Edit: whoops, use d2.28 instead of 2.23.. Yea would be slightly under 10
9.97TF is it?
Close enough, as Cerny said I suppose.
Last edited:
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
PlayStation 5 System Architecture Deep Dive |OT| Secret Agent Cerny
- Thread starter vestan
- Start date
- Status
9.97TF is it?
Close enough, as Cerny said I suppose.
Wonder if first party studios can make game mechanics where the player can be teleported into an entire different worlds instantaneously, similar to the portal scene in Doctor Strange. 🤔
I find it fascinating how little importance non-dev folks ascribe to this stuff when it's become clear over the past few years that storage and memory are critical bottlenecks, if not the most critical bottlenecks, in graphical power.
Damnit Transistor ! Trying to parse this thread for knowledgeable members and I can't recognize anyone anymore! Who came up with that idea?!
I'm getting it for exclusive games, so the power was never going to make or break my decision, but this thing better be cheaper or I expect MS to make up significant ground in the US with a stronger box and attractive Gamepass offerings.
Chances are it is bigger, they need a strong cooling solution for these clock speeds. This could be generating more heat than Xbox, as you go up in freq, efficiency drops.
Instantly yes... plus like the 3-second teleportation animation.
I think he did a good job. This is one of the only conferences I've seen were I felt like I actually learned something about how hardware and game design works. I do wish they would have shown some tech demos demonstrating some of it in action.
Yes, exactly as I said to you. That drop would put PS5 at 9.97TF of compute, which you misleadingly refer to as the "9 TF area". Rounding is typically done upward once more than halfway to the next integer. For example, if XSX were 11.97TF, would you say it was "in the 11TF area"? Further, you classify this truncated lower number as the "real performance" of the PS5, when in fact the manufacturer says it's the probable minimum, rarely dipped to.
You were clearly trying to maximize the perceived difference. It would be both more accurate and more beneficial to discussion to say something like "It looks like PS5 may sometimes perform 2.2TF below the sustained 12+TF performance of Xbox Series X, rather than 1.9TF below. There's always gonna be a gap."
Note that improved pixel fillrate isn't a magic wand. The PS4 Pro had better pixel fillrate than Xbox One X (by about the same proportion), yet the usual rendering gap between them was still about as big as raw TF implied. Though maybe the pixel fill stopped it from being even worse, I'm not educated enough to know.
More NPCs, better AI, better physics and more physics objects, larger worlds and better and more world simulation.
That sounds awesome, maybe Kratos can have some of those powers.
This is not accurate and assumes that XSX is not using hardware solutions to remove i/o bottlenecks and simply using an off the selves ssd
Well, I assume they are going to be bandwidth limited well before being ROP bound so I doubt it will be much of a factor.
Exactly, picture an epic battle between Kratos and Thor that starts in the portal room and each phase knocks em to a different planet within the nine realms ending the fight in Asgard.
I'd be shocked if Barlog wasn't thinking of this when the SSD was pitched to 1st parties.
Lol now anything less than what you described will be a disappointment. 😋😛
It honestly kind of bugs me that so many people are insisting that Sony charge no more than $400, otherwise it's a rip off or something like that. I just spent nearly $600 to upgrade my PC last last year. The PS5 blows it out of the park easily. If they charge $500 for it, you're getting great value for that price. The specs for this thing is far superior to the PS4 when you compare both to PC hardware in each consoles respective year of release. No one thought the PS4 was special, where as this year it's going to take a beastly PC to beat the PS5 when it comes out.
Oh yeah and the Xbox is even better. If they want to charge $600 for that, it would still be an amazing deal. If it's $500 then shit, hats off to MS for providing such great value for the consumers. I'm glad neither company held back their systems trying to fit as much as possible into a small box that could only be sold for $400(although that could still happen, but I definitely don't think either will hit that price).
EDIT: When I think more about it. I guess most of these people live only in the console bubble where they feel consoles should always cost around $300-400. Anybody who plays on PC and upgrades it every 3-4 at least will see how much better value these new consoles provides compared to what you can get on the current PC market. Even when next gen parts comes out the story isn't going to change much if at all.
