Originally Posted by ChevChelios
RX 490 probably cut down Vega
or maybe 485 - cut down Vega and 490 - full Vega
Vega is HBM memory bus *(HBM and HBM2 use same bus) and cannot be "cut down" because the data bus uses more than 80% of the pin interconnects on the chip. You can't "cut down" the HBM system to a 3092 or 2048... they'd have to be built that way to start with. There's a large amount of public ignorance about HBM and memory systems in general in graphics cards. The HBM system is a rigid solution with a buffer system between the ram and the GPU's processing.
GDDR5 and GQDR5 are both "DDR5" and can be operated by the GDDR5 controller. One transmits Double and one transmits Quadruple. Both are on 192/256/384/512 scaleable buses.
We may see a 512bit "Vegalaris" downclock with GDDR5 and then a full power "Vegalaris" with GQDR5 memory spanking nVidia's ti in the face. But Vega was specifically named and scheduled as an HBM/2 product.
We're CERTAINLY not at the upper end of the Polaris architecture with the rinky-dink little 2300 shader RX-480. Wouldn't surprise me if we see a 2800 and 3600 shader "Polaris 9" with 512bit still within the thermal envelope of a 2-slot card.
2016 is far from over but I can guarantee you there will be no "cut down" HBM/2 gpu made. There may be bank-disables like Fury Nano/Nitro/FPD but those are full-sized chips with bad shader banks etched out.
What we will see is AMD just sitting back to see what nVidia does for their big-end harpoon-cards. Then out-build them. No matter how awesome the Pascal architecture seems, its a junky 1970's car compared to the 14nm that AMD is working with. (its even starting to appear that the "16nm" of Pascal is a misnomer)Edited by prjindigo - 6/15/16 at 3:52am