Vector Digital Signal Processors (DSPs) have been employed in every generation of wireless communication systems due to their ability to compute complex algorithms, efficiently meeting the growing demand for higher data rates and user capacity. Compute-intensive wireless kernels, such as Beamforming (BF), equalization, and Channel Estimation (CE), require data parallelization to perform multiple matrix multiplications, inversions, and vector decomposition operations in real time. In this paper, we present the software implementation of hybrid beamforming for massive Multiple Input Multiple Output (mMIMO) 5th Generation Cellular Networks (5G) New Radio (NR). Specifically, we implement the linear Zero-Forcing (ZF) algorithm in a Reduced Instruction Set Computer (RISC)-V-based single core processor, computing the precoding matrix for a system composed of M transmit antennas at the eNodeB and K receivers, each with N antennas.Our single-core implementation demonstrates that optimizing the the algorithm with the Cholesky decomposition for computing the precoding matrix results in throughout and latency gains, when comparing the core's scalar and vector processors. The vector processor achieves a speedup up to 107.73 and 108.36 in throughput and latency, respectively, when computing the ZF kernel.