Oh yeah and the Xbox is even better. If they want to charge $600 for that, it would still be an amazing deal. If it's $500 then shit, hats off to MS for providing such great value for the consumers. I'm glad neither company held back their systems trying to fit as much as possible into a small box that could only be sold for $400(although that could still happen, but I definitely don't think either will hit that price).
EDIT: When I think more about it. I guess most of these people live only in the console bubble where they feel consoles should always cost around $300-400. Anybody who plays on PC and upgrades it every 3-4 at least will see how much better value these new consoles provides compared to what you can get on the current PC market. Even when next gen parts comes out the story isn't going to change much if at all.
Exactly what i was picturing when i was thinking about that battle lol.
Would the Pro's higher pixel fillrate explain why many Pro enhanced games had better framerates than their X enhanced versions?
They talked about how they have some stuff to help with this.
but is it? 560 vs 448 is 1.25x so roughly the same as gpu perf difference, can't see much bandwidth/tflop advantagege here for xsx + ps5 has all 16 with full speed and xbox 10 and 6 with half
The console version of witcher 3has way way less NPCs in cities and around compared with PC. It is a huge difference, it runs at the lowest setting compared with PC. Disparities like that should not happen anymore and a game like Cyberpunk is really made for that. Similarly distance lod in some games come with a big CPU hit. That's something that becomes more noticeable at 4K as well and right now consoles usually run at the lowest setting compared with PC.
Frame rate, 60fps will be much more common as most games probably won't tax these GPUs anytime soon specially in the earlier years. You might even see 120fps performance modes in call of duty or similar this year. A game like modern warfare easily runs at 120fps on these consoles at ~1440p with some tweaks. Along with it you will see much better performance consistency.
Hardware solutions like? And if they are why didn't they point it out?
Sony literally is putting stuff in their APU (including SRAM) using up valuable die space specifically to tackle all the received bottlenecks that could impede their targetted SSD performance. And spent a great deal of time talking about it. Sony is targeting 1 second load times vs the shown 8ish second load times of the XSX.
And to be specific, MS also has some form of hardware solution to assist with their SSD performance, but its nowhere near as intricate as sony's implementation.
On is 2.4GB/s... the other is 5.5GB/s. One showed actual customizations to their APU to facilitate their advertised speeds the other didn't... what am I missing here?
I am honestly finding it amusing and baffling that people are touting the PS5 SSD like it is some revolutionary feature that is going to add more polygons and raytracing. It's like Sony fans are grasping for something to latch onto. Sure, the PS5 SSD has incredible read speeds. Sure, it is a good feature. But it wont improve graphics. The GPU, CPU, and TFLOPS will impact graphics WAY MORE than a SSD. The PS5 SSD will shave seconds off loading times over XSX, yes. XSX has suspend and resume and its own fast SSD architecture though that that the PS5 SSD will be a meager benefit and likely used more by first party developers. Comparing loading times between the PS5 and XSX literally no one is going to care about.
Why fast storage changes everything
The specs on this page represent only the tiniest fraction of the potential of the storage solution Microsoft has engineered for the next generation. In last year's Project Scarlett E3 teaser, Jason Ronald - partner director of project management at Xbox - described how the SSD could be used as 'virtual memory', a teaser of sorts that only begins to hint at the functionality Microsoft has built into its system.
On the hardware level, the custom NVMe drive is very, very different to any other kind of SSD you've seen before. It's shorter, for starters, presenting more like a memory card of old. It's also rather heavy, likely down to the solid metal construction that acts as a heat sink that was to handle silicon that consumes 3.8 watts of power. Many PC SSDs 'fade' in performance terms as they heat up - and similar to the CPU and GPU clocks, this simply wasn't acceptable to Microsoft, who believe that consistent performance across the board is a must for the design of their consoles.
