We develop and implement a parallel flatPERM algorithm with mutually interacting parallel flatPERM sequences and use it to sample self-avoiding walks in 2 and 3 dimensions. Our data show that the parallel implementation accelerates the convergence of the flatPERM algorithm. Moreover, increasing the number of interacting flatPERM sequences (rather than running longer simulations) improves the rate of convergence. This suggests that a more efficient implementation of flatPERM will be a massively parallel implementation, rather than long simulations of one, or a few parallel sequences. We also use the algorithm to estimate the growth constant of the self-avoiding walk in two and in three dimensions using simulations over 12 parallel sequences.