Using Hardware Software Codesign for Optimised Implementations of High-Speed and Defence in Depth CAESAR Finalists

In this work, we present five optimised implementations on a Xilinx-Zynq7200 SoC for the high-speed and defence in depth finalists of the CAESAR competition for finding authenticated encryption ciphers. We eliminated the standard interfaces used during the competition. Through optimised interfaces between hardware and software, we were able to get both performance improvements as well as reduction in used programmable logic. The performance of our implementations is comparable to pure hardware implementations, but our implementations are 50% smaller. Compared to pure SW implementations we are 16 times faster. Comparing the different algorithms, we come to the conclusion that Colm allows the fastest implementation.