General Next Generation Rumors and Discussions [Post GDC 2020]

[MrFox]

When they give a typical this is probably what they have seen in real game...

It's fine to give an idea. Then you look at a game with a lot of smoke animation and cloud data which compress like shit, and the other company uses a futuristic first person shooter with lots of metal patterns being more compressible data. The "typical" figure of B/W for one cannot be compared with the other if the didn't use the same compression ratio as a foundation.

 

[BRiT]

Will smoke animations and cloud data actually be static? And compressed on SSD? And read over and over again? Aren't those dynamic or procedural generations? So you wouldn't be reading them off storage over and over, but have it in ram and generate from there?

 

[chris1515]

Will smoke animations and cloud data actually be static? And compressed on SSD? And read over and over again? Aren't those dynamic or procedural generations? So you wouldn't be reading them off storage over and over, but have it in ram and generate from there?

Maybe better to baked the fluid simulation and render it with volumetric particle. But we had pretty rendered volumetric cloud and volumetric light this generation, probably better to continue like this.

 

[MrFox]

Will smoke animations and cloud data actually be static? And compressed on SSD? And read over and over again? Aren't those dynamic or procedural generations? So you wouldn't be reading them off storage over and over, but have it in ram and generate from there?

I guess we'll have more procedural. It was just a bad example, because in VFX anything with smoke and explosions fill up hundreds of terabytes overnight... I don't really know what would become more or less compressible, maybe we will see more focus from devs to exploit higher compression ratios, considering the limited storage and bandwidh advantage.

 

[Deleted member 7537]

The typical figure for lossless compression ratio have no place in hardware specs. It has nothing to do with the hardware capabilities. Basically, data entropy is unrelated to hardware. There is a direct link to the compression algorithm however but lossless codecs are not very far from each other.

Data types reaching 4:1 ratio will cap at 22GB/s on ps5 and 6GB/s on xbsx.

Data types reaching 2:1 ratio will cap at 11GB/s on ps5 and 4.8GB/s on xbsx.

Incompressible data will be "above 5GB" (cerny's precise wording, so. 5.5 have margins for OS, downloads, GC, etc) and "2.4GB guaranteed" on xbsx (unknown margins).

Everything will be somewhere between these extremes and will depend on game data. We should expect different types of data varying wildly. Average figures, from hypothetical game data compression ratio, are ridiculously misleading. They do this with LTO tapes with imaginary capacity.

Yeah, I guess some first party games may want to optimize to achieve as much compression as possible, but I guess that above certain point there aren't any significant gains and, as you say, not all data can be compressed.

 

[BRiT]

Yeah. Like in the Spider-Man breakdown, they had tons and tons of data they had to pair down to fit on the optical disc. Those limits would still be there, somewhat.... Somewhat relieved if they compress better. Maybe a larger relief would be using UHD disc for 100 GB physical instead of trying to fit on only a 50GB disc or making the hard call to use 2 discs for installation. But that's still just a sliver of the terabytes of data they'd love to be able to leverage.

Now that neither system has anemic CPUs like Jaguar and both have hefty GPUs, it will be interesting to see what devs come up with for their genuine next-gen engines.

 

[iroboto]

Yeah. Like in the Spider-Man breakdown, they had tons and tons of data they had to pair down to fit on the optical disc. Those limits would still be there, somewhat.... Somewhat relieved if they compress better. Maybe a larger relief would be using UHD disc for 100 GB physical instead of trying to fit on only a 50GB disc or making the hard call to use 2 discs for installation. But that's still just a sliver of the terabytes of data they'd love to be able to leverage.

Now that neither system has anemic CPUs like Jaguar and both have hefty GPUs, it will be interesting to see what devs come up with for their genuine next-gen engines.

sounds like the tools needed to make open world monsters.
1 world, multiple games... etc.
heh, sounds honestly very exhausting.

 

[AzBat]

Has MS stated how many hours users spend playing 360 or OG games vs XBO games?

On December 17, 2015 Major Nelson posted this...

