• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

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

Status
Not open for further replies.

Moses85

Member
599€???
giphy.gif


Chances are rising with this price to get one of these boxes.
 

hemo memo

Gold Member
Indeed. I dont expect them to be sprung on us at all. My friend and former coworker at GameStop for many years who is still there believes they'll know in advance(conference call, a weekly payroll increase or marketing stuff). He'll warn me and I'll warn you guys subsequenly if it happens that way.

Please do. Thank you.
 
I am down for this, it's sorta frustrating to wake up every day and keep watching the June 11th event reaction videos on youtube and wait patiently until we hear and something more concrete about PS5 :(

It's what I'm hoping for to be honest. They could just record a short video with the launch details and price. Then they can have another event where we see more of the hardware and games. It's how I would do it if I was really concerned about the competition undercutting my product.
 

Bo_Hazem

Banned
Oh, this is amazing, bro. So beautiful nature. Mmmm, parallax effect motion, nice editing.

Thanks, bro, was testing the new drone that got a zoom feature update lately, so that dolly/parallx effect is done natively with the zoom feature instead of pure editing sorcery ;) . This is the first flight with the Mavic Air 2. We're active here if you're interested in photography/videography:


Did you manage to find the PS5 somewhere in there? Last known whereabouts of Cerny?

Mark Cerny isn't a true person, he's a program, made out of pure artificial intelligence, the highest in that regard. :lollipop_sunglasses:
 
Last edited:

Mr Moose

Member
Screenshot or die!
Mine just says "PS5 is coming | Discover a deeper gaming experience - Lightning speed | Breathtaking immersion | Stunning games" and has a trailer and a link to the PS5 page. Not sure if it's the same as the one a few weeks back, too lazy to check, I think its slightly different.
 

Bo_Hazem

Banned
Bo_Hazem Bo_Hazem @sircaw @Bryank75 do you think Sony might wait until October to start talking PS5 again, by talking I am referring to another event where you see in depth gameplay demos, UI reveal, price and pre order info and probably a couple of unannounced titles gets revealed and then November the console launches. Or do you think Sony already has an event edited out and ready to be aired the moment Microsoft reveals Series S and pricing for Series S and X?

Can't say, Sony is pretty unpredictable. I can see something happening this month, it's not good to go radio silent this month. We need gameplay from Spiderman MM or Demon's Souls at least.

I think the new cards will make Sony and Microsoft calculate their pricing wisely.
 

IntentionalPun

Ask me about my wife's perfect butthole
Pretty much but I don’t blame him because that’s the rule not the exception regarding random people on twitter talking about PS5 tech.
It's almost like we shouldn't be posting what random people on twitter say on this forum....

The PS5 I/O is over twice as fast + even better compression.. it's gonna do things the XSX can't. We'll really have to wait and see what that entails though as far as games (other than much better load times, which is enough for me to easily make PS5 my lead platform)
 
Last edited:

hemo memo

Gold Member
It's almost like we shouldn't be posting what random people on twitter say on this forum....

I’ve been saying this for a long time yet we still get posts from random fake insiders on here. Same fake insiders that got proven fake and wrong over and over and over again. People are just so desperate to the point of following every random rumor without logic.
 
Incorrect, the i/o block decompresses in real-time

Edit: on PS5 at least , can not confirm XsX. Although also likely.

Got a source for that? As I said, it was according to Sony’s IO patent that data streams into RAM before the IO block checks in the data into where it’s needed by the application.
 

IntentionalPun

Ask me about my wife's perfect butthole
I’ve been saying this for a long time yet we still get posts from random fake insiders on here. Same fake insiders that got proven fake and wrong over and over and over again. People are just so desperate to the point of following every random rumor without logic.
We'll literally have like half the posts on the page be about some Xbox fanboy on twitter being a moron, and I'll click their profile and they'll have 12 followers lol Just as bad as the fake insiders to me.. who f-ing cares?

I complain about it weekly lol
 

Shmunter

Member
Got a source for that? As I said, it was according to Sony’s IO patent that data streams into RAM before the IO block checks in the data into where it’s needed by the application.
Yes directly from Cerney, in his GCC announcement. Sorry, you’ll need to sift through it.

Edit : Found a summary, you’re welcome...

“The next component is the Dedicated DMA Controller. This controller directs the game on where to send the data flow coming out of the decompressed data from the decompressor. That is the job of this chip. It gets the decompressed data and tells the game where to send it. Cerny states " this equates to ANOTHER Zen2 core or two worth of performance ". So again it is a powerful unit for processing the data but it is much less intensive since it is just basically a traffic controller. You go here, you go there, you go there. So it doesn't need a gigantic amount of power to handle the decompressed data flow.”

 
Last edited:

hemo memo

Gold Member
It's almost like we shouldn't be posting what random people on twitter say on this forum....

