Thanks for sharing both links. Seeing Amdahl’s original framing alongside Gustafson’s optimistic reinterpretation really helps clarify where the architectural limits start to dominate once workloads scale.
Nice write-up. I’ve always felt that Amdahl’s Law gets oversimplified in software discussions.
The real bottlenecks tend to be architectural rather than purely parallelizable tasks.