Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.
MSFT now is not what it used to be.
And the best devs there are in Azure right now. While the best Sony's devs are in the ICE which essentially produces all these tools.
This is simply not true, vs is still the benchmark in ide... Ste you even a programmer? Have you ever programmed something? Vcr doesn't count.... Geez guys at least something stick to reality
 
I don't know what the situation is with VS and doubt Microsoft are at any disadvantage with their own software but do remember Richard and John briefly comment in one of the DF videos about how Sony's dev environment/tools are quite a bit better than Microsoft's. That surprised me.

Can't remember which video so can't link right now.
 
I don't know what the situation is with VS and doubt Microsoft are at any disadvantage with their own software but do remember Richard and John briefly comment in one of the DF videos about how Sony's dev environment/tools are quite a bit better than Microsoft's. That surprised me.

Can't remember which video so can't link right now.
That's talking about api support and tools, not compilers and ide, tools get rewritten to support new hardware and if Sony finalized design before ms they had more time to work on sw.
 
That's talking about api support and tools, not compilers and ide, tools get rewritten to support new hardware and if Sony finalized design before ms they had more time to work on sw.

Sure, but Richard and John were talking about Xbox One and PS4. How it will go with PS5 and XSX we wait and see.
 
So six pieces of hardware (SSD, Kraken block, DMA controller, coherency engines, 2x co-processors) and one piece of software: a new ID-based API.

In the end this gives 5.5GB/s raw, 8-9GB/s compressed (block theoretically capable of 22GB/s) giving a theoretical 100x I/O multiplier.

That's how I currently understand things.
I think you are spot on. However there is still one thing that has been mostly absent from the SSD discussions: random IOPS.

The sequential speeds are clear but I feel the PS5 might have an even larger advantage than the sequential difference makes us believe.

One big difference is that Sony uses 12 channels for 12 chips while Msft uses 16 chips on 4 channels. That means - with my non expert understanding - that there might be a factor three or four advantage in random IOPS for the PS5.

That combined with all of the co-processors to reduce any latency and inefficiency, I think that will be where the real profit is at. Cerny focused on the point that they want the SSD speed to translate to actual performance 1 to 1.
 
How would you feel if it was at 9:20pm? 😅
Only dealer ,windows central and few hardcore fans would watch that tbh 🤔😂(since those are the ones pushing this fake info)

Ps5 clock was 2ghz at one point in time when xsx clcok was 1.6ghz at that time in github. Why is ms only allowed to increase to 1.85 in retail and not sony to 2.25?🤔

Both increased 0.25 ghz due to rdna 2 being able to do it and still fall into their initial TDP profile design.
 
Last edited:
Only dealer ,windows central and few hardcore fans would watch that tbh 🤔😂(since those are the ones pushing this fake info)

Ps5 clock was 2ghz at one point in time when xsx clcok was 1.6ghz at that time in github. Why is ms only allowed to increase to 1.85 in retail and not sony to 2.25?🤔

Both increased 0.25 ghz due to rdna 2 being able to do it and still fall into their initial TDP profile design.
No that was the ram frequency. From the start MS targeted 12tf (at least since january 2019). They had to with XBX's 6tf.
 
Last edited:
In all seriousness, is Cerny throwing some Shade when this slide comes on?
8AYatUE.jpg
 
I think you are spot on. However there is still one thing that has been mostly absent from the SSD discussions: random IOPS.

The sequential speeds are clear but I feel the PS5 might have an even larger advantage than the sequential difference makes us believe.

One big difference is that Sony uses 12 channels for 12 chips while Msft uses 16 chips on 4 channels. That means - with my non expert understanding - that there might be a factor three or four advantage in random IOPS for the PS5.

That combined with all of the co-processors to reduce any latency and inefficiency, I think that will be where the real profit is at. Cerny focused on the point that they want the SSD speed to translate to actual performance 1 to 1.

Also don't forget priorities for access to ssd. We know Sony has 6 priority levels. Not how many MS has (standard is 2).
 
That is not a small claim. I'm gonna need to see the receipts for this one.

It's under NDA. But you can see how good it is by some posts of confused developers that think that Sony tools use DX/Vulkan and then somehow port their game to PS4/5. Or that Sony internal studios "develop for PCs".
 
