Collective communication operations such as broadcast operations and gather operations are commonly used as fundamental communication mechanisms in many applications: the broadcast operation transmits a message from a single processor to all the other processors, and the gather operation collects messages from all the processors to a single processor. In many cases, however, only point-to-point communication is supported by the system and the collective communication operations must be implemented as a collection of point-to-point messages. Since the efficiency of the collective communication operations affects the overall performance of the applications, it is very important to present efficient implementation of these operations. Efficient collective communication operations in homogeneous parallel computing environments have been proposed. However, it is pointed out that they may become inefficient in heterogeneous parallel computing environments. So, we design the efficient collective communication operations for heterogeneous parallel computing environments.