we know that there is no RT until RDNA2 which is releasing next year.

Depending on how accurate all of this is, this does bring into light why Sony brought in more engineers etc as we discussed.

There's a few of us that basically said it was likely that Sony intended to release the PS5 Holiday 2019.

However, the reasons for it vary. I still believe it was delayed due to AMD being late with NAVI, especially when it comes to the inclusion of some form of hardware RT assist.

Hence, regardless of the state of supporting PS4 titles on PS5, Sony wouldn't have been able to launch this year if they wanted to use RT capable Navi GPU cores.

On top of that, using Ryzen 2 CPU cores would have been cutting it close although it probably would have been feasible to include in a console releasing this year as long as they didn't mind not having final SOC silicon until relatively late in the year.

Regards,
SB

 

[Jay]

Technically speaking it's never been a reasonable comparison at all. That's why we got the whole nvidia flops vs AMD flops. TF is a single number performing 1 major function of what the GPU must do. We make assumptions that the rest of the pipeline and bandwidth is beefed up equally to support the TF, but we know that's actually not the case. But we use the term anyway. Sadly.

It's a reasonable comparison within the same architecture and a lot of the time derivitves.
Depending on how customised the next consoles are TF will give a specific measure of an aspect of the pipeline also. But as I said, not in isolation. It's useful for us here to measure a specific thing.
Current gen consoles measuring TF, there isn't much wrong with it. Each console pretty much falls in line rendering wise in relation to the TF.

Moving forward, I could easily see MS/Sony use a different metric like RDNA Ray's to take into account RT.

Your comparing Nvidia, AMD, different generations etc, that's where it's meaning becomes a lot less meaningful.

 

[Deleted member 13524]

There's a few of us that basically said it was likely that Sony intended to release the PS5 Holiday 2019.

However, the reasons for it vary. I still believe it was delayed due to AMD being late with NAVI, especially when it comes to the inclusion of some form of hardware RT assist.

Hence, regardless of the state of supporting PS4 titles on PS5, Sony wouldn't have been able to launch this year if they wanted to use RT capable Navi GPU cores.

On top of that, using Ryzen 2 CPU cores would have been cutting it close although it probably would have been feasible to include in a console releasing this year as long as they didn't mind not having final SOC silicon until relatively late in the year.

Regards,
SB

If the decision to postpone the PS5 to 2020 really was made in 2017, then I doubt they'll use the same SoC that could have come out this year.

 

[JoeJ]

So unless RDNA2 is bringing dedicated INT32 ALUs like Volta/Turing, you'll always need more TFLOPs on AMD's architecture to achieve the same FP+INT throughput.
As it stands, a 7 TFLOPs RDNA GPU will not get the same practical perfromance as a 7 TFLOPs Turing. Which is why you need a ~9TFLOP 5700XT to be competitive against a 7.5 TFLOPs RTX 2070.

Makes sense, but is not necessarily true. For example, GTX480 had exact same compute performance as GTX670 for me, but 670 has twice the TF.
I don't know how Turing performs, and i also don't know about Navi, but architectural changes do not always result in better performance everywhere, even if any changes always look promising on paper.
If RDNA is sort of AMDs Kepler, then 7TF could be even less raw power than we think, but can be also the other way around.
But 7TF sounds low, no matter how we talk about it.

As tot the technical debate, how much does RT hardware add to RTX? Isn't it something like 10%? What could be achieved on a large GPU with 50% of it dedicated to raytracing? Someone compare the rumoured die-sizes (300 mm²) with other GPUs and determine how much of that die would likely be 7TFs of compute and give us a die size for the RT side of things. That'll give some impression of what might be going on under the hood of this hypothetical machine.

If they push RT so heavily, could the low TF bottleneck shading?
Personally i think ist's a no if we target GI, but a yes if we want accurate sharp reflections.

However, 7TF could be also a compromise to keep the pricepoint. (SSD added)

 

[AbsoluteBeginner]

If the decision to postpone the PS5 to 2020 really was made in 2017, then I doubt they'll use the same SoC that could have come out this year.

Not like anything changed. They would still aim for same manufacturing process, so decision of moving release date shouldnt impact size of chip.

Now, if 2020 was 5nm and they aimed for 7nm in 2019 then I would agree.

 

[Deleted member 13524]

Not like anything changed. They would still aim for same manufacturing process, so decision of moving release date shouldnt impact size of chip.

2020 allows for N7+ or at the very least N6. If TSMC's forecasts are anything to go by, we won't be seeing many new chips using 7nm in 2020.

It could even mean that a holiday 2020 release allows for 5nm, since TSMC is on track for 5nm mass production in Q2 2020.

 

[Nisaaru]

On top of that, using Ryzen 2 CPU cores would have been cutting it close although it probably would have been feasible to include in a console releasing this year as long as they didn't mind not having final SOC silicon until relatively late in the year.

You assume the CPU release dates are defined by how "finished" the designs are than have more to do with process technology, cost, marketing/product stategies and maybe even timed verification/test windows to catch unlikely problems related to esoteric setups(MMU/IO/Cache/Server/OS usages) not reasonable for consoles.