No that was the ram frequency. From the start MS targeted 12tf (at least since january 2019). They had to with XBX's 6tf.
I understand that . I m talking about clock change. 9.2TF refers to github last year when ps5 had 2.0ghz clock. Dealer and windeows central hang on to that and say ps5 is 9m2 tf and sony last minute increased clock to 2.25 ghz. I m pointing out xsx in 2019 github has clock of 1.6 ghz with 56 cu to compensate. Once they got rdna2 final silicon they increased the clock to 1.85 with 52 cu. Meaning final rdna2 allowed them to go 0.25 ghz higher clock while still be in the same TDP design threshold. Same for sony. Sony also increased clock 0.25 ghz to 2.25ghz once they got the final rdna2 chip.

Both increased the gpu clock 0.25 ghz because rdna2 allowed it. Not because they both had no choice and wanted to increase clock. Console design doesn't work like that.
All the TDP is set way in advance . If they suddenly increase the TDP of gpu something else has to take a hit which would create design implications .

However in this case rdna 2 allowed both to increase clock while staying in the same TDP design criteria . So 9.2 tf ps5 doesn't exist just like 10.4 tf xsx doesn't exist (1.6ghz clock of last year github for xsx)
 
I think you are spot on. However there is still one thing that has been mostly absent from the SSD discussions: random IOPS.

The sequential speeds are clear but I feel the PS5 might have an even larger advantage than the sequential difference makes us believe.

One big difference is that Sony uses 12 channels for 12 chips while Msft uses 16 chips on 4 channels. That means - with my non expert understanding - that there might be a factor three or four advantage in random IOPS for the PS5.

That combined with all of the co-processors to reduce any latency and inefficiency, I think that will be where the real profit is at. Cerny focused on the point that they want the SSD speed to translate to actual performance 1 to 1.

Yeah I have no idea about IOPS or the other tech stuff around the SSDs. I do wonder if the main reason both have expandable storage is down to the fact of the chips being soldered to the PCB and faulty/worn out chips would render a console dead without a way of replacing them.

I would've liked to see the SSD chips on the XSX PCB but frustratingly in both DF's and Austin's videos constructing a XSX they didn't show it. Not sure if that was deliberate on Microsoft's part?
 
I understand that . I m talking about clock change. 9.2TF refers to github last year when ps5 had 2.0ghz clock. Dealer and windeows central hang on to that and say ps5 is 9m2 tf and sony last minute increased clock to 2.25 ghz. I m pointing out xsx in 2019 github has clock of 1.6 ghz with 56 cu to compensate. Once they got rdna2 final silicon they increased the clock to 1.85 with 52 cu. Meaning final rdna2 allowed them to go 0.25 ghz higher clock while still be in the same TDP design threshold. Same for sony. Sony also increased clock 0.25 ghz to 2.25ghz once they got the final rdna2 chip.

Both increased the gpu clock 0.25 ghz because rdna2 allowed it. Not because they both had no choice and wanted to increase clock. Console design doesn't work like that.
All the TDP is set way in advance . If they suddenly increase the TDP of gpu something else has to take a hit which would create design implications .

However in this case rdna 2 allowed both to increase clock while staying in the same TDP design criteria . So 9.2 tf ps5 doesn't exist just like 10.4 tf xsx doesn't exist (1.6ghz clock of last year github for xsx)
I understand, I am telling you that those 1.6ghz-ish clocks (of Arden) were always about the ram frequency, not the GPU.
 
It's under NDA. But you can see how good it is by some posts of confused developers that think that Sony tools use DX/Vulkan and then somehow port their game to PS4/5. Or that Sony internal studios "develop for PCs".
That just says to me that Sony's integration is good - which you'd expect from a platform holder of a video game console. VS is the industry standard, and for good reason. Failing to have proper integration, especially for their Multiplatform third party partners, would be a massive blemish.
However, this doesn't translate to it being "better than Microsoft's". Speaking from my somewhat limited experience, MS is still the best in this particular business. Sorry, no offence, but I'm not buying it without some serious backing.

