Baseless Next Generation Rumors with no Technical Merits [post E3 2019, pre GDC 2020] [XBSX, PS5]

[BRiT]

So, The best news of the day that no one is talking about (yet) ...

With all the power efficiency gains AMD talked about for their GPU and CPU, a SOC with LockHart's rumored specifications would be in the 22 Watt range. Sure would fit with the Surface product line. Does that sound about right?

 

[anexanhume]

So, The best news of the day that no one is talking about (yet) ...

With all the power efficiency gains AMD talked about for their GPU and CPU, a SOC with LockHart's rumored specifications would be in the 22 Watt range. Sure would fit with the Surface product line. Does that sound about right?

Don’t forget the PS4 portable.

 

[lynux3]

https://www.resetera.com/threads/ps5-and-xbox-series-speculation-ot12-aw-hell-transistors-running-this-again.173318/page-30

A lot of Eras say OBR/Ariel is Navi 10 (or Navi 10 lite). Is't it real?

It comes from the _rogame tweet.

 

[iroboto]

f_**ing guy, who verified Brit?

 

[McHuj]

While RDNA2 having 50% performance per watt is a good thing, don’t set yourself up with unreasonable expectations.

If I were a betting man, I would bet that this is an efficiency gain in the shader not the overall GPU. Things like caches and IO certainly won’t see gains like that, so I expect the overall GPU to see less drastic gains in power efficiency.

 

[anexanhume]

While RDNA2 having 50% performance per watt is a good thing, don’t set yourself up with unreasonable expectations.

If I were a betting man, I would bet that this is an efficiency gain in the shader not the overall GPU. Things like caches and IO certainly won’t see gains like that, so I expect the overall GPU to see less drastic gains in power efficiency.

Bunch of VRS gains getting claimed I bet.

 

[Proelite]

It comes from the _rogame tweet.

Won't be the first time he is wrong about something. 14gbps GDDR6 is terribly off.

 

[Globalisateur]

No, no it isn't. Era can show me where Oberon is specified as Navi 10 Lite and I will drop months of cult leadership here and admit I am wrong (which, I'm pretty sure I'm not).

Ariel - Navi 10 Lite. RDNA1 chip
Oberon - RDNA2 chip
Github Oberon regression tests (for native, BC1 and BC2) - running against Ariel iGPU testlist (therefore CANNOT contain any trace of RDNA2 features).

Ariel - Gonzalo platform (13E9, 13F8)
Oberon - Flute platform (13F9)

Indeed. Also I often read people arguing about the different steppings, A0, B0 about those chips and what they could mean for the silicon modifications. But that's not looking at the whole picture. Steppings are puny compared to whole different SOCs ! Ariel and Oberon are 2 different SOCs ! Oberon could (and should) be a very different chip than Ariel.

 

[Pinstripe]

Over 400 pages and people still argue about RDNA1 vs RDNA2?

I thought Navi10/RDNA1 was needed for GCN/PS4 backward compatibility. So why suddenly the RDNA2 in PS5 conjecture?

 

[AbsoluteBeginner]

It comes from the _rogame tweet.

And its wrong. I guess he was "trigger" happy when posting these tweets. But yea, no where in Github does it say Oberon is Navi 10. Although, its running regression tests for Navi 10 chip (Ariel)

 

[goonergaz]

Then it comes down to a 9.2TF vs 12TF difference.... About what we have now with Pro vs One X. Exclusives could show a difference perhaps, but 3rd party games a resolution/fps difference.

The comparison isn’t right, that’s just the GPU - the X has more advantages over pro so I don’t think the gap will be as big if everything else is equal or PS5 has some advantage(s).

 

[N2O]

Over 400 pages and people still argue about RDNA1 vs RDNA2?

I thought Navi10/RDNA1 was needed for GCN/PS4 backward compatibility. So why suddenly the RDNA2 in PS5 conjecture?

To be fair, most of time is arguing 9.2TF or not.

