This paper sets speed records on well-known Intel chips for the Curve25519 ellipticcurve Diffie-Hellman scheme and the Ed25519 digital signature scheme. In particular, it takes only 159 128 Sandy Bridge cycles or 156 995 Ivy Bridge cycles to compute a Diffie-Hellman shared secret, while the previous records are 194 036 Sandy Bridge cycles or 182 708 Ivy Bridge cycles. There have been many paper...

This paper presents new speed records for AES software, taking advantage of (1) architecture-dependent reduction of instructions used to compute AES and (2) microarchitecture-dependent reduction of cycles used for those instructions. A wide variety of common CPU architectures—amd64, ppc32, sparcv9, and x86—are discussed in detail, along with several specific microarchitectures.