Since launch on Nov. 12, fans have been enjoying Xbox 360 games on Xbox One for the first time ever, logging more than 9 million hours spent playing Xbox 360 games on Xbox One. And the most popular games fans from around the world have been spending those hours playing are Fallout 3, Gears of War 3, Just Cause 2, Assassin’s Creed II and DiRT 3.

https://majornelson.com/2015/12/17/...backward-compatibility-titles-playable-today/

I was thinking there was something more recent. If I find it I will post it.

Tommy McClain

 

[AzBat]

Here's another one from February 27, 2018...

We’re thrilled with the continued excitement for Xbox One Backward Compatibility—the community has played more than 840 million hours of Xbox 360 games on Xbox One. Compatibility is important to Xbox, to developers and their games, and our community. Preserving the art form of video games is part of our DNA, which is why Xbox One is the only console designed to play the best games of the past, present and future. We’re excited to continue to deliver on our vision of compatibility and grow our current Backward Compatible library, which includes more than 460 Xbox 360 games, 11 Xbox One X Enhanced Xbox 360 titles, and 13 Original Xbox games.

https://majornelson.com/2018/02/27/...ture-come-to-xbox-one-backward-compatibility/

EDIT:

Another data point...

https://twitter.com/x/status/872571508358496256

Mike Nichols is a former Marketing Chief of Xbox.

Phil Spencer chimed in too..

https://twitter.com/x/status/872584688040202240

GamesIndustry.biz did an article on June 8th 2017...

https://www.gamesindustry.biz/artic...f-xbox-one-owners-use-backwards-compatibility

Tommy McClain

 

[iroboto]

On December 17, 2015 Major Nelson posted this...



https://majornelson.com/2015/12/17/...backward-compatibility-titles-playable-today/

I was thinking there was something more recent. If I find it I will post it.

Tommy McClain

The hours for current gen titles being played on next gen devices will be massive.
You can't drop the major top 10 games off the planet. All of them are MP and have huge populations.

 

[Silent_Buddha]

The theoretical max value on compressed bandwidth on PS5 SSD is 22GB/s. 8/9 GB/s is just a realistic number that should be reachable in most conditions. But it can actually reach 20GB/s in ideal conditions :



https://www.neogaf.com/threads/next...leaks-thread.1480978/page-1508#post-257428542

Along the same lines, XBSX is 6+ GB/s when using the hardware decompression block. This was stated as in what a developer would typically see and not a theoretical maximum. I view it as the minimum attainable with hardware decompression.

Regards,
SB

 

[Silent_Buddha]

35=>8.5 seconds loading, not switching for XB1X to XBSX

that's what? 1/4th the time?

PS5 is doing like 1/10th.

While closer to an apples to apples comparison than switching apps, it's still not quite.

The State of Decay demonstration was the first initial loading of a stock XBO game in Backwards Compatibility mode from the game's menu. The PS3 demo was showing how quickly fast travel transitions are sped up.

In practical terms

• State of decay has to load everything from scratch.
  • This will include things that only happen on first initialization into a game from menu.
• Spiderman still has some assets resident that are reused in each scene.
  • Pedestrians and vehicles.
  • Universally used textures (glass and asphalt for example).

So basically, the takeaway is that both systems are roughly the same based on their respective SSD speeds. So in a similar fast travel like scenario the XBSX will still be slower at ~1.6-1.7 seconds.

However, this likely represents the worst case for both system. At least WRT XBSX, it's been stated that games have to be coded to take advantage of the numerous technologies they've implemented for their custom storage subsystem. While anything will see a speed up even without that happening the most dramatic increases will occur when it is specifically coded for.

I expect this to be the same for the PS5. After all a 1/10th speedup is nice and all, but we have a minimum 55x increase over the 2.5" HDD in the PS4. In the case of Spiderman it's potentially up to a 275x increase as they coded the game expecting a worst case 20 Mb/s scenario.

I expect both systems to improve upon this dramatically for games coded for their respective architectures.

Regards,
SB

 

[Strange]

While closer to an apples to apples comparison than switching apps, it's still not quite.

The State of Decay demonstration was the first initial loading of a stock XBO game in Backwards Compatibility mode from the game's menu. The PS3 demo was showing how quickly fast travel transitions are sped up.

In practical terms

• State of decay has to load everything from scratch.
  • This will include things that only happen on first initialization into a game from menu.
• Spiderman still has some assets resident that are reused in each scene.
  • Pedestrians and vehicles.
  • Universally used textures (glass and asphalt for example).

