• 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.

MadAnon

Member
I'm personally sceptical about the 2ghz clock in the final console. That's silicon lotto territory. You just can't reliably output millions of chips at such clocks.
 
Reading both Era and Gaf...where are we standing now?

Getting confused about all the talk chaos.

thumb_my-mind-is-ful-of-fk-msmagen-com-my-mind-is-52763240.png
 
Last edited:
Reading both Era and Gaf...where are we standing now?

Getting confused about all the talk chaos.

Nothing is clear tbh.

Ms is aiming for 12Tflops rdna. Sony's APU tests gave a result of about 9.2Tflops RDNA. Could be an old APU for the Ps5 that wad supposed to launch this year before Sony decided to delay it for better hardware and launch next year but according to rumours the Oberon Apu (the one supposed to be Ps5) was still being tested in November.

Doesn't add up to previous rumours because:

A) The Apu is missing Raytracing hardware (which Cerny confirmed)

B) Insiders and journos said that developers (according to the target specs given to them by both Ms and Sony) confirmed that Ps5 was more powerful than Xsx.

Kleegamefan confirmed that XSX is 12 tflops rdna and that Ps5 was a bit ahead of Xsx, but not by much (10% max)
 
Last edited:
It would make sense if Arden is the final PS5 chip, given they all have the Shakespeare links. But then what is the 2GHz chip all about, and what is the Xbox chip(s)?

Bonus conspiracy! - Arden is the Xbox chip and it can also cover all the PS5 configurations to run in the Azure datacentres! :messenger_face_screaming: :messenger_grinning_sweat:
 

Gamernyc78

Banned
It would make sense if Arden is the final PS5 chip, given they all have the Shakespeare links. But then what is the 2GHz chip all about, and what is the Xbox chip(s)?

Bonus conspiracy! - Arden is the Xbox chip and it can also cover all the PS5 configurations to run in the Azure datacentres! :messenger_face_screaming: :messenger_grinning_sweat:

Bonus bonus conspiracy thy both have the same exact chip and modifications and we are all a bunch of dummies 😂😂😂 12. 5 TF all across the board!!!
 

Fake

Gold Member
It would make sense if Arden is the final PS5 chip, given they all have the Shakespeare links. But then what is the 2GHz chip all about, and what is the Xbox chip(s)?

Bonus conspiracy! - Arden is the Xbox chip and it can also cover all the PS5 configurations to run in the Azure datacentres! :messenger_face_screaming: :messenger_grinning_sweat:
Now you mention Microsoft not been talking about Azure implementation as much I like.
 

McHuj

Member
A) The Apu is missing Raytracing hardware (which Cerny confirmed)

We don't know what the PS5 raytracing solution is. It may even be something separate and different from RDNA2 that Sony provided on their own so it might not show up in these test/soc descriptions.
 

R600

Banned
Reading both Era and Gaf...where are we standing now?

Getting confused about all the talk chaos.
We are standing here :

Ariel and Oberon are the same chips, one pre silicon 2nd newer revision. On all accounts they are the same chip. Both have PS4/Pro BC.

Arden and Sparkman are same as well. Have Xbox BC and I assume Sparkman might be to Arden, what Oberon is to Ariel.

Dante and Prospero are both codenames for Xbox and PS5 dev kits, and both have Shakespear name.

Judging by BC mode in Arden, as well as BW/bus width (matching Scarlett video) I cant see how they get to the point of thinking Arden is PS5 (it was rumored to be Xbox back in January).
 

McHuj

Member
Cerny confirmed the APU have RT missing?

Of course not. He simply said way back when that PS5 has RT hardware and left it at that. Forums speculate what that means, but the reality his statement can mean anything above the RDNA1 baseline. We won't know until the reveal.

In my opinion, XSX will have the same solution that AMD will have with RDNA2 GPU's. PS5 may have that or something else.
 

Bryank75

Banned
Nothing is clear tbh.

Ms is aiming for 12Tflops rdna. Sony's APU tests gave a result of about 9.2Tflops RDNA. Could be an old APU for the Ps5 that wad supposed to launch this year before Sony decided to delay it for better hardware and launch next year but according to rumours the Oberon Apu (the one supposed to be Ps5) was still being tested in November.