I started to pay attention to some of that stuff years back, around when I recall John Carmack talking about MS integration and how it shaped some coding approaches during RAGE's development. I believe he was talking during a QuakeCon key note, so I can probably dig it up if you'd like? Anyway, he talked about Microsoft's top tier integrated tools - the crazy expensive ones that about six people on the planet actually understand - and how he finally started listening to their suggestions, and how Carmack then used that to improve their entire code base, ready for their first big push into console focused tech, and the difference it made. That was back in the 360 era - and Microsoft haven't been resting on their laurels.
 
This I find very difficult to believe.
It's not unheard of. For instance, the most popular application platform in the world, Chromium (used by Vivaldi, Chrome, Opera, Brave etc), are using VS as just a frontend and debugging tool, the compiler from MS is not used at all. clang is used on all platforms, same as Sony is using.
 
Last edited:
That just says to me that Sony's integration is good - which you'd expect from a platform holder of a video game console. VS is the industry standard, and for good reason. Failing to have proper integration, especially for their Multiplatform third party partners, would be a massive blemish.
However, this doesn't translate to it being "better than Microsoft's". Speaking from my somewhat limited experience, MS is still the best in this particular business. Sorry, no offence, but I'm not buying it without some serious backing.

I started to pay attention to some of that stuff years back, around when I recall John Carmack talking about MS integration and how it shaped some coding approaches during RAGE's development. I believe he was talking during a QuakeCon key note, so I can probably dig it up if you'd like? Anyway, he talked about Microsoft's top tier integrated tools - the crazy expensive ones that about six people on the planet actually understand - and how he finally started listening to their suggestions, and how Carmack then used that to improve their entire code base, ready for their first big push into console focused tech, and the difference it made. That was back in the 360 era - and Microsoft haven't been resting on their laurels.

I'm not sure I want to start the discussion about VS pros and cons.
It's not about VS but how the development pipeline works.
Any specifics will be both under MSFT and Sony NDA. So. Take it or leave it.
I don't really care.
 
It's not unheard of. For instance, the most popular application platform in the world, Chromium (used by Vivaldi, Chrome, Opera, Brave etc), are using VS as just a frontend and debugging tool, the compiler from MS is not used at all. clang is used on all platforms, same as Sony is using.
Clang is indeed superior to MSVC. I have found the latter to be too permissive.
I just thought the previous poster was referring to the IDE itself.
 
Ascend you share a moderate point of view please don't ruined it with "100 GB is instantly accessible".
Right now, everything we are talking about is what both Sony and MS have claimed. I stated that MS claims developers get instant access to 100GB. Whether that's true or not, or what it really means in practice, still needs to be seen. And the same applies for all the PS5 tech.

These 3.3 seconds is if MS removed all Bottlenecks (remember Sony pushed the Bottleneck narrative), thing that we already don't know.

And PS5 I/O is equal to ~ 9 zen cores and Xbox ~5, if Im not wrong.
Isn't that exactly the same ratio as the 4.8 GB/s and the 9 GB/s speeds for the SSDs? That should ring a bell....
 
Last edited:
Yes, it's almost twice the speed. But it needs to be seen in context. That's the only way things can be looked at objectively and evaluated fairly.
The whole purpose of the SSD, and everyone agrees, is to remove the limits of the hard drive. Games needed to be designed with the hard drive in mind, because it was the major bottleneck for a long time. So everything as of now, needs to be looked at, in comparison to hard drives. Every developer is jumping from hard drives as a standard to SSDs as a standard.

So in reality, how much does twice the SSD speed of the PS5 REALLY matter? Let's do a quick calculation...
Assume you have to load 16GB of data into RAM. You have an HDD that can transfer 50MB/s, the XSX that can do 4.8 GB/s, and the PS5 that can do 9 GB/s.
The HDD takes 16000/50 = 320 seconds
The XSX SSD takes 16 / 4.8 = 3.3 seconds
The PS5 SSD takes 16 / 9 = 1.8 seconds

The jump from 320 seconds to 3.3 seconds is HUGE. The jump from 320 to 1.8 seconds is definitely bigger, but is it really THAT much more significant compared to the XSX? Compared directly against the XSX, it is almost 90% faster, sure. But the difference between the PS5 SSD and the XSX SSD compared to the HDD is 1.5 seconds over 320 seconds, which is a mere 0.5%. Sure, maybe the HDD can do 100MB/s instead of 50 MB/s, but that would still make the difference a mere 1% between the consoles.
Slamming each other and bragging about 0.5% - 1% difference in storage improvement is not exactly what I would call good use of one's time.
It's not so much the load times, not many will feel slighted by a couple of seconds more loading. The difference is in the ability to stream when the ssd becomes an extension of the entire memory pipeline.

