3 ms·

Hi there. Can you share some pointers to the alternate algorithm that was mentioned? Thanks and congrats to you as well for reducing the calculation time!

• Supranational 7y ago
  The goal was to extract as much parallelism from a single squaring operation as possible using the available hardware resources. We used a Xilinx FPGA. There will be a paper detailing the exact algorithm in the coming weeks. In the interim, if you are interested in digging into code, you can find some of the primitives for the design here: https://github.com/supranational/primitives https://github.com/supranational/primitives