The form factor is cute, the 2.4GB/s of guaranteed throughput is impressive, but it's the software APIs and custom hardware built into the SoC that deliver what Microsoft believes to be a revolution - a new way of using storage to augment memory (an area where no platform holder will be able to deliver a more traditional generational leap). The idea, in basic terms at least, is pretty straightforward - the game package that sits on storage essentially becomes extended memory, allowing 100GB of game assets stored on the SSD to be instantly accessible by the developer. It's a system that Microsoft calls the Velocity Architecture and the SSD itself is just one part of the system.
"Our second component is a high-speed hardware decompression block that can deliver over 6GB/s," reveals Andrew Goossen. "This is a dedicated silicon block that offloads decompression work from the CPU and is matched to the SSD so that decompression is never a bottleneck. The decompression hardware supports Zlib for general data and a new compression [system] called BCPack that is tailored to the GPU textures that typically comprise the vast majority of a game's package size."
PCI Express 4.0 connections hook up both internal and optional external SSDs directly to the processor.
The final component in the triumvirate is an extension to DirectX - DirectStorage - a necessary upgrade bearing in mind that existing file I/O protocols are knocking on for 30 years old, and in their current form would require two Zen CPU cores simply to cover the overhead, which DirectStorage reduces to just one tenth of single core.
"Plus it has other benefits," enthuses Andrew Goossen. "It's less latent and it saves a ton of CPU. With the best competitive solution, we found doing decompression software to match the SSD rate would have consumed three Zen 2 CPU cores. When you add in the IO CPU overhead, that's another two cores. So the resulting workload would have completely consumed five Zen 2 CPU cores when now it only takes a tenth of a CPU core. So in other words, to equal the performance of a Series X at its full IO rate, you would need to build a PC with 13 Zen 2 cores. That's seven cores dedicated for the game: one for Windows and shell and five for the IO and decompression overhead."
Asset streaming is taken to the next level, but Microsoft wasn't finished there. Last-gen, we enjoyed a 16x increase in system memory, but this time it's a mere 2x - or just 50 per cent extra if we consider Xbox One X as the baseline. In addition to drawing more heavily upon storage to make up the shortfall, Microsoft began a process of optimising how memory is actually used, with some startling improvements.
"We observed that typically, only a small percentage of memory loaded by games was ever accessed," reveals Goossen. "This wastage comes principally from the textures. Textures are universally the biggest consumers of memory for games. However, only a fraction of the memory for each texture is typically accessed by the GPU during the scene. For example, the largest mip of a 4K texture is eight megabytes and often more, but typically only a small portion of that mip is visible in the scene and so only that small portion really needs to be read by the GPU."
Microsoft has partnered with Seagate for its proprietary external 1TB SSD expansion. It's very short, quite weighty for its dimensions and actually presents rather like a memory card.
As textures have ballooned in size to match 4K displays, efficiency in memory utilisation has got progressively worse - something Microsoft was able to confirm by building in special monitoring hardware into Xbox One X's Scorpio Engine SoC. "From this, we found a game typically accessed at best only one-half to one-third of their allocated pages over long windows of time," says Goossen. "So if a game never had to load pages that are ultimately never actually used, that means a 2-3x multiplier on the effective amount of physical memory, and a 2-3x multiplier on our effective IO performance."
A technique called Sampler Feedback Streaming - SFS - was built to more closely marry the memory demands of the GPU, intelligently loading in the texture mip data that's actually required with the guarantee of a lower quality mip available if the higher quality version isn't readily available, stopping GPU stalls and frame-time spikes. Bespoke hardware within the GPU is available to smooth the transition between mips, on the off-chance that the higher quality texture arrives a frame or two later. Microsoft considers these aspects of the Velocity Architecture to be a genuine game-changer, adding a multiplier to how physical memory is utilised.
The Velocity Architecture also facilitates another feature that sounds impressive on paper but is even more remarkable when you actually see it play out on the actual console. Quick Resume effectively allows users to cycle between saved game states, with just a few seconds' loading - you can see it in action in the video above. When you leave a game, system RAM is cached off to SSD and when you access another title, its cache is then restored. From the perspective of the game itself, it has no real idea what is happening in the background - it simply thinks that the user has pressed the guide button and the game can resume as per normal.
