This paper presents high-performance Advanced Vector Extensions 512 (AVX-512) implementations of two Korean standard cryptographic algorithms: the Lightweight Secure Hash (LSH) function and Lightweight Encryption Algorithm–Galois/Counter Mode (LEA-GCM) authenticated encryption. For LSH, we apply three optimization strategies: single-message processing using AVX-512 512-bit vector registers, dual-message parallel processing through register interleaving, and multi-core parallelization using a dynamic queue-based pthread thread pool. For LEA-GCM, we propose an end-to-end optimization that replaces scalar counter-mode (CTR) encryption with 16-block AVX-512 parallel processing and Streaming SIMD Extensions(SSE)-based Galois Hash (GHASH) authentication with VPCLMULQDQ-based four-block parallel processing. Performance evaluation on an Intel Core i7-1165G7 (Tiger Lake) processor shows that LSH-256 achieves an average of 1.16× throughput improvement and LSH-512 achieves an average of 1.61× improvement over the Korea Internet and Security Agency (KISA) AVX2 reference implementation. Dual-message interleaving achieves an average superlinear speedup of 2.28×, driven by instruction-level parallelism (ILP), and the thread pool delivers speedups of 3.73× to 5.13× across eight logical cores (four physical cores with hyperthreading). The optimized LEA-GCM implementation achieves a 3.26× throughput improvement over the KISA SSE-based reference and a 12.1× improvement over the pure software implementation for 4096-byte inputs, with correctness verified against KISA official test vectors.
Constant-time screening of post-quantum cryptographic implementations must simultaneously examine secret-dependent execution structure, binary instructions, and timing. Because tests observe different phenomena, no single result supports assessment. We present CT-KAT, which generates test harnesses from configurations and connects functional validation, structural analysis across builds, instruction scanning, and timing measurements. CT-KAT records measurement validity, input relationships, and manual-review status, withholding a no-issue status when evidence is insufficient. We evaluate 25 target-operation pairs and 206 build combinations from 24 pairs with build artifacts. Structural-review targets increase from 11 under one deployment build to 13 across the full matrix; instruction scanning expands the review set to 22. Three of 28 timing comparisons show repeatable differences, but all change public and secret values together. In KyberSlash, structural tests show no signal while binary evidence identifies secret-related divisions. Thus, evidence must be preserved with its interpretation conditions.
FAEST is a National Institute of Standards and Technology post-quantum signature candidate based on the Vector Oblivious Linear Evaluation-in-the-Head paradigm, whose signing performance is dominated by repeated Advanced Encryption Standard Counter-based Pseudorandom Generator calls. The reference implementation provides no FAEST-specialized acceleration for Advanced RISC Machine platforms. This paper proposes a three-layer Advanced Reduced Instruction Set Computer Machine NEON Single Instruction Multiple Data optimization: a register-resident 256-byte S-box with Table Lookup/Table Lookup with Extension-based SubBytes and four-way/eight-way parallel Advanced Encryption Standard processing; a fixed-length Pseudorandom Generator specialized for the FAEST tree structure; and Portable Operating System Interface for Unix thread-based parallelization of independent Vector Oblivious Linear Evaluation instances. Evaluated on all 12 parameter sets of FAEST v2 on Raspberry Pi 4 (without Advanced Reduced Instruction Set Computer Machine version 8 crypto-extensions) and Apple M2 (with hardware Advanced Encryption Standard support), the proposed method achieves signing speedups of up to 136.9x on Raspberry Pi 4 and 330.1x on Apple M2 over the pure-C reference. On Raspberry Pi 4, the NEON implementation outperforms OpenSSL; on Apple M2, the NEON-plus-Portable Operating System Interface for Unix thread configuration outperforms hardware-accelerated OpenSSL across all parameters, confirming that NEON SIMD combined with task-level parallelization can exceed hardware-accelerated single-thread throughput on Advanced Reduced Instruction Set Computer Machine-based platforms.