Sequential and Parallel Performance IV

• We can use the same idea to improve the parallel algorithm

• Now the top P=logNproc digits label processor number and we transfer these to digits stored inside processor. This involves communication but after this transfer, all computations are internal to a processor

  • Total Communication time is 2(d-P). (1-2L) . Tcomm

Previous slide

Next slide

Back to first slide

View graphic version