Doesn't add up to previous rumours because:

A) The Apu is missing Raytracing hardware (which Cerny confirmed)

B) Insiders and journos said that developers (according to the target specs given to them by both Ms and Sony) confirmed that Ps5 was more powerful than Xsx.

Kleegamefan confirmed that XSX is 12 tflops rdna and that Ps5 was a bit ahead of Xsx, but not by much (10% max)
Excellent summary!
 

MadAnon

Member
We are standing here :

Ariel and Oberon are the same chips, one pre silicon 2nd newer revision. On all accounts they are the same chip. Both have PS4/Pro BC.

Arden and Sparkman are same as well. Have Xbox BC and I assume Sparkman might be to Arden, what Oberon is to Ariel.

Dante and Prospero are both codenames for Xbox and PS5 dev kits, and both have Shakespear name.

Judging by BC mode in Arden, as well as BW/bus width (matching Scarlett video) I cant see how they get to the point of thinking Arden is PS5 (it was rumored to be Xbox back in January).
Dante has nothing to do with Shakespeare. Why would PS codenames be consistent across the whole project including devkits while XBox are all over the place?
 
Last edited:

R600

Banned
Dante has nothing to do with Shakespeare.
Of course, sorry, Dante - writer and poet.

Arden - full Shakespear work of plays and poems.

Oberon, Gonzalo, Prospero, Flute, Oberon - CHARACTERS in Shakespear books. Oberon being King of feries.

BTW these seem to be internal AMD codenames. As they give for example Notebook APUs (Van Gogh, Renoir - artists).
 

R600

Banned
Why would PS codenames be consistent while XBox are all over the place?
PS are consistant. All of them are characters in Shakespear.

Arden is book of plays and poems from Shakespear. Dante is poet and writer.

Biggest issue for Arden being PS5 is it containing Xbox and X BC compatability as well as 320bit bus (Scarlet video). Also, Brad Sams leaked it back in January as being Xbox chip.

Oberon and Ariel have PS4/Pro and 256bit bus.

I think AMD gives codenames to chip, and they have different names then MS/Sony internal codenames (this was a case during Durango/Orbis last gen).
 

Aceofspades

Banned
It would make sense if Arden is the final PS5 chip, given they all have the Shakespeare links. But then what is the 2GHz chip all about, and what is the Xbox chip(s)?

Bonus conspiracy! - Arden is the Xbox chip and it can also cover all the PS5 configurations to run in the Azure datacentres! :messenger_face_screaming: :messenger_grinning_sweat:

Now Arden is PS5 too?! Wasn't confirmed by MS army to be Xbox? Seems like PS5 will be too powerful it will emulate Xbox One too 🤣

Komachi and other Twitter dude are full of BS.
 
Last edited:

R600

Banned
I think Sony aimed at 1.8GHZ at 7nm and went with 7P revision, therefore Oberon - 2.0GHZ.

Its very hard to imagine Navi with RT and 9.2TF at these clocks in console (standard, not PC case one). Its very much pushing the boundaries IMO.
 
I think Sony aimed at 1.8GHZ at 7nm and went with 7P revision, therefore Oberon - 2.0GHZ.

Its very hard to imagine Navi with RT and 9.2TF at these clocks in console (standard, not PC case one). Its very much pushing the boundaries IMO.
2 ghz doesn’t make sense in my opinion as it’s asking for yield issues and heating issues but I guess soon we ll see .
 

Aceofspades

Banned
I think Sony aimed at 1.8GHZ at 7nm and went with 7P revision, therefore Oberon - 2.0GHZ.

Its very hard to imagine Navi with RT and 9.2TF at these clocks in console (standard, not PC case one). Its very much pushing the boundaries IMO.

Sony is emulating a bigger CU chip by clocking what they have ready at the time (36CUs). Navi development has started with Sony+AMD
PS5 won't be 36CUs or 40CUs, it will be bigger.
 
Last edited:
Nothing is clear tbh.

