Add hybrid Neon/Neon Keccak-f1600, don't use scalar ASM on Cortex-A72

pq-code-package / mlkem-c-aarch64

ML-KEM implementation optimized for aarch64

Apache License 2.0

9 stars 6 forks source link

Closed hanno-becker closed 1 week ago

hanno-becker commented 1 week ago

Benchmarks suggest that something isn't yet working as intended. Both Gv2 and Gv3 get slower.