The real-time difference is significant in feeding new material to render while you're playing. A game that takes advantage of double the speed on PS5 for streaming purpose cannot be matched by a system twice as slow in this regard. Visible concessions would need to be made, think about it - twice as many textures, or models, or animations while playing. Important to not conflate this with scene complexity, the scene budgets will still be within the rendering ceiling of the GPU, it is about detail achieved through variety, faster LOD switching, etc.

XsX is good, but PS5 cranks next gen possibilities to 11. "IF" a 3rd party game dares to show off PS5 potential, the delta will become apparent.
SSDs are going to create a paradigm shift regarding how data is handled. The combination of extremely fast read speed, the x100 faster random access, no need to arrange data on disk, and no more inside or outside of plate reading means games can work differently than they have until now. The funny thing is, MS and Sony talked about the exact same thing, streaming the world that the camera sees and nothing else, in real-time, for the next frame.

Both are doing it, Sony can do it faster. I'm not sure how the faster streaming of the PS5 will eventually look like in non-exclusive games. I mean, the SSD can stream x2 more textures, models, and other forms of data but all that extra data needs to sit in memory and PS5 doesn't have more or faster memory than XSX. And the CPU and GPU also need to be able to handle all of that stuff in order to render it to screen, but PS5 doesn't have a more powerful CPU and GPU. So I guess we will have to wait and see how these things will affect next-gen, it's going to be really interesting.


Could do it the same way Microsoft got 16GB on a 320-bit bus; Mixed density chips.,

10*64GB chips + 2*96GB Chips = 832GB. Which is pretty close to 825GB.
Unless you're saying that Mark Cerny was lying about SSD capacity - something which is very easy to debunk.
PS5 actually uses 12*64 GiB modules which is 768GiB, translate it to GB and you get 824.63GB.


XSX translated to GiB is actually 931GB.
 

For the Xbox 20/20 May update, we will focus on giving you a first look at next-gen gameplay, trailers and sneak peeks from a wide variety of publishing partners and independent developers across the globe and industry, including Ubisoft's recently announced "Assassin's Creed Valhalla," and hear from game creators about what they're doing with their games on Xbox Series X.

We'll be confirming games that utilize our new Smart Delivery feature, which ensures that you always play the best version of the games you own for your console, across generations. And all the games you'll see will be Xbox Series X Optimized, meaning they are built to take advantage of the powerful Xbox Series X features that make games look and feel incredible, including 4K resolution at up to 120 frames per second, DirectStorage, hardware-accelerated DirectX raytracing, super-fast load times and much more.
 
Last edited:
In all seriousness, is Cerny throwing some Shade when this slide comes on?
8AYatUE.jpg

I thought so the moment I saw it.

Proof will be in the games, though. I still expect the PS5 to ultimately outsell the XSX, despite the (supposed) power difference. The PlayStation brand simply has too much weight behind it.
 
SSDs are going to create a paradigm shift regarding how data is handled. The combination of extremely fast read speed, the x100 faster random access, no need to arrange data on disk, and no more inside or outside of plate reading means games can work differently than they have until now. The funny thing is, MS and Sony talked about the exact same thing, streaming the world that the camera sees and nothing else, in real-time, for the next frame.

Both are doing it, Sony can do it faster. I'm not sure how the faster streaming of the PS5 will eventually look like in non-exclusive games. I mean, the SSD can stream x2 more textures, models, and other forms of data but all that extra data needs to sit in memory and PS5 doesn't have more or faster memory than XSX. And the CPU and GPU also need to be able to handle all of that stuff in order to render it to screen, but PS5 doesn't have a more powerful CPU and GPU. So I guess we will have to wait and see how these things will affect next-gen, it's going to be really interesting.




