code

Our scheme is currently under submission and constant revision. But you can find some of our code here:

---

C:

Our C implementation can be found here.

The library provides a reference implementation, an optimized implementation, an Intel AVX2 optimized implementation, an Intel AVX-512/GFNI optimized implementation, and an Arm NEON optimized implementation.

AES-NI is used by default, if available, in the optimized versions.

They can be built in this way:

• Reference: build with CMake with the option: -DMAYO_BUILD_TYPE=ref.
• Optimize: build with CMake with the option: -DMAYO_BUILD_TYPE=opt.
• Intel AVX2 Optimize: build with CMake with the option for avx2 op: -DMAYO_BUILD_TYPE=avx2.
• Intel AVX-512/GFNI Optimize: build with CMake with the option: -DMAYO_BUILD_TYPE=gfni. This requires a host CPU supporting AVX-512 and GFNI; where VAES is also present, AES-CTR uses the 512-bit VAES path.
• Arm NEON Optimize: build with CMake with the option: -DMAYO_BUILD_TYPE=neon.

---

Arm Cortex-M4:

Our Arm Cortex-M4 implementation can be found here. It has also been merged into pqm4.

---

Sage:

Our pure sage implementation can be found here.

---

Golang:

Our golang implementation can be found here.

---