Closed jedisct1 closed 5 days ago
Fair enough. LGTM, although shouldn't we have an alternate reference like D23?
I'm not sure. 10 years from now, benchmarks conducted on an AMD Zen 4 CPU are going to feel old and irrelevant.
Even the benchmark from D23 is outdated, because since then, OpenSSL got an AES-GCM implementation specialized for AVX-512.
Ok good point. Happy to leave it, and we can revisit it if someone mentions it. At least some of the referenced papers have benchmarks.
Also remove the reference to the original AEGIS paper as a source for performance measurements.
This paper is getting old and doesn't reflect the performance of AEGIS and AES-GCM on modern CPUs. It also doesn't include parallel variants.
Part of #50