So basically, the takeaway is that both systems are roughly the same based on their respective SSD speeds. So in a similar fast travel like scenario the XBSX will still be slower at ~1.6-1.7 seconds.

However, this likely represents the worst case for both system. At least WRT XBSX, it's been stated that games have to be coded to take advantage of the numerous technologies they've implemented for their custom storage subsystem. While anything will see a speed up even without that happening the most dramatic increases will occur when it is specifically coded for.

I expect this to be the same for the PS5. After all a 1/10th speedup is nice and all, but we have a minimum 55x increase over the 2.5" HDD in the PS4. In the case of Spiderman it's potentially up to a 275x increase as they coded the game expecting a worst case 20 Mb/s scenario.

I expect both systems to improve upon this dramatically for games coded for their respective architectures.

Regards,
SB

Fast travel has a fixed amount of data that is required to be loaded, and it resulted in 8 seconds for original PS4 (I suppose SSD, because I played spiderman and if HDD could do it in 8 seconds it's really short then. I don't even recall it fast traveling in 8 seconds and I put a SSD in my PS4 pro from day 1)
that same amount of data gets loaded in 1/10 of the time for the demo.

Loading everything from scratch took 35 seconds on the XB1X (I also suppose SSD) and it was taken down to 8.5 seconds., which is ~1/4 of the time.

I'm comparing their relative loading speeds when compared to their last gen hardware w/out the tech, 1/10 to 1/4 instead of 0.8 to 8.5 seconds.

If Cerny has his way we'd expect the 35 seconds to be slashed to sub 1 second (as he advertised). That would probably require some more software changes though.

People may say that the spiderman demo was modded to provide the tech demo. Ya of course it was modded because we couldn't move through the map at the shown speed!
(remember sony said that spiderman's swinging speed (which constitutes the fastest way to move around at sustained speed) was capped in order to prevent loading as shown in the ps4 pro version demo)

 

[PSman1700]

It would appear that PS5's design was constrained by BC requirements after all, resulting in a forced solution as poopy as Nintendo's.

It would seem like that, yes. Development started in 2015, extreme high clocked GPUs didnt exist back then. They probably wanted '2x PS4 Pro and BC' and went with something close to 9TF. Stadia, SXS happened, and were seeing boosts. This is honestly the best explanation to the whole thing. Otherwise we would have seen something like 44CU at high(er) clocks.

*Lots of black plastic

Molten plastic j/k dont be mad

The hours for current gen titles being played on next gen devices will be massive.
You can't drop the major top 10 games off the planet. All of them are MP and have huge populations.

There's a reason BC was important for both sony and ms.

XBSX is 6+ GB/s when using the hardware decompression block

That's some beastly speeds, now i see why MS sells those SSD-memcards, most SSD drives for pc are way below that. Going to be intresting to see huge open world games what they can do with that. Back in 2001, i thought the first halo had impressive open areas (thx to hdd?).

 

[Deleted member 7537]

@Tkumpathenurpahl I am fine, thanks. Just very disappointed as you can see that they designed a flawed hardware because of BC requirements. That strategy was a failure on PS3, it was a failure on Wii U, I don't see why it should work on PS5. Just look at Switch and PS4 success, both recent designs without any kind of BC.

PS5 didn't need BC, and also, currently BC is broken on PS5 ! From their own statements, it's not working, that's ridiculous. All of that for 4% (or even less, like 2%) of games available.

And the yields are currently bad, frequency is too high:

https://www.chiphell.com/forum.php?mod=viewthread&tid=2201057&page=2&mobile=2#pid44551026

https://blog.us.playstation.com/202...ls-of-playstation-5-hardware-technical-specs/

A quick update on backward compatibility – With all of the amazing games in PS4’s catalog, we’ve devoted significant efforts to enable our fans to play their favorites on PS5. We believe that the overwhelming majority of the 4,000+ PS4 titles will be playable on PS5.

We’re expecting backward compatible titles will run at a boosted frequency on PS5 so that they can benefit from higher or more stable frame rates and potentially higher resolutions. We’re currently evaluating games on a title-by-title basis to spot any issues that need adjustment from the original software developers.

