This article is a complete waste of time. It reads like a children's story or a marketing announcement but it's not actually saying anything meaningful or making any technical point beyond just stating "If we use vectors then maybe we don't need speculation" but without providing much evidence except that highly parallel workloads already have enough work to do. Go figure. It mentions in the article we already have GPUs for this. CPUs are famously burdened with workloads that usually aren't GPU workloads. But now there's a declaration of a RVV requirement or something. (I say this as a SIMD programmer who likes a lot about RVV)