 

[Shifty Geezer]

Might be interesting to see how 2SE@2Ghz+ vs 3SE@1675 performs -- faster front-ends vs more front-ends.

Looks like the only 'interesting' part about next gen tech. At least DF will still be able to earn a living comparing fast versus wide.

 

[Kaotik]

Over 400 pages and people still argue about RDNA1 vs RDNA2?

I thought Navi10/RDNA1 was needed for GCN/PS4 backward compatibility. So why suddenly the RDNA2 in PS5 conjecture?

Sounds like you bought into the "RDNA is halfway between GCN and "real RDNA"" narrative which some sites cooked up.
RDNA2 has its roots in GCN just like RDNA1

 

[N2O]

In recent year i start using DF(and VG tech etc) to check out which game have half assed optimization, for me it's more interesting than just comparing which console is better.

 

[lynux3]

And its wrong. I guess he was "trigger" happy when posting these tweets. But yea, no where in Github does it say Oberon is Navi 10. Although, its running regression tests for Navi 10 chip (Ariel)

Flute is also supposed to be Navi 10, right?

 

[AbsoluteBeginner]

Flute is also supposed to be Navi 10, right?

Why do you people think Flute is Navi 10?

Point of my entire post here - https://forum.beyond3d.com/posts/2106613/ is that Flute is not Gonzalo, and Oberon is not Ariel.

Ariel is chip. Oberon is chip. Gonzalo is platform. Flute is platform. Ariel is found in Gonzalo benchmark, Oberon is found in Flute benchmark (check PCI IDs).

PCI ID for Ariel is different then the one for Oberon. Oberon is later then Ariel. Ariel is only one found to have any resemblance to Navi 10 through driver data mining, Oberon was never found there. Only reason why people think Oberon is Navi 10 is because it runs regression test on Ariel testlist specifications.



But entire point of regression test is testing functionality (native/bc1/bc2) of previous version of chip (Ariel), therefore RT/VRS would never be shown there, because Ariel (ie Navi 10 Lite) doesn't have one.

Its very very simple to grasp and entire logic cannot be explained any other way.

Ariel - Navi 10 Lite chip, RDNA1, perhaps used by Sony to get BC working on RDNA architecture
OBR - RDNA2 chip, regression tested against Ariel testlist in Github leak

 

[lynux3]

Why do you people think Flute is Navi 10?

Point of my entire post here - https://forum.beyond3d.com/posts/2106613/ is that Flute is not Gonzalo, and Oberon is not Ariel.

Ariel is chip. Oberon is chip. Gonzalo is platform. Flute is platform. Ariel is found in Gonzalo benchmark, Oberon is found in Flute benchmark (check PCI IDs).

PCI ID for Ariel is different then the one for Oberon. Oberon is later then Ariel. Ariel is only one found to have any resemblance to Navi 10 through driver data mining, Oberon was never found there. Only reason why people think Oberon is Navi 10 is because it runs regression test on Ariel testlist specifications.

View attachment 3642

But entire point of regression test is testing functionality (native/bc1/bc2) of previous version of chip (Ariel), therefore RT/VRS would never be shown there, because Ariel (ie Navi 10 Lite) doesn't have one.

Its very very simple to grasp and entire logic cannot be explained any other way.

Ariel - Navi 10 Lite chip, RDNA1, perhaps used by Sony to get BC working on RDNA architecture
OBR - RDNA2 chip, regression tested against Ariel testlist in Github leak

https://twitter.com/x/status/1153157871963521030

 

[PSman1700]

Here a informative post over at resetera, they are using your posts to highlight.

https://www.resetera.com/threads/ps...stors-running-this-again.173318/post-29667625

 

[chris1515]

Here a informative post over at resetera, they are using your posts to highlight.

https://www.resetera.com/threads/ps...stors-running-this-again.173318/post-29667625

In the full Github there is no sign at all of RT and VRS for Oberon. Saying it is RDNA2 is adding information to it. There is no proof at all.