In his presentation, Mark Cerny provided a snapshot into the Top 100 most-played PS4 titles, demonstrating how well our backward compatibility efforts are going. We have already tested hundreds of titles and are preparing to test thousands more as we move toward launch. We will provide updates on backward compatibility, along with much more PS5 news, in the months ahead. Stay tuned!

It's quite incredible that Sony has been forced to come out and clarify this because some people just want to misrepresent what Cerny said. He said, btw, that most of the 100 most played games on PS4 work fine on PS5 and some needed rework, he did not say only 100 PS4 games work on PS5.

 

[Silent_Buddha]

That's some beastly speeds, now i see why MS sells those SSD-memcards, most SSD drives for pc are way below that. Going to be intresting to see huge open world games what they can do with that. Back in 2001, i thought the first halo had impressive open areas (thx to hdd?).

Just keep in mind when doing comparisons that even with that, the PS5's SSD is still faster and it also has hardware decompression capabilities to boost its effective throughput.

Regards,
SB

 

[Remij]

Just keep in mind when doing comparisons that even with that, the PS5's SSD is still faster and it also has hardware decompression capabilities to boost its effective throughput.

Regards,
SB

How much does sequential throughput really equate to effective performance though? Is there something I'm missing here, or could the 2 drives be closer in general performance than many people think? Maybe initial loads will be more instantaneous on PS5, but loading in assets during gameplay?

Basically why I'm asking is that on the PC side, random reads are far more indicative of real world performance in games, right? Is there some reason why this will no longer be true on consoles this gen?

 

[Deleted member 7537]

How much does sequential throughput really equate to effective performance though? Is there something I'm missing here, or could the 2 drives be closer in general performance than many people think? Maybe initial loads will be more instantaneous on PS5, but loading in assets during gameplay?

Basically why I'm asking is that on the PC side, random reads are far more indicative of real world performance in games, right? Is there some reason why this will no longer be true on consoles this gen?

In games like Star Citizen actually affects performance.

 

[dobwal]

I don't think it is a matter of "possible", Cerny flat stated it, but he also stated that was best case and would not be the norm. (Did he actually say "Maybe 10% of the time" or is my mind sticking that in there?)


Unrelated thought:
Going from memory. If i get a figures wrong, apologies in advance. They should be ballpark correct.

Roughly speaking, the Sony SSD solution is around 2x as fast. I think Shifty posted some napkin math based on an older post that went, roughly, 62mb per 16 ms frame to change everything. MS is at 66mb per 16ms frame and Sony is around 100MB per 16 ms frame. As cool as the Sony solution is, and I will leave it to others to debate the real world implications thereof, I do have to wonder. MS specifically mentioned that most of the swapping and memory in use was about textures. Which often went unused or were only partially used. There is supposed to be something about only loading (SSD to RAM) the part of a texture that is actually going to be used, rather than the whole thing. Wouldn't that, assuming it works and MS is not misrepresenting the situation, have a huge effect on your needed bandwidth between the SSD and the RAM? As well as how much RAM was in actual use?

Cerny mentioned the same thing during during his PS5 talk. If I’m not mistaken Cerny said you need about 30 seconds worth of data.

I imagine alot of swapping is texture based because it would chew up a lot of data just to store the highest quality version of each texture in a scene in VRAM.

So while the game is storing a bunch of data for many frames downstream, its also streaming a bunch of higher resolution texture data as late as possible to accommodate IQ.

 

[manux]

Cerny mentioned the same thing during during his PS5 talk. If I’m not mistaken Cerny said you need about 30 seconds worth of data.

I imagine alot of swapping is texture based because it would chew up a lot of data just to store the highest quality version of each texture in a scene in VRAM.

So while the game is storing a bunch of data for many frames downstream, its also streaming a bunch of higher resolution texture data as late as possible to accommodate IQ.

Just to add the details in in case someone missed them. Cerny said previous gen required roughly 30s worth of data in ram due to random access to disk being slow. For ps5 he claimed data for next 1s is needed. This means the 16GB vs. 8GB comparison is not valid as ps5 uses the available memory much more efficiently and there is a lot less unneeded data in ram.