The design of a hardware accelerator for supporting CACSD computations is considered in relation toH∞design. The key algorithms are identified and parallel solutions formulated. The speedup of these techniques is then assessed with respect to trial implementations on a network of transputers.