Ms is aiming for 12Tflops rdna. Sony's APU tests gave a result of about 9.2Tflops RDNA. Could be an old APU for the Ps5 that wad supposed to launch this year before Sony decided to delay it for better hardware and launch next year but according to rumours the Oberon Apu (the one supposed to be Ps5) was still being tested in November.

Doesn't add up to previous rumours because:

A) The Apu is missing Raytracing hardware (which Cerny confirmed)

B) Insiders and journos said that developers (according to the target specs given to them by both Ms and Sony) confirmed that Ps5 was more powerful than Xsx.

Kleegamefan confirmed that XSX is 12 tflops rdna and that Ps5 was a bit ahead of Xsx, but not by much (10% max)
Excellent summary!

Yep, quite good summary. But also was told before, both XSX and PS5 will use Big Navi chips, which is in AMD roadmap.

But who cares, only AMD's test data are legit, so big NO!
 

R600

Banned
Sony is emulating a bigger CU chip by clocking what they have ready at the time (36CUs).
PS5 won't be 36CUs or 40CUs, it will be bigger.
But its not. It says right there "full chip 18WGPs native clock - 2.0GHZ".

When they emulate PS4 pro they use 18WGPs at 911MHZ.

With 56CU chip they would get :

2-3TF more (depending on clock)
20-30% less Pixel Fillrate duo to lower clock
Bigger chip and worse yields
Requirement for wider bus, therefore bigger die size and yet still less BW per TF.

Oberon vs Arden difference is half of that in X vs Pro. Except die size in this case will be even smaller for PS side.
 
Last edited:

ManaByte

Gold Member
I really wouldn’t trust what “journalists” claim about the hardware. They go where the ad traffic comes from. In 2013 they were sucking up to the Xbox One and trying to cover-up its issues because the 360 was the more popular system. In 2019 that’s the PlayStation 4 so they’ll say stuff just to keep Sony fans clicking on their bullshit as long as possible.
 

MadAnon

Member
PS are consistant. All of them are characters in Shakespear.

Arden is book of plays and poems from Shakespear. Dante is poet and writer.

Biggest issue for Arden being PS5 is it containing Xbox and X BC compatability as well as 320bit bus (Scarlet video). Also, Brad Sams leaked it back in January as being Xbox chip.

Oberon and Ariel have PS4/Pro and 256bit bus.

I think AMD gives codenames to chip, and they have different names then MS/Sony internal codenames (this was a case during Durango/Orbis last gen).
But it really doesn't containt Xbox BC modes. If anything it has the same BC scheme as Oberon (native, gen1, gen0) but no clocks or CUs for each mode. Sparkman does have connection to Xbox One in it's only gen0 BC mode as you pointed out yesterday (4RBs).

I'm bothered enough to not fully buy into Oberon is 100% PS5 final chip. Have a suspicion Arden is PS5 and Sparkman is SeX. Maybe I'm just overthinking out of boredom.
 
Last edited:
  • Thoughtful
Reactions: TLZ

Aceofspades

Banned
But its not. It says right there "full chip 18WGPs native clock - 2.0GHZ".

When they emulate PS4 pro they use 18WGPs at 911MHZ.

With 56CU chip they would get :

2-3TF more (depending on clock)
20-30% less Pixel Fillrate duo to lower clock
Bigger chip and worse yields
Requirement for wider bus, therefore bigger die size and yet still less BW per TF.

Oberon vs Arden difference is half of that in X vs Pro. Except die size in this case will be even smaller for PS side.

But it will, Sony wouldn't mind getting bigger chip and worse yields.

Your fillrate comparison doesn't make sense because if it held any merit that would mean PS2 is way stronger than PS5 and XsX combined.
 
Last edited:

Fake

Gold Member
I really wouldn’t trust what “journalists” claim about the hardware. They go where the ad traffic comes from. In 2013 they were sucking up to the Xbox One and trying to cover-up its issues because the 360 was the more popular system. In 2019 that’s the PlayStation 4 so they’ll say stuff just to keep Sony fans clicking on their bullshit as long as possible.
I believe more in journalists than insiders. At least they put their reputation at risk unlike a random guy that just claim to be an insider.

Actually devs are in my top list of trust ahead anyone. Aka codemasters confirming the devkit design.
 
