Our benchmark improves the state-of-the-art by 9x in transaction generation time and by 3.4x in memory usage. Along the way, we also design efficient gadgets for variable-base multi-scalar ...