The PS5 I/O is over twice as fast + even better compression.. it's gonna do things the XSX can't. We'll really have to wait and see what that entails though as far as games (other than much better load times, which is enough for me to easily make PS5 my lead platform)

My lead platform is PS5 because of PS Studios games regardless of power.
 

IntentionalPun

Ask me about my wife's perfect butthole
what does it say? I am unable to view the tweet nor the message posted by the original gaf member
XSS has advanced VRS, PS5 doesn't appear to have any... XBox has specific hardware for Mesh Shaders, PS5 only has primitive shaders (but primitive shaders are pretty good.)

Probably more absolute made up bullshit since that's all this forum seems to want to talk about lol
 

IntentionalPun

Ask me about my wife's perfect butthole
My lead platform is PS5 because of PS Studios games regardless of power.
Well I mean for 3rd parties.

I also don't think the graphics will be better on XSX.. but even if they are, I'd choose load time over graphics for 3rd party games unless for some strange reason framerates were also worse.
 

IntentionalPun

Ask me about my wife's perfect butthole
Dude, much more than 2x. You just... It's really fucking fast, no joke.
Raw speed wise it's a bit over 2x as fast as XSX. Then as I mentioned, it has better compression. But compression is just for textures and you can't just multiply that in for all I/O.
 
Last edited:
Yes directly from Cerney, in his GCC announcement. Sorry, you’ll need to sift through it.

Edit : Found a summary, you’re welcome...

“The next component is the Dedicated DMA Controller. This controller directs the game on where to send the data flow coming out of the decompressed data from the decompressor. That is the job of this chip. It gets the decompressed data and tells the game where to send it. Cerny states " this equates to ANOTHER Zen2 core or two worth of performance ". So again it is a powerful unit for processing the data but it is much less intensive since it is just basically a traffic controller. You go here, you go there, you go there. So it doesn't need a gigantic amount of power to handle the decompressed data flow.”



The summary you quoted doesn’t negate what I understand Sony’s IO patent to be saying. The reddit post you quoted from is also off the mark on several other points.

Nothing in Cerny’s talk negates what the IO patent with his name on says, either.

Also, data streaming from the SSD into RAM before being checked-in by the IO Complex doesn’t mean it’s not in “real time”. The IO Complex doesn’t need to wait until the whole thing is copied across before it begins.

The same would be true of what Nvidia aims to do, from my understanding, too. Which is why I brought it up.
 
Last edited:

Shmunter

Member
The summary you quoted doesn’t negate what I understand Sony’s IO patent to be saying. The reddit post you quoted from is also off the mark on several other points.

Nothing in Cerny’s talk negates what the IO patent with his name on says, either.

Also, data streaming from the SSD into RAM before being checked-in by the IO Complex doesn’t mean it’s not in “real time”. The IO Complex doesn’t need to wait until the whole thing is copied across before it begins.

The same would be true of what Nvidia aims to do, from my understanding, too. Which is why I brought it up.
Sorry, you’re completely wrong. It wouldn’t even be logical to have a decompressor bolted onto the i/o if it wasn’t decompresssing during you know....I/o

Not the case for a pc GPU, unless there is something I’m not aware off
 
Last edited:
Raw speed wise it's a bit over 2x as fast as XSX. Then as I mentioned, it has better compression. But compression is just for textures and you can't just multiply that in for all I/O.
It's not about i/o itself, in raw numbers. Data management in the PS5 in the most inconvenient, uncomfortable case is significantly superior to the XsX solution. Just wait and see what I'm talking about. Ok?
 

ethomaz

Banned

VRS tiers is due old hardware limitation.

Tier 0: No support.
Tier 1: Hardware support with limitations (eg. RDNA)
Tier 2: Actual hardware support (eg. RDNA 2).
Tier 3? Future GPUs hardware (look at the table below there is feature with N in Tier 2).

That is how Direct X works... they add new Tier with new hardware so you have a abstraction layer for old and new hardware each one supporting a set of features (Tiers)

tiers.png


 
Last edited:
Sorry, you completely wrong. It wouldn’t even be logical to have a decompressor bolted onto the i/o if it wasn’t decompresssing during you know....I/o

Not the case for a pc GPU, unless there is something I’m not aware off

Yes there is something you’re not aware of. As I said in the last reply, the SSD initially streaming into RAM before the IO Complex performs check-in doesn’t mean that it’s not real-time, or that it’s somehow now not part of “I/O” any more.

This was in response to you saying PC is inefficient as the data needs to be copied to a GPU’s VRAM before the GPU can decompress it.

But according to Sony’s on IO patent the PS5 essentially works the same way.

