This document summarizes research on the latency performance of encoding with random linear network coding (RLNC). It finds that doubling the generation or symbol size increases latency by approximately 2x for a full vector encoding algorithm. Switching to an on-the-fly encoding algorithm provides more moderate latency gains of 1.41x for generation size 8, 1.62x for size 16, and 1.75x for size 32. Future work could confirm these results on different CPU architectures, further investigate results from 1kB to 1MB data sizes, and evaluate if multi-threading can decrease encoding latency.