Vectorization and Parallelization of Loops in C/C++ Code

Modern computer processors can support parallel execution of a program by using their multicores. Computers can also support vector operations by using their extended SIMD instructions. To make a computer program run faster, the time-consuming loop computations in the program can often be parallelized and vectorized to utilize the capacity of multicores and extended SIMD instructions. In this paper, the vector multiplication and the matrix multiplication will be used as examples to illustrate how to perform parallelization and vectorization of loops in a C/C++ program when using Microsoft Visual C++ compiler or GNU gcc (g++) compiler. An overview of the Intel