In this work we present the algorithm behind the updated version of the N-body dynamics code Swifter-SyMBA, which we have dubbed “Swiftest”. This code was developed by our group and contains upgrades aimed at improving the performance of Swifter-SyMBA in a shared-memory parallel computing environment. The development of Swiftest was motivated by recent models of martian moon accretion from a circumplanetary debris disk and terrestrial planet accretion. These models generally assume that all collisions between bodies lead to perfect accretion. However, recent work on collisional dynamics and fragmentation has shown that this is not a realistic description of the collisional environment. In agreement with previously published results, Swiftest considers four main types of collisional outcomes: (1) perfect mergers, in which the total mass of the colliding bodies in conserved in a single resultant body, (2) simplified hit and runs, in which the bodies collide at an oblique angle, resulting in an intact target body and either an intact projectile or a fragmented projectile, dependent on the angle and velocity of the collision, (3) disruptions of both the projectile and target bodies resulting in the formation of collisional debris, and (4) super catastrophic disruptions of both the projectile and target bodies resulting in the formation of larger quantities of smaller mass collisional debris than in a disruption. In this work we present the effect of collisional regimes on planetary accretion simulations when compared to simulations that only consider perfect mergers between colliding planetary bodies. We show that the incorporation of collisional regimes in Swiftest, along with the inclusion of fragments generated during imperfect collisional events, presents a more realistic and meaningful environment in which to study planetary accretion than perfect merger simulations.