Last edited:

Gamernyc78

Banned
I believe more in journalists than insiders. At least they put their reputation at risk unlike a random guy that just claim to be an insider.

Actually devs are in my top list of trust ahead anyone. Aka codemasters confirming the devkit design.

I know one group of ppl we shouldn't trust! Talking heads, super fanboys or ex pr heads for the opposing team aka Mistermediax, colteastwood, penello, etc
 
Last edited:

ManaByte

Gold Member
I believe more in journalists than insiders. At least they put their reputation at risk unlike a random guy that just claim to be an insider.

Actually devs are in my top list of trust ahead anyone. Aka codemasters confirming the devkit design.

Journalists don’t care about their reputation. They only care about traffic. In 2013 polygon flat out lied about the PS4 on their launch stream claiming people should buy a Xbox One over it because the PS4 doesn’t even support a remote control.

Then you had that idiot on IGN who said the human eye can’t see 1080p. You think he cares about his reputation? No, he failed upwards and now works for EA.
 

R600

Banned
But it really doesn't containt Xbox BC modes. If anything it has the same BC scheme as Oberon (native, gen1, gen0) but no clocks or CUs for each mode. Sparkman does have connection to Xbox One in it's only gen0 BC mode as you pointed out yesterday (4RBs).

I'm bothered enough to not fully buy into Oberon is 100% PS5 final chip. Have a suspicion Arden is PS5 and Sparkman is SeX. Maybe I'm just overthinking out of boredom.
Gen0 and Gen1 only point to which last gen console BC is meant for.

Gen0 in Oberon case - PS4 (same clocks - 800MHZ)

Gen1 in Oberon case - Pro (same clocks - 911MHZ)

Gen0 in Arden case - XBone (same RBs)

Gen1 in Arden case - XBX (same RBs)

Xbone is only console last gen with 16ROPs - thereforr 4RBs (unlike PS4, Pro or X).

Fact is, Oberon Gen0/1 is tied to PS4/Pro. Arden is tied to Xbone/XbX.
 
Gen0 and Gen1 only point to which last gen console BC is meant for.

Gen0 in Oberon case - PS4 (same clocks - 800MHZ)

Gen1 in Oberon case - Pro (same clocks - 911MHZ)

Gen0 in Arden case - XBone (same RBs)

Gen1 in Arden case - XBX (same RBs)

Xbone is only console last gen with 16ROPs - thereforr 4RBs (unlike PS4, Pro or X).

Fact is, Oberon Gen0/1 is tied to PS4/Pro. Arden is tied to Xbone/XbX.
Very good info. Thx for your hard work btw.
 

Aceofspades

Banned
Gen0 and Gen1 only point to which last gen console BC is meant for.

Gen0 in Oberon case - PS4 (same clocks - 800MHZ)

Gen1 in Oberon case - Pro (same clocks - 911MHZ)

Gen0 in Arden case - XBone (same RBs)

Gen1 in Arden case - XBX (same RBs)

Xbone is only console last gen with 16ROPs - thereforr 4RBs (unlike PS4, Pro or X).

Fact is, Oberon Gen0/1 is tied to PS4/Pro. Arden is tied to Xbone/XbX.

And this recent development of Komachi and the other dude mixing up both codenames tells you that they know nothing.

Edit: I'm now on the camp that is waiting for the reveals , I don't trust no one, not Klee nor matt or any Twitter people . Im just just waiting for the reveals now.
 
Last edited:

HeisenbergFX4

Gold Member
And this recent development of Komachi and the other dude mixing up both codenames tells you that they know nothing.

Edit: I'm now on the camp that is waiting for the reveals , I don't trust no one, not Klee nor matt or any Twitter people . Im just just waiting for the reveals now.

They will be close in power and it will come down to games, trust me.
 

Ovech-King

Member
I believe Klee and he confirmed it’s 12.08 rdna TF .

also Phil wouldn’t just retweet gamespot article where it mentions 12 Navi tf. Do you know the back lash if it’s inaccurate ?

to me xsx is 12 Navi TF

What worrying me about klee claims is that he then said we won't be impressed by the numbers. Sounds like lower tflops number yet very efficient to me
 
I think a lot of people are assuming Lockhart will be much weaker than PS5.