PS5 actually uses 12*64 GiB modules which is 768GiB, translate it to GB and you get 824.63GB.


XSX translated to GiB is actually 931GB.


How dare you make the number lower!!!!!!!!

3dbf4a375442c5965d92d486cbfce1d5.gif
 
Source?
show assets vs liabilities for each company please.

Didn't I directly reply to that already? Yes, it seems like it backfired so you try to act like you didn't see it.


Sony's Short-Term Debt & Capital Lease Obligation for the quarter that ended in Dec. 2019 was $9,135 Mil. Sony's Long-Term Debt & Capital Lease Obligation for the quarter that ended in Dec. 2019 was $8,225 Mil. Sony's Total Stockholders Equity for the quarter that ended in Dec. 2019 was $38,786 Mil. Sony's debt to equity for the quarter that ended in Dec. 2019 was 0.45.


Microsoft's Short-Term Debt & Capital Lease Obligation for the quarter that ended in Mar. 2020 was $3,748 Mil. Microsoft's Long-Term Debt & Capital Lease Obligation for the quarter that ended in Mar. 2020 was $70,110 Mil. Microsoft's Total Stockholders Equity for the quarter that ended in Mar. 2020 was $114,501 Mil. Microsoft's debt to equity for the quarter that ended in Mar. 2020 was 0.65.


So assuming both will pay their debt, it's ~$40B left for Sony and -$4B debt is still to be paid by Microsoft.:goog_unsure:
 
Last edited:
The only case when I heard normally the word unit is in "Unit testing" only in that phrase/moment maybe is a problem in the translation to english but even, we
usually use the same terminology in english doesn't matter if you talk to a japanese or indian for put an example.
Yes "units" of code a something that is referred to. However, API's are not called "units" as far as I've ever heard. Sorry. May even be the case where you are, but that doesn't make it common nomenclature. Maybe that's the issue here? Maybe you're talking about units of code produced during Agile "Sprints?"
He clearly misunderstood what Cerny said on the GDC presentation (and he even quoted it in one of his previous posts), Cerny says that "the Playstation5 has a unit called the Geometry Engine..." the Playstation5! That's the console, the hardware...
Even if we give him the benefit of the doubt, and speculate that Cerny was talking about a software feature and for some reason called a unit... he would've said the Playstation5 API has a unit called the Geometry Engine!

Sigh...I'm bored so thought I'd have look see myself at the Velocity Architecture and compare it to what I think I learnt from The Road to PS5 video.

XSX:
So two pieces of hardware (SSD, decompression block) and two pieces of software (DirectStorage, SFS). In the end this gives 2.4GB/s raw, 4.8GB/s compressed (the block theoretically capable of >6GB/s) compared to 120MB/s for One X giving a 40x I/O multiplier).

PS5:
So six pieces of hardware (SSD, Kraken block, DMA controller, coherency engines, 2x co-processors) and one piece of software: a new ID-based API.

In the end this gives 5.5GB/s raw, 8-9GB/s compressed (block theoretically capable of 22GB/s) giving a theoretical 100x I/O multiplier.

That's how I currently understand things.
Very good resumen post mate.
 

$70B in debt


$5.8B in debt
Liabilities not just debt.
 
So this PL3 guy on Twitter says 7.2 GB/s vs 5.5GB/s and I wondered specifically how he came up with that. Someone correct me if I have anything wrong.

He is taking this comment from Andrew Goossen in the DF article:

Andrew Goossen said:
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.

And doing the calculation 2.4GB/s x3 for effective I/O it = 7.2GB but he then compares this with PS5's raw 5.5GB/s. Also the SFS software along with BCPack seems to be targeted specifically at textures. I also assume all this software aimed at making things as efficient as possible (a good thing BTW) still has to work withing the confines of the 2.4GBraw/4.8GB compressed figures?

Going back to Andrew's comment above he says there is an effective 2-3x multiplier for physical RAM. During the Road to PS5 video Mark showed that more of the RAM is working so most of PS5's(~12GB?) RAM is being used versus PS4 where ~2GB is being used while the rest remains idle. This equates to a ~6x effective multiplier.

