Its a secret sauce. Just like everything else sony do. Its so good they dont talk about it.
The information is out there, and even if Sony tried to explain 99 % of gamers would not understand or appreciate it anyway and it would be mis understood.
Maybe you should focus on what XSX is doing, and ask questions like how have Microsoft optimised for 14 CU in extra long shader arrays when everyone else uses less or equal to 10 CU per shader array in every design since dawn of time.
How did Microsoft optimise L1 cache for the bigger 14 CU shader array. Microsoft said no comment on L1 at Hotchips, would not talk about it.
How did Microsoft optimise LDS data store, as it has to store parameters from 14 CU instead of 10 CU, and LDS is normally a nottleneck even with 10 CU.
MS, How does the front end deliver work optimally to 14 CU per array instead of the usual 10 CU ? What is the typical performance hit ?
It seems 6800 XT is converting allot of work to primitive shaders in drivers and hardware, is XSX doing the same optimisations ?
It is not Sony who is missing performance, why do Sony need to explain anything especially when XSX design is very non standard dual purpose Server design and different to everyhing else evermade ?