Abstract
Current processors typically embeds many cores running at high speed. We may then ask ourselves whether software parallelism is practical for low level arithmetic operations. In this paper we generalize the bipartite modular multiplication method of Kaihara and Takagi. We present a multipartite multiplication. We carefully analyze its asymptotic complexity and measure its practical efficiency and scalability for nowadays multi-core architectures. We present several experimental results which illustrate the efficiency of our method and which bring a positive answer to the above question for a wide range of operand's sizes.