When I aim for parallelism, I seem to fare a lot worse than when I aim for concurrency. (In the parallel case, I hit Amdahl's law hard and my throughput for 1,

When I aim for parallelism, I seem to fare a lot worse than when I aim for concurrency. (In the parallel case, I hit Amdahl's law hard and my throughput for 1, 2, 4 cores goes something like 1, 1.8, 2.4.)