Programmers don't wanne use AVX3, becuase they affraid on one computer it can not work at all and on another will be verry slow (like, as slow as AVX2, becuase on all know processors when AVX is run even if not AVX3 iteself, the processor slowing himself down

to make shure he is not overheating, that is the problem

this Zen 5 is probably the first processor that dosen't slowing down when running AVX/3, and if you wonder about what will hapen on future programs you can be shure that programmers will not wright for only 1 processor that can run AVX3

this is whay Zen 5 is bad design choice (AVX3 or get out), AMD think they can tell programmers which vectorization to use they living in a movie. but I also remember AMD is low budget technology corporation and despite the (HUGE) discounts TSMC gived them, the nodes are limited, I don't think even if they had this "3nm" that TSMC gived to apple would maked mutch of a diffrence, but if they did gived them this node I blive they wouldn't need to reduce low-bit SIMD performance at all, could be AMD try to save money and choosed not to buy "3nm" from TSMC, could be that blueprints were alredy developed for TSMC N4 and then that was before "3nm" was ready (this things costs many money....), for customers I think the best is to decide if they can take advantage of AVX3