 

[AbsoluteBeginner]

2020 allows for N7+ or at the very least N6. If TSMC's forecasts are anything to go by, we won't be seeing many new chips using 7nm in 2020.

It could even mean that a holiday 2020 release allows for 5nm, since TSMC is on track for 5nm mass production in Q2 2020.

AMD already confirmed PS5 chip is made on 7nm so no doubt its not 5nm. 7nm+ I guess it could happen, although it yields ~10% in performance, its no silver bullet.

 

[Deleted member 13524]

AMD already confirmed PS5 chip is made on 7nm so no doubt its not 5nm.

They did? Do you have a source for that?

 

[AbsoluteBeginner]

They did? Do you have a source for that?

 

[Deleted member 13524]

Oh ok then, I stand corrected.

Seems like a lost opportunity to not use 5nm, though this might still be an opportunity for them to launch a Pro console at the same time.

 

[Proelite]

Whatever PS5 that's delayed from 2019 to 2020 by a decision from 2017 will be identical at the silicon level.
The SSD might be upgraded. The ram amount can be increased by higher density (bus width would be set in stone) and ram speed increased via faster GDDR6. Clocks might also be bumped up by a modest amount on the APU.

But the APU silicon itself won't have any updates.

 

[Deleted member 11852]

IIRC I think @DSoup made this exact prediction. 2019 delayed to 2020 for BC. I suspect being 2017, ray tracing must have been a point as well, DXR demos were revealed by end of 2018. And we know that there is no RT until RDNA2 which is releasing next year.

Wow, good memory. I did indeed speculate this almost a year ago. Mostly I was figuring B/C potentially being the swing factor but it could have been a bunch of technologies that Sony may have felt it worth hanging out another year for - all to eliminate any erosion of their PS4 customers base who I assume they want/expect buy PS5.

 

[BRiT]

So a user who has been vetted for inside information has revealed that PS5 was due to launch in 2019 (around this time) but they were held up by BC issues.
According to the poster, "It is my belief that this is one of (but not the only) reasons Sony delayed PlayStation 5 to holiday 2020."
https://www.resetera.com/threads/ne...ly-fan-noise-and-hot-air.129193/post-25291605

IIRC I think @DSoup made this exact prediction. 2019 delayed to 2020 for BC. I suspect being 2017, ray tracing must have been a point as well, DXR demos were revealed by end of 2018. And we know that there is no RT until RDNA2 which is releasing next year.

Depending on how accurate all of this is, this does bring into light why Sony brought in more engineers etc as we discussed.

We talked about it earlier, September 28, 2018. Here's my post on that part, https://forum.beyond3d.com/posts/2044232/

 

[AbsoluteBeginner]

Sony delaying launch of console would be nothing new. PS5 also got delayed and I doubt they didnt know about delay before 2005.

 

[Nisaaru]

AMD already confirmed PS5 chip is made on 7nm so no doubt its not 5nm. 7nm+ I guess it could happen, although it yields ~10% in performance, its no silver bullet.

7nm+ to me always looked like a "must have" for TDP reasons than your "10% headroom" for the speculated CUs/Clocks here.

 

[JoeJ]

Whatever PS5 that's delayed from 2019 to 2020 by a decision from 2017 will be identical at the silicon level.

Could it be, with recent consoles being more similar to each other and PC, they now have the option to wait on latest tech? (Which they did not had in the past because of more custom chips?)

Also, still wondering about 800 GB/s, what else (beside RT) could justify this?
Pushing AVX-256 to be the new SPUs?
SSD data transfer / decompression?

 

[DavidGraham]

AMD already confirmed PS5 chip is made on 7nm so no doubt its not 5nm. 7nm+ I guess it could happen, although it yields ~10% in performance, its no silver bullet.

The cost of new nodes is always cost prohibitive, even NVIDIA abstained from doing 7nm until it's cheaper and the yields are better. A console is much more susceptible to these factors.

everyone arriving later should be able to
a) see how it performs on production game code (and what developers actually do with it)
b) make modifications where necessary to improve performance (at least wrt) beating the performance of nvidia's first generation of RT acceleration.

Not for a design that was set in stones before RTX, the consoles had RT designs that predate RTX for sure, AMD was among the companies that developed DXR, so they designed their RT version around that time, or slightly after, but NVIDIA was just faster in getting their RT version to market.

 

[iroboto]

Not for a design that was set in stones before RTX, the consoles had RT designs that predate RTX for sure, AMD was among the companies that developed DXR, so they designed their RT version around that time, or slightly after, but NVIDIA was just faster in getting their RT version to market.

That's reasonable consideration. There is a time delay for things - a reasonable expectation is that everyone who was there at the discussion table was building their own RT solution. But that doesn't necessarily mean everyone had the same targets or goals of what they wanted to accomplish with their solution. Some may have been more ambitious, others more basic to get out the door faster.

 

[DavidGraham]

But that doesn't necessarily mean everyone had the same targets or goals of what they wanted to accomplish with their solution. Some may have been more ambitious, others more basic to get out the door faster.

Then we have to consider the possibility that AMD is dedicating more hardware to RT than NVIDIA, a notion that I find hard to believe.
A more convincing possibility would be Sony customizing the GPU to handle more RT.