As a first demonstrator of the EDRA approach, we developed a DAE hardware accelerator for a computationally intensive task from the bioinformatics domain, namely the RAxML Phylogenetic Analysis application*.
As shown in the figure above, RAxML’s most computationally intensive task (blue square) is offloaded to the hardware accelerator (step 1); the host CPU sends all input data to the FPGA (step 2), and as soon as processing is done, it receives back the results (step 3). Please note that you need to have an AWS account in order to try the EDRA demo.
*A. Stamatakis: “RAxML Version 8: A tool for Phylogenetic Analysis and Post-Analysis of Large Phylogenies”. In Bioinformatics, 2014