We saw Xbox Series X hardware cycling between Forza Motorsport 7 running in 4K60 Xbox One X mode, State of Decay 2, Hellblade and The Cave (an Xbox 360 title). Switching between Xbox One X games running on Series X, there was around 6.5 seconds delay switching from game to game - which is pretty impressive. Microsoft wasn't sharing the actual size of the SSD cache used for Quick Resume, but saying that the feature supports a minimum of three Series X games. Bearing in mind the 13.5GB available to titles, that's a notional maximum of around 40GB of SSD space, but assuming that the Velocity Architecture has hardware compression features as well as decompression, the actual footprint may be smaller. Regardless, titles that use less memory - like the games we saw demonstrated - should have a lower footprint, allowing more to be cached.
Why fast storage changes everything
The specs on this page represent only the tiniest fraction of the potential of the storage solution Microsoft has engineered for the next generation. In last year's Project Scarlett E3 teaser, Jason Ronald - partner director of project management at Xbox - described how the SSD could be used as 'virtual memory', a teaser of sorts that only begins to hint at the functionality Microsoft has built into its system.
On the hardware level, the custom NVMe drive is very, very different to any other kind of SSD you've seen before. It's shorter, for starters, presenting more like a memory card of old. It's also rather heavy, likely down to the solid metal construction that acts as a heat sink that was to handle silicon that consumes 3.8 watts of power. Many PC SSDs 'fade' in performance terms as they heat up - and similar to the CPU and GPU clocks, this simply wasn't acceptable to Microsoft, who believe that consistent performance across the board is a must for the design of their consoles.
The form factor is cute, the 2.4GB/s of guaranteed throughput is impressive, but it's the software APIs and custom hardware built into the SoC that deliver what Microsoft believes to be a revolution - a new way of using storage to augment memory (an area where no platform holder will be able to deliver a more traditional generational leap). The idea, in basic terms at least, is pretty straightforward - the game package that sits on storage essentially becomes extended memory, allowing 100GB of game assets stored on the SSD to be instantly accessible by the developer. It's a system that Microsoft calls the Velocity Architecture and the SSD itself is just one part of the system.
"Our second component is a high-speed hardware decompression block that can deliver over 6GB/s," reveals Andrew Goossen. "This is a dedicated silicon block that offloads decompression work from the CPU and is matched to the SSD so that decompression is never a bottleneck. The decompression hardware supports Zlib for general data and a new compression [system] called BCPack that is tailored to the GPU textures that typically comprise the vast majority of a game's package size."
PCI Express 4.0 connections hook up both internal and optional external SSDs directly to the processor.
The final component in the triumvirate is an extension to DirectX - DirectStorage - a necessary upgrade bearing in mind that existing file I/O protocols are knocking on for 30 years old, and in their current form would require two Zen CPU cores simply to cover the overhead, which DirectStorage reduces to just one tenth of single core.
"Plus it has other benefits," enthuses Andrew Goossen. "It's less latent and it saves a ton of CPU. With the best competitive solution, we found doing decompression software to match the SSD rate would have consumed three Zen 2 CPU cores. When you add in the IO CPU overhead, that's another two cores. So the resulting workload would have completely consumed five Zen 2 CPU cores when now it only takes a tenth of a CPU core. So in other words, to equal the performance of a Series X at its full IO rate, you would need to build a PC with 13 Zen 2 cores. That's seven cores dedicated for the game: one for Windows and shell and five for the IO and decompression overhead."
Asset streaming is taken to the next level, but Microsoft wasn't finished there. Last-gen, we enjoyed a 16x increase in system memory, but this time it's a mere 2x - or just 50 per cent extra if we consider Xbox One X as the baseline. In addition to drawing more heavily upon storage to make up the shortfall, Microsoft began a process of optimising how memory is actually used, with some startling improvements.
"We observed that typically, only a small percentage of memory loaded by games was ever accessed," reveals Goossen. "This wastage comes principally from the textures. Textures are universally the biggest consumers of memory for games. However, only a fraction of the memory for each texture is typically accessed by the GPU during the scene. For example, the largest mip of a 4K texture is eight megabytes and often more, but typically only a small portion of that mip is visible in the scene and so only that small portion really needs to be read by the GPU."
