Trumpstyle said:
Guys this doesn't disprove github, there's nothing in there about the gpu being Navi 10, this is just a random assumption some people have. Same as either being rdna1 or rdna2.
Amd confirming PS5 is RDNA2 and that the architecture has clock improvements actually make github more possible as 2ghz looked hard even if they were using Tsmc 7nm EUV. They also comfirmed that RDNA2 has ipc gains (Perf/ghz).
As Github brought Oberon back from the dead another idea of mine is back from the dead. That it's possible that no insider actually knows the TF number for PS5, that they or their sources are guessing it based on game performance from the gpu inside the devkit. Based on my previous list, the insiders seems to believe ps5 is at ~11tf. An 9.2TF Rdna2 would only need 20% stronger TFs to be equal to 11TF Rdna1.
It's also possible that Oberon is just BC stuff and the 36CU/2ghz is just PS5 BC boost mode and nothing else or just outdated.
|
Github is undeniably real based on real test. But it's not overall picture. The data was incomplete, the file was messed up, some folder are mixed each other. And then we still don't know when the test was happened .
Even Komachi and Tum Appisak said Oberon and Ariel were built under navi 10 arch . and also RDNA 2 able to do a higher clock but that's also add another theory that they also can make the chip bigger with power efficiency.
