This paper shows how to securely authenticate messages using just $$29$$ bit operations per authenticated bit, plus a constant overhead per message. The authenticator is a standard type of “universal” hash function providing information-theoretic security; what is new is computing this type of hash function at very high speed. At a lower level, this paper shows how to multiply two elements of a field of size $$2^{128}$$ using just $$9062 \approx 71\,\cdot \, 128$$ bit operations, and how to multiply two elements of a field of size $$2^{256}$$ using just $$22164 \approx 87\,\cdot \, 256$$ bit operations. This performance relies on a new representation of field elements and new FFT-based multiplication techniques. This paper’s constant-time software uses just 1.89 Core 2 cycles per byte to authenticate very long messages. On a Sandy Bridge it takes 1.43 cycles per byte, without using Intel’s PCLMULQDQ polynomial-multiplication hardware. This is much faster than the speed records for constant-time implementations of GHASH without PCLMULQDQ (over 10 cycles/byte), even faster than Intel’s best Sandy Bridge implementation of GHASH with PCLMULQDQ (1.79 cycles/byte), and almost as fast as state-of-the-art 128-bit prime-field MACs using Intel’s integer-multiplication hardware (around 1 cycle/byte).

This paper presents extremely fast algorithms for code-based public-key cryptography, including full protection against timing attacks. For example, at a 2 128 security level, this paper achieves a reciprocal decryption throughput of just 60493 cycles (plus cipher cost etc.) on a single Ivy Bridge core. These algorithms rely on an additive FFT for fast root computation, a transposed additive FFT for fast syndrome computation, and a sorting network to avoid cache-timing attacks.