Soldato
And without sounding argumentative or obtuse, why? There's no benefit other than MOAR COREZ for the hell of it, and muddying the waters of AMD's versatile modular design structure. How much of a product segmentation ballache are we going to get if a CPU-laden IO die doesn't have all 8 working cores? Are we going down Intel's bullship route of +8 and +6 SKUs? Are the +8 SKUs restricted to the 7900X and 7950X? Surely the gaming-focussed 7800X would benefit from a +8 setup too, or is that getting jipped with a +6? What about the thermals?
It's just not needed.
I would highly, highly doubt that. How does it cost less to port previous gen cores to a new node, integrate them into the IO die, adding L3 cache (thus making it a monolithic CPU) and then messing about with Microsoft to get the Windows scheduler updated, rather than adding some new traces to the Raphael package and slapping a 3rd pre-existing CPU chiplet on?
Don't forget the IO die is TSMC 6nm now so it's nowhere near as big as the old GloFlo slab o silicon, and will be packing some RDNA CUs now. I can think of better things to glue onto an IO die than previous-generation CPU cores.
Last edited:
