Accelerating HQC Key Encapsulation Mechanism with AVX-512

Post-Quantum Cryptography (PQC) aims to secure digital communications against adversaries powered by quantum computers. To be useful, PQC algorithms must be both secure and fast. The Hamming Quasi-Cyclic (HQC) key encapsulation mechanism is a primary code-based alternative to lattice-based standards such as ML-KEM. However, initial benchmarks on Intel processors using AVX2 show that HQC is 9.5x slower than ML-KEM. In this paper, we show that careful analysis and optimization of the implementation can substantially narrow this gap. Using AVX-512 instructions in conjunction with architecture-independent refinements, we significantly accelerated internal HQC operations, including the Reed-Muller decoding step, polynomial multiplications, and the SHA3 hash function. These improvements narrow the performance gap between HQC and ML-KEM to 5.6x. Our contribution brings high-performance implementations of alternative schemes like HQC that are essential for long-term cryptographic agility.