If they announce X3D at the same time as Zen6 is announced, with only catch that it will take say 2 months longer for release, there won't be Osborne effect - not saying nothing will make people disappointed as there is no way normal Zen6 will be much quicker than 9800X3D, and anybody on 7800X3D would certainly want X3D model.
In general, that's probably a fare statement with respect to 1T. Here's the issue, that's going to be HIGHLY application dependent. There are some apps that will happily live in 48MB of L3 that weren't happy at all in 32MB that will do surprisingly well on Zen6. I expect that Zen6x3d will have the same characteristics that Zen5x3d had, in that accessing the 3d cache took a few cycles more than the first 32MB. On gaming benchmarks where people are splitting hairs over the difference between 148fps and 152fps, that latency difference will be evident.
Then you move on to the games that didn't fit into the 128MB 3d cache neatly, but do better with 144MB like Zen6 has, then move on to the games that have such ugly memory access patterns that the 3d cache just doesn't help much, then the better IoD link and likely improved performance of the memory controllers will help.
And what of situations where 8 cores becomes a limiting factor? Having 12 cores in the same segment will improve things, especially for games that like to do a lot of background shader work.
There are lots of reasons why a full single CCD 12 core CCD 3d cache chip might be notably faster than the 9850x3d, and I haven't even touched on the potentially higher core clocks and improved IPC of the cores themselves that are hotly debated here.
Facebook
Twitter