Figure 2.

Performance results. A) The speedup increases linearly with the number of used cores. Setting more workers (5) than existing cores (4) does not improve the results. B) The (super linear) speedup exceeds the theoretical maximum of number of processing units due to faster tasks. C) With a conservative average load of 15% for other tasks, the computer is overloaded when R/parallel claims 100%. By reducing the percentage for R/parallel (with an optional argument) we can recover responsiveness and keep working on other tasks while running our calculations.