Vector operations like AVX512 will not magically make common software faster. The number of applications that deal with regular operations on large blocks of data is pretty much limited to graphical applications, neural networks and bulk cryptographic operations. Even audio processing doesn't benefit that much from vector operations because a codec's variable-size packets do not allow for efficient vectorization (the main exception being multi-channel effects processing as used in DAW).
Thanks for the correction. I hadn't considered bulk memory operations to be part of SIMD operation but it makes sense -- they operate on a larger grain than word-size so they can do the same operation with less micro-ops overhead.
