High-Throughput Sequence Translation Using CUDA