This is the sort of playing with the numbers that makes things so annoying. I'm sure both systems will have amazing looking games ( Killzone:SF and Ryse still hold up great technically).
There is no need to muddy the waters like this IMO.
 
I don't know what the situation is with VS and doubt Microsoft are at any disadvantage with their own software but do remember Richard and John briefly comment in one of the DF videos about how Sony's dev environment/tools are quite a bit better than Microsoft's. That surprised me.

Can't remember which video so can't link right now.

Why did it surprise you?? Looking back at all the dx12 magic talk Sony had its own solutions and apis that were as good and better. Ppl push tht narrative that Microsoft is a software company but they sure haven't been better in software in terms of consoles. From os, to features like Shareplay, to API implementations Sony has trumped them.
 
From os, to features like Shareplay, to API implementations Sony has trumped them.
You might want to look back at the start of this gen and the first few years, because Microsoft was doing a much better job at updates and new features in the OS. API wise, Sony has always been doing better, that's been said a lot by many developers
 
Sigh...I'm bored so thought I'd have look see myself at the Velocity Architecture and compare it to what I think I learnt from The Road to PS5 video.

XSX


So two pieces of hardware (SSD, decompression block) and two pieces of software (DirectStorage, SFS). In the end this gives 2.4GB/s raw, 4.8GB/s compressed (the block theoretically capable of >6GB/s) compared to 120MB/s for One X giving a 40x I/O multiplier).

PS5


So six pieces of hardware (SSD, Kraken block, DMA controller, coherency engines, 2x co-processors) and one piece of software: a new ID-based API.

In the end this gives 5.5GB/s raw, 8-9GB/s compressed (block theoretically capable of 22GB/s) giving a theoretical 100x I/O multiplier.

That's how I currently understand things.
Not quite, for the last time, probably ;-)

Raw: 2.4GB/s Vs 5.5GB/s
Typical: ??? Vs 8/9 GB/s
Max theoretical: 4.8GB/s | 6GB/s (for textures) Vs 22GB/s

You are all going to be shocked at how fast Sony's SSD is compared to MS's. The biggest difference is going to be in the typical numbers. And we don't know XSX typical numbers.
 
Last edited:
Could do it the same way Microsoft got 16GB on a 320-bit bus; Mixed density chips.,

10*64GB chips + 2*96GB Chips = 832GB. Which is pretty close to 825GB.
Unless you're saying that Mark Cerny was lying about SSD capacity - something which is very easy to debunk.
Why are you comparing ram to storage? They are using a mixture of 1 and 2 GB chips for ram by the way
 
Last edited:
You might want to look back at the start of this gen and the first few years, because Microsoft was doing a much better job at updates and new features in the OS.
At the start of the generation, the Xbox One OS was so barebones, of course they added features more quickly (and most of those catching up to PS4 OS). They only got it more feature-rich than PS4 in the last couple of years.
 
Why did it surprise you?? Looking back at all the dx12 magic talk Sony had its own solutions and apis that were as good and better. Ppl push tht narrative that Microsoft is a software company but they sure haven't been better in software in terms of consoles. From os, to features like Shareplay, to API implementations Sony has trumped them.

I guess I'm not surprised but more wondering why a dumb, dumb like me can figure this out but nobody on Twitter seem to be calling him out and showing him (I don't do Twitter myself so can't).

Not quite, for the last time, probably ;-)

Raw: 2.4GB/s Vs 5.5GB/s
Typical: ??? Vs 8/9 GB/s
Max theoretical: 4.8GB/s | 6GB/s (for textures) Vs 22GB/s

You are all going to be shocked at how fast Sony's SSD is compared to MS. The biggest difference is going to be in the typical numbers. And we don't know MS typical numbers

I'm just being fair. You might turn out correct but right now Microsoft specs say 4.8GB compressed (with no texture qualifier) so will give the benefit of doubt for now.
 
I guess I'm not surprised but more wondering why a dumb, dumb like me can figure this out but nobody on Twitter seem to be calling him out and showing him (I don't do Twitter myself so can't).



I'm just being fair. You might turn out correct but right now Microsoft specs say 4.8GB compressed (with no texture qualifier) so will give the benefit of doubt for now.

Those guys on Twitter are a lost cause lol super fanatics. Don't bother.
 

