Abstract-The paper describes software implementation research of CRC computation algorithms. Table-driven and matrix-driven algorithms were presented schematically. Also different implementations of the matrix-driven algorithm such as single-byte; two-byte and four-byte were researched. Graphical results of a computer experiment on supercomputer cluster to determine the speed of CRC32 software implementation were described. It is shown that a high-speed four-byte matrix-driven algorithm should be used in embedded systems and industrial data transmission systems. Research of the matrix-driven algorithms acceleration of relative table-driven shows that even two-bytes matrix-driven algorithm ahead of ~29%, while the four-bytes -by ~54%, which is a significant increasing in speed with respect to the table-driven algorithm.