Microsoft has partnered with Seagate for its proprietary external 1TB SSD expansion. It's very short, quite weighty for its dimensions and actually presents rather like a memory card.
As textures have ballooned in size to match 4K displays, efficiency in memory utilisation has got progressively worse - something Microsoft was able to confirm by building in special monitoring hardware into Xbox One X's Scorpio Engine SoC. "From this, we found a game typically accessed at best only one-half to one-third of their allocated pages over long windows of time," says Goossen. "So if a game never had to load pages that are ultimately never actually used, that means a 2-3x multiplier on the effective amount of physical memory, and a 2-3x multiplier on our effective IO performance."
A technique called Sampler Feedback Streaming - SFS - was built to more closely marry the memory demands of the GPU, intelligently loading in the texture mip data that's actually required with the guarantee of a lower quality mip available if the higher quality version isn't readily available, stopping GPU stalls and frame-time spikes. Bespoke hardware within the GPU is available to smooth the transition between mips, on the off-chance that the higher quality texture arrives a frame or two later. Microsoft considers these aspects of the Velocity Architecture to be a genuine game-changer, adding a multiplier to how physical memory is utilised.
The Velocity Architecture also facilitates another feature that sounds impressive on paper but is even more remarkable when you actually see it play out on the actual console. Quick Resume effectively allows users to cycle between saved game states, with just a few seconds' loading - you can see it in action in the video above. When you leave a game, system RAM is cached off to SSD and when you access another title, its cache is then restored. From the perspective of the game itself, it has no real idea what is happening in the background - it simply thinks that the user has pressed the guide button and the game can resume as per normal.
We saw Xbox Series X hardware cycling between Forza Motorsport 7 running in 4K60 Xbox One X mode, State of Decay 2, Hellblade and The Cave (an Xbox 360 title). Switching between Xbox One X games running on Series X, there was around 6.5 seconds delay switching from game to game - which is pretty impressive. Microsoft wasn't sharing the actual size of the SSD cache used for Quick Resume, but saying that the feature supports a minimum of three Series X games. Bearing in mind the 13.5GB available to titles, that's a notional maximum of around 40GB of SSD space, but assuming that the Velocity Architecture has hardware compression features as well as decompression, the actual footprint may be smaller. Regardless, titles that use less memory - like the games we saw demonstrated - should have a lower footprint, allowing more to be cached.
I think Sony should continue using checkerboard rendering honestly. Every checkerboard rendered game I've played looks amazing on my TCL R625 and that would certainly help alleviate any perceived power gap
Again, it's not just about loading. The SSD will benefit both through game design and streaming.
The next CoD will be a great showcase of the differences between the two consoles so we can all move on with our lives arguing about specs when all the matters will be the games. Both will do well.
It's not just load times, it's asset streaming within the game.
Yes, it would be good to hear directly that VRS is present in PS5. As for DLSS, neither PS5 nor XSX will have that. Nvidia's solution requires tensor cores which AMD doesn't provide on their GPUs. Don't worry too much about it, though. The effectiveness of DLSS is very much oversold, in my opinion, and other reconstruction techniques can be just as good (or better, depending on how you define "better").
Right now people see numbers and a demo from MS. We don't have anything more than some numbers from Sony but no demos. People are judging without being able to see the full picture.
We had a video of Spiderman loading in less than a second a year ago.
Thats much faster than what Microsoft showed off this week.
No multiplatform game is going to be designed with the PS5's SSD in mind, so to see the real difference the better SSD makes will be comparing first party titles, which is always difficult. Hopefully we get some first party games that make the cost of the advanced SSD worth it.
They don't have dedicated hardware for it, but MS does have an equivalent to DLSS in the Series X.
Holy crap, despite being a super technical GDC talk the dang video still has almost 6 million views right now trending #1 on YouTube. People are so thirsty for some PS5 news
That's also a demo that was optimized for the PS5, the Xbox demos were just back compat games.
Ok. I did say I am aware MS also didsome hardware stuff too? But lets look at what they actually did.
