In this paper we evaluate FPGA technology for use in frequency-domain sonar beamforming. We compare our results with those obtained using Analog Devices' SHARC DSPs to highlight a number of FPGA features which make them attractive for signal processing. First, the CORDIC algorithm maps well onto FPGAs; in our case, this allows complex arithmetic to be done in either rectangular or polar forms and, thus, greatly reduces the hardware requirements for the algorithm. Second, the ability of FPGAs to support multiple memory ports eliminates memory bottlenecks and allows multiple processing elements to be placed into each FPGA, increasing performance via parallelism. Finally, custom computational units and pipelining enable the FPGAs to complete an inner loop computation every cycle while DSPs require multiple cycles per inner loop. Based on these comparisons and considering their 8 to 16 times performance advantage over the SHARC DSP for this algorithm, FPGAs appear to be a reasonable alternative to DSPs in this case.