+--------------------------------+----------------------------------------+ | big | small | +--------------------------------+----------------------------------------+ | + | * | * | + | +----------------+---------------+------------------+---------------------+ | | | | | | 0.0592733sec | 1.38934sec | 0.000124173sec | 0.000160601sec | parallel +----------------+---------------+------------------+---------------------+ | | | | | | 0.0123527sec | 19.3726sec | 1.449e-06sec | 6.15e-07sec | sequential | | | | | +----------------+---------------+------------------+---------------------+ i) There is a difference between the 2 sets of data. With big data, the program perform faster in parallel mod, while with smaller data it perform faster in sequential. The difference comes from the "cost" of opening and closing new threads. If the data is big enough, it pays off, but if it's too small, it's just make the process more complicated. ii) There is a difference between sum operation and multiplication. It is related to the multiplication algorithm, which is n^3 on square matrices, in compare to the linear complexity of the sum operation.