$70B in debt


$5.8B in debt

406gfm.jpg
 
Last edited:
It's not so much the load times, not many will feel slighted by a couple of seconds more loading. The difference is in the ability to stream when the ssd becomes an extension of the entire memory pipeline.

The real-time difference is significant in feeding new material to render while you're playing. A game that takes advantage of double the speed on PS5 for streaming purpose cannot be matched by a system twice as slow in this regard. Visible concessions would need to be made, think about it - twice as many textures, or models, or animations while playing. Important to not conflate this with scene complexity, the scene budgets will still be within the rendering ceiling of the GPU, it is about detail achieved through variety, faster LOD switching, etc.

XsX is good, but PS5 cranks next gen possibilities to 11. "IF" a 3rd party game dares to show off PS5 potential, the delta will become apparent.
The XSX claims it can do the same thing, streaming from the SSD. Yes. The PS5 is faster. Yes, the streaming itself cannot be matched by a slower system. But consider what it means in practice.
Streaming is not a new thing. That's what's been done all the time from hard drives. But hard drives are slow. It has been stated clearly that the slow hard drive means filling up RAM with data for at least the next 30 seconds or so.

Now let's jump to the SSDs... So if the PS5 can load data in 0.3 seconds for example, the XSX would need 0.6 seconds, roughly speaking. The expected difference at this point is that the PS5 needs lower RAM allocation than the XSX, because the additional data that would need to be loaded in that 0.3 seconds, would simply be put in RAM beforehand instead. The XSX will have less efficient RAM usage, and at this point that is the only main known drawback in practice for the XSX regarding the whole SSD & I/O aspect. How significant it will be, once again, remains to be seen.
 
Last edited:
So this PL3 guy on Twitter says 7.2 GB/s vs 5.5GB/s and I wondered specifically how he came up with that. Someone correct me if I have anything wrong.

He is taking this comment from Andrew Goossen in the DF article:



And doing the calculation 2.4GB/s x3 for effective I/O it = 7.2GB but he then compares this with PS5's raw 5.5GB/s. Also the SFS software along with BCPack seems to be targeted specifically at textures. I also assume all this software aimed at making things as efficient as possible (a good thing BTW) still has to work withing the confines of the 2.4GBraw/4.8GB compressed figures?

Going back to Andrew's comment above he says there is an effective 2-3x multiplier for physical RAM. During the Road to PS5 video Mark showed that more of the RAM is working so most of PS5's(~12GB?) RAM is being used versus PS4 where ~2GB is being used while the rest remains idle. This equates to a ~6x effective multiplier.

This is the sort of playing with the numbers that makes things so annoying. I'm sure both systems will have amazing looking games ( Killzone:SF and Ryse still hold up great technically).
There is no need to muddy the waters like this IMO.
4,8GB/s + 50% Bpack = 7,2GB/s :)
 
SSDs are going to create a paradigm shift regarding how data is handled. The combination of extremely fast read speed, the x100 faster random access, no need to arrange data on disk, and no more inside or outside of plate reading means games can work differently than they have until now. The funny thing is, MS and Sony talked about the exact same thing, streaming the world that the camera sees and nothing else, in real-time, for the next frame.

Both are doing it, Sony can do it faster. I'm not sure how the faster streaming of the PS5 will eventually look like in non-exclusive games. I mean, the SSD can stream x2 more textures, models, and other forms of data but all that extra data needs to sit in memory and PS5 doesn't have more or faster memory than XSX. And the CPU and GPU also need to be able to handle all of that stuff in order to render it to screen, but PS5 doesn't have a more powerful CPU and GPU. So I guess we will have to wait and see how these things will affect next-gen, it's going to be really interesting.




PS5 actually uses 12*64 GiB modules which is 768GiB, translate it to GB and you get 824.63GB.


XSX translated to GiB is actually 931GB.

How are you getting an odd number using a base 2 multiplier? Has to be an even number.
 

... Xbox Series X features that make games look and feel incredible, including 4K resolution at up to 120 frames per second, DirectStorage, hardware-accelerated DirectX raytracing, super-fast load times and much more.

So now the narrative is super-fast load times 👀
 
Status
Not open for further replies.
Top Bottom