Interesting to me is the L3 cache structure. With Zen2, it was 16mb for each group of 4 cores. With Zen3, it is 32mb for each group of 8 cores. Same amount total. But now you can run a bigger size of task without having to go to RAM. So it should help running big PG tasks, and increasing the thread count from 4 to 8 without taking a speed hit.