Even if the poster making the claims has some actual info,why is it news?? Look at almost every generation of graphics card in the last decade?? There have been second generation refinements with a slightly updated feature set or better performance/watt:
1.)GTX400 to GTX500 series
2.)HD5000 to HD6000 series
3.)Kepler 1 to Kepler 2
4.)Maxwell 1(GTX750TI) to Maxwell 2(GTX900 series)
5.)GCN1.2(Fury/Tonga) to Polaris
So,AMD will release a slightly refined Vega on a slightly refined process node - it just seems like a midlife update of what they have now.
Its no different than with CPUs.
Clock for Clock there is next to no difference from Fuji and Vega:
on one hand I call BS on this for the above reason, on the other hand I this could be true and while working features might help performance more of the chip being used will also mean more power, so a clocked and power staved heater of a chip will just have be throttled even more.
Vega is still a modern day R600, Rankine (NV30) and to a lesser degree Fermi (NVC0)
I don't thing a refresh/die srink is going to fix the core problems with core of there current architecture sadly.
Nah,the R600 and NV30 were worse by far - both of them not only were late but debuted on newer process nodes when compared to the 8800GTX and 9700PRO which were made on the previous generation ones. Plus the R600 and NV30 did poorly on the modern APIs of the day,ie, DX9 and DX10,and the R600 had terrible AA performance which also affected the follow up HD3000 series. The NV30 issues were compounded by it also having a dual slot cooler in an era of single slot coolers(which Nvidia themselves mocked in a famous video) and Half Life 2 having issues with the FX series.
At least even with Vega being late,its OKish in modern APIs and is not using a better node than the competition,which is a generation ahead. Its not quite Fermi either since Nvidia still managed to debut the fastest card at launch.