They have dedicated silicon specifically for decompression (this is why MS compression is better than that of the PS5) and that's why they can get their throughput up from 2.4GB/s to 4.8GB/s. Thats also the hardware stuff I was referring to. And then they have an API stack built to talk with this SSD (which should go without saying)...
and that's it.
Everything else about the form factor is just talk. Its a 2330 NVMe SSD. And probably using system RAMfor its SSD cache (similar to what they do with their surface laptops.
Sony's SSD solution is a lot better than this. That's literally why it's at 5.5GBand at 8GB equivalent when dealing with compressed data. I still don't see what the discussion here is.
Why is its hard to accept that sony invested more in their SSD than they did in their GPU? Like sony literally used up space in there APU for SSD stuff that could have gone to throw in a few more CUs for the GPU. Isn't that obvious that they felt having a faster SSD was more important than having a few more TFs?
I think what a lot of people are concerned about is that the SSD won't be able to be fully utilised in that way outside of first-party games. Though I assume the faster speeds should allow the PS5 to have more "effective RAM" due to the size of the buffer being smaller?
I've seen a few devs--including 1 or 2 in these threads here--talk about this. I know we rarely give experts the benefit of the doubt, but sometimes we should--for our own sanity's sake lol.
Every day I wake up I say a small prayer that they do.
Xbox Series X: A Closer Look at the Technology Powering the Next Generation - Xbox Wire
A few months ago, we revealed Xbox Series X, our fastest, most powerful console ever, designed for a console generation that has you, the player, at its center. When it is released this holiday season, Xbox Series X will set a new bar for performance, speed and compatibility, all while allowing...
news.xbox.com
Then why does even their official website peg compressed data throughput at 4.8GB/s
This. With the raw graphical power likely being only maybe 3-3.5x the Pro's, I think reconstruction will be really important for making some crazy impressive games. I want every game to go the Spider-Man route but at 40-60fps if possible.
Xbox Series X: A Closer Look at the Technology Powering the Next Generation - Xbox Wire
A few months ago, we revealed Xbox Series X, our fastest, most powerful console ever, designed for a console generation that has you, the player, at its center. When it is released this holiday season, Xbox Series X will set a new bar for performance, speed and compatibility, all while allowing...
Then why does even their official website peg compressed data throughput at 4.8GB/s
I might be reading this wrong, but isn't the 6GB/s from MS kind of like the 22GB/s number Cerny mentioned? The way I'm interpreting it is that there are 3 relevant numbers:
Raw Read Speed (5.5GB/s on PS5, 2.4GB/s on XSX)
Compressed Read Speed (8-9GB/s on PS5, 4.8GB/s on XSX)
Maximum Potential Throughput (22GB/s on PS5, 6GB/s on XSX)
Or am I fucking this up?
Give it 5 years from now and i wouldn't be surprised if we start seeing 3rd party titles launched only on the PS5 and PC just because of the SSD, it'll be hard for devs to not go with their wildest imaginations with the PS5 SSD because of the XSX SSD.
If that 5GB/s SSD really is a game changer as Sony claims then i wouldn't be surprised if more 3rd party developers opt for PS5 exclusives just to make some crazy game ideas that only a PS5 canl do.
The PS5 was designed with developers in mind the most afterall.
The PS5 was designed with developers in mind the most afterall.
can't say for games as i'm not a game dev but as an enterprise architect for my corp (i design and ran the VM infrastructure and also setup the SAN and brocade switches), storage does indeed seems to be a bottleneck for many modern application like SQL and Mail database and file access system. No joke, you can add the fastest RAM and CPU but at some point, the bottleneck i've encountered always comes down to storage IOPS. If only we can get this type of nvme ssd at high enough capacity and density for enterprise.
example, i am doing some migration now and the migration was slowed down due to disk iops, despite the server not really and full cpu and ram usage.
Last edited:
Speaking of which are devs here allowed to freely talk about what they love/hate about the next gen consoles on ERA since everything's out in the open? 🤔
I doubt that will ever happen. Somethings will just be better.
- Status