This repository accompanies with the paper Algorithmic Views of Vectorized Polynomial Multipliers -- NTRU Prime at International Conference on Applied Cryptography and Network Security, Part II, 2024. The paper is also available at ePrint 2023/1580.

Authors:

• Vincent Hwang [email protected]
• Chi-Ting Liu [email protected]
• Bo-Yin Yang [email protected]

Overview of This Artifact

Raspberry pi 4

• Cortex-A72.
• 64-bit Armv8.0-A.
• No hardware support for cryptographic operations.

Schemes and Parameters

• NTRU Prime
  • ntrulpr761
  • sntrup761

Our Environment for Benchmarking

• Ubuntu 22.04.1
• GCC 11.2.0

Minimum Requirements (Theoretically)

• A hardware supporting 64-bit Armv8.0-A.
• A C compiler.

Additional Tests for Compilations

• Raspberry pi 4, Uubntu 22.04.1, clang 14.0.0.
• Apple M1, macOS Monterey, clang 13.1.6.
• Apple M1, macOS Monterey, gcc 11.3.0.

Structure of This Artifact

• cycles: Code for accessing cycle counters.
• enable_ccr: Something that should be done prior to accessing cycle counters.
• hash: aes, fips202, sha2.
• randombytes: Randombytes. From system by default, switch to chacha20 for pseudorandom.
• sort: crypto_sort_int32 and crypto_sort_uint32.
• ntrulpr761
• sntrup761

How to Test for Correctness

Go to the folders {scheme}/{implementation}/ and type make where {scheme} is one of the following.

• ntrulpr761
• sntrup761

The following binaries will be produced.

• test
• testvectors

test tests for the key encapsulation mechanisim. OK KEYS means success and ERROR KEYS means that the resulting keys are different. testvectors prints all the bytes computed from crypto_kem_keypair, crypto_kem_enc, and crypto_kem_dec. Implementations targeting the same parameter should have the same testvectors. We compare our implementations with the reference implementations.

How to Benchmark

Go to the folders {scheme}/{implementation}/ and type make speed where {scheme} is one of the following.

• ntrulpr761
• sntrup761 The binary speed will be produced.

Additionally, there are various profiling code as follows (compile with make XXX where XXX is one of the following targets).

• ntrulpr761
  • aarch64_gt_rader_bruun, aarch64_gt, aarch64_gt_schönhage
    • speed_arith
    • speed_encode_decode
    • speed_hash
    • speed_polymul
    • speed_rand
    • speed_sort
• sntrup761
  • aarch64_gt_rader_bruun, aarch64_gt, aarch64_gt_schönhage
    • speed_arith
    • speed_encode_decode
    • speed_hash
    • speed_polymul
    • speed_rand
    • speed_sort
  • ref
    • speed_arith
    • speed_encode_decode
    • speed_polymul
    • speed_sort