The weight computation is the most intensive computational part of the space-time adaptive processing (STAP). The traditional sample matrix inverse (SMI) algorithm has a poor numerical stability due to that the dynamic range of the space -time snapshots is great, relatively speaking, the QR decomposition based SMI (QRD-SMI) algorithm has a better numerical characteristic and parallelism, which may suit for this condition. For the purpose of real time implementation of STAP, the performance comparison of STAP with two weight computational algorithms is presented. The test results show that both algorithms have satisfactory performances in resisting clutter and jamming, the traditional SMI algorithm has a better performance in clutter rejection, but QRD-SMI algorithm shows a stronger robustness to the effect of target self-nulling. Through comparing the above two algorithms in algorithm performances, numerical characteristics and parallelism in a comprehensive way, the QRD-SMI algorithm is more suitable for real time implementation of STAP.