Lockhart and PS5 are similar in performance, with the slight edge for PS5 because of RT or whatever suits your fancy.

XSX is meant to play the same role as XOX did this gen.

XSX = ~12 TF
Lockhart/PS5 = ~9 TF
 

R600

Banned
And this recent development of Komachi and the other dude mixing up both codenames tells you that they know nothing.

Edit: I'm now on the camp that is waiting for the reveals , I don't trust no one, not Klee nor matt or any Twitter people . Im just just waiting for the reveals now.
Its from AMD (now entire repo and tweets are deleted).

Its there, black on white.
 

FranXico

Member
Komachi and other Twitter dude are full of BS.
They are not full of shit, the test data they found appears to be related to consoles (namely BC on PS5), but there is no certainty about which exact tests, which chips (mind you, internal prototype chips usually start as subsets of what ends up in devkits and commercial products), or how far along those tests were made.

That data was found at some point mid this year, it is not clear from which tests or when exactly. There are also a few inconsistencies here and there, but that's just because only AMD knows what goes on in AMD.

If we assume that those were tests being done on final hardware, and that the full number of CUs is the same as the number of CUs for PS4 Pro compatibility, and based on certain comments found therein, then the PS5 GPU does not support RT, and its maximum theoretical throughput is around 9.2TF. Which would make Cerny a baldfaced liar.

I expect a DF (Richard Leadbetter) article fully supporting this stance on Monday.

I will assume PS5 at 9TF until official reveal. We'll see.
 

Fake

Gold Member
Journalists don’t care about their reputation. They only care about traffic. In 2013 polygon flat out lied about the PS4 on their launch stream claiming people should buy a Xbox One over it because the PS4 doesn’t even support a remote control.

Then you had that idiot on IGN who said the human eye can’t see 1080p. You think he cares about his reputation? No, he failed upwards and now works for EA.
You're doing generalization. Cold down a bit.
Not all game journalist are those kind of person you mention. Its just like saying all mods are shit because resetera.

I not forcing you to believe in game journalist. You believe whatever thafuck you want. I'll keep devs in my trust list. Not here to convence you otherwise.
 
I think a lot of people are assuming Lockhart will be much weaker than PS5.

Lockhart and PS5 are similar in performance, with the slight edge for PS5 because of RT or whatever suits your fancy.

XSX is meant to play the same role as XOX did this gen.

XSX = ~12 TF
Lockhart/PS5 = ~9 TF

This doesn't make a whole lot of sense. Lockhart loses its value proposition by targeting PS5 level of performance (provided PS5 is actually around 9-9.5 TFLOPs). MS's made it pretty clear Lockhart is a low-cost alternative, and preferably it'd be tuned for streaming. Therefore it wouldn't need to be anywhere near PS5 in terms of power since it wouldn't be focusing as much on local (non-streaming) gaming.

$199 Lockhart/XSES, $499 XSEX. Hell, maybe even $399 XSEX if MS is willing to lose that much on each system sold. Lockhart at (purported) PS5 specs loses any strong selling point or value proposition to MS and customers.
 
  • Like
Reactions: TLZ

Mass Shift

Member
They are not full of shit, the test data they found appears to be related to consoles (namely BC on PS5), but there is no certainty about which exact tests, which chips (mind you, internal prototype chips usually start as subsets of what ends up in devkits and commercial products), or how far along those tests were made.

That data was found at some point mid this year, it is not clear from which tests or when exactly. There are also a few inconsistencies here and there, but that's just because only AMD knows what goes on in AMD.

If we assume that those were tests being done on final hardware, and that the full number of CUs is the same as the number of CUs for PS4 Pro compatibility, and based on certain comments found therein, then the PS5 GPU does not support RT, and its maximum theoretical throughput is around 9.2TF. Which would make Cerny a baldfaced liar.

I expect a DF (Richard Leadbetter) article fully supporting this stance on Monday.

I will assume PS5 at 9TF until official reveal. We'll see.

Indeed. I'm holding on to hope until then.
 
Status
Not open for further replies.
Top Bottom