This doesn’t mean it’s not real-time. Neither does it mean it’s an extra step on PC, either, if DirectStorage and Nvidia will be streaming straight from SSD to VRAM.

System RAM (kernel) is used as a staging area and buffer for the IO Complex before it is checked “back” into System RAM (user) for the application.

At least that’s what Sony’s patent says is how their IO Complex works.

None of this contradicts what Cerny said at all. Check-in is “eliminated” as far as the CPU can see as it doesn’t see it happening, but the IO Complex is still performing that critical task of decompressing, decrypting etc from kernel memory to user memory all on the same RAM.

That’s why I was curious for a source and politely asked for one, and why I believe what Nvidia and Microsoft aim to do is fundamentally the same as far as data flow goes.

A decompressor can start working as soon as the first chunk of data hits RAM, and also deallocate the original staging area RAM once it has dealt with it.
It’s not some Windows file copy where you need to wait for the whole thing to he copied over and occupy its full space before that can begin. That applies to PS5 and Nvidia + future DirectStorage.

Instead of just telling me I’m wrong over and over, and telling me I’m welcome for some reddit amateur analysis of what’s going on; how about we all just focus on primary sources and actually trying to figure out how all of this cool shit works.

To my knowledge Nvidia’s approach to decompressing data bound for the GPU isn’t inefficient (or less efficient than PS5) because it’s decompressing from and to its VRAM. I’m happy to be corrected—I’m only here to learn. I asked for a source in case I’d missed out on some other new patent or primary source that supersedes what I already learned from the patent someone else helpfully provided in this thread months ago.
 
Last edited:

Shmunter

Member
Yes there is something you’re not aware of. As I said in the last reply, the SSD initially streaming into RAM before the IO Complex performs check-in doesn’t mean that it’s not real-time, or that it’s somehow now not part of “I/O” any more.

This was in response to you saying PC is inefficient as the data needs to be copied to a GPU’s VRAM before the GPU can decompress it.

But according to Sony’s on IO patent the PS5 essentially works the same way.

This doesn’t mean it’s not real-time. Neither does it mean it’s an extra step on PC, either, if DirectStorage and Nvidia will be streaming straight from SSD to VRAM.

System RAM (kernel) is used as a staging area and buffer for the IO Complex before it is checked “back” into System RAM (user) for the application.

At least that’s what Sony’s patent says is how their IO Complex works.

None of this contradicts what Cerny said at all. Check-in is “eliminated” as far as the CPU can see as it doesn’t see it happening, but the IO Complex is still performing that critical task of decompressing, decrypting etc from kernel memory to user memory all on the same RAM.

That’s why I was curious for a source and politely asked for one, and why I believe what Nvidia and Microsoft aim to do is fundamentally the same as far as data flow goes.

A decompressor can start working as soon as the first chunk of data hits RAM, and also deallocate the original staging area RAM once it has dealt with it.
It’s not some Windows file copy where you need to wait for the whole thing to he copied over and occupy its full space before that can begin. That applies to PS5 and Nvidia + future DirectStorage.

Instead of just telling me I’m wrong over and over, and telling me I’m welcome for some reddit amateur analysis of what’s going on; how about we all just focus on primary sources and actually trying to figure out how all of this cool shit works.

To my knowledge Nvidia’s approach to decompressing data bound for the GPU isn’t inefficient (or less efficient than PS5) because it’s decompressing from and to its VRAM. I’m happy to be corrected—I’m only here to learn. I asked for a source in case I’d missed out on some other new patent or primary source that supersedes what I already learned from the patent someone else helpfully provided in this thread months ago.
You’re reaching for the stars when the most simple answer and even stated info is in front of you. You’ll just need to wait for a tear down to convince yourself.
 
You're wrong. Data that is unpacked from ssd, stored and ready to be read in memory. The whole process is very fast. Your method or from the patent, it's wrong, a little dumb and meaningless.

Another person telling me I’m wrong.
Have you got a source for that? As I said, I’m only going by what Sony’s own patent said, which to my knowledge is the deepest dive we’ve had yet on what they’re doing, and also in no way contradicted by Cerny’s Road to PS5 sermon.

This isn’t my own idea or theory. If you want to say the patent is wrong then I’m interested to know why as I’ve clearly missed something.
 

geordiemp

Member
Another person telling me I’m wrong.
Have you got a source for that? As I said, I’m only going by what Sony’s own patent said, which to my knowledge is the deepest dive we’ve had yet on what they’re doing, and also in no way contradicted by Cerny’s Road to PS5 sermon.

This isn’t my own idea or theory. If you want to say the patent is wrong then I’m interested to know why as I’ve clearly missed something.

The source is the schematic of Cernys presentation, which shows local SRAM on the IO and the flow of data is clear.

You have seen it posted on GAF allot.
 
Status
Not open for further replies.
Top Bottom