2017.12.10 01:30:51 (939653601638666241) from Daniel J. Bernstein, replying to "Adam Langley (@agl__)" (939632838487375872):
Sure, but you're adding even more roughness by omitting some of the identifying information for the different CPUs involved!
2017.12.09 02:09:35 (939300964472905728) from Daniel J. Bernstein, replying to "Adam Langley (@agl__)" (939299369978798080):
Something else: There's a second KEM, ntrulpr4591761, in the NTRU Prime submission. 58756/94508/128316 Haswell cycles keypair/enc/dec.
2017.12.09 05:46:20 (939355510515548161) from "Adam Langley (@agl__)":
Thanks! (Had to wait to get home before I could update it.)
2017.12.09 23:58:33 (939630375998980097) from Daniel J. Bernstein, replying to "Adam Langley (@agl__)" (939355510515548161):
Yet another thing: there's a big difference in speeds between, e.g., E3-1220 v2 (3.1GHz Ivy Bridge) and E3-1220 v5 (3GHz Skylake).
2017.12.10 00:08:20 (939632838487375872) from "Adam Langley (@agl__)":
Yes, of course. These aren't benchmarks on a common machine, just an extraction of the claims from the PDFs. As noted, there's a lot of variance in machines and in implementation effort & ability, thus speeds are rough.