On Thu, 10 Apr 2025 13:19:05 GMT, Ferenc Rakoczi <d...@openjdk.org> wrote:
>> By using the aarch64 vector registers the speed of the computation of the
>> ML-KEM algorithms (key generation, encapsulation, decapsulation) can be
>> approximately doubled.
>
> Ferenc Rakoczi has updated the pull request incrementally with two additional
> commits since the last revision:
>
> - Code rearrange, some renaming, fixing comments
> - Changes suggested by Andrew Dinn.
src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp line 5684:
> 5682: VSeq<2> vs5(vs3[1], delta);
> 5683: kyber_montmul16(vs5, vz, vs5, vs_front(vs2), vq);
> 5684: // add results in pairs storing in vs3
Suggestion:
// add results in pairs storing in vs3
// vs3[0] <- montmul(a0, b0) + montmul(montmul(a1, b1), z0);
// vs3[1] <- montmul(a0, b1) + montmul(a1, b0);
src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp line 5686:
> 5684: // add results in pairs storing in vs3
> 5685: vs_addv(vs_front(vs3), __ T8H, vs_even(vs3), vs_odd(vs3));
> 5686: vs_addv(vs_back(vs3), __ T8H, vs_even(vs1), vs_odd(vs1));
Suggestion:
// vs3[2] <- montmul(a2, b2) + montmul(montmul(a3, b3), z1);
// vs3[3] <- montmul(a2, b3) + montmul(a3, b2);
vs_addv(vs_back(vs3), __ T8H, vs_even(vs1), vs_odd(vs1));
src/hotspot/cpu/aarch64/stubGenerator_aarch64.cpp line 5687:
> 5685: vs_addv(vs_front(vs3), __ T8H, vs_even(vs3), vs_odd(vs3));
> 5686: vs_addv(vs_back(vs3), __ T8H, vs_even(vs1), vs_odd(vs1));
> 5687: // montmul result by constant vc and store result in vs1
Suggestion:
// vs1 <- montmul(vs3, montRSquareModQ)
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/23663#discussion_r2044712516
PR Review Comment: https://git.openjdk.org/jdk/pull/23663#discussion_r2044714830
PR Review Comment: https://git.openjdk.org/jdk/pull/23663#discussion_r2044726778
