I'm guessing the reason AVX512 took so long is that doubling the size increases complexity by a lot. This doesn't introduce wider vectors, so I don't see why it

I'm guessing the reason AVX512 took so long is that doubling the size increases complexity by a lot. This doesn't introduce wider vectors, so I don't see why it would take so long. But who knows if AMD sees value in these new instructions/if there's market demand for them.