It's obviously less dramatic than running across a dozen cores at once, but per-core performance has continued to improve over time