The results show that the hardware compute engines can achieve similar performance in arithmetic precision by using a combination of: (i) fixed-point arithmetic, (ii) function table lookup, and (iii) function interpolation, compared to computation that directly uses double precision floating point.Molecular dynamics is an approach that uses classical mechanics to model the behavior of a molecular system using the Newtonian equations of motion. A biomolecular simulation using software could spend up to 99% of the total computation time in calculating the non-bonded interactions between particles, which is a significant bottleneck in biomolecular simulations.The primary motivations for this research are: (i) special-purpose computers for MD simulation have become an interesting application, and (ii) FPGA technology is becoming a viable alternative to ASIC technology. The objective of this thesis is to design two types of FPGA-based compute engines for computing the non-bonded interactions: (i) Ewald Direct Space, and (ii) Lennard-Jones.