Hi,
On Sat, 18 Oct 2025 at 21:46, KAZAR Ayoub <ma_kazar@esi.dz> wrote:
>
> Hello,
>
> I’ve rebenchmarked the new heuristic patch, We still have the previous improvements ranging from 15% to 30%. For
regressionsi see at maximum 3% or 4% in the worst case, so this is solid.
Thank you so much for doing this! The results look nice, do you think
there are any other benchmarks that might be interesting to try?
> I'm also trying the idea of doing SIMD inside quotes with prefix XOR using carry less multiplication avoiding the
slowpath in all cases even with weird looking input, but it needs to take into consideration the availability of
PCLMULQDQinstruction set with <wmmintrin.h> and here we go, it quickly starts to become dirty OR we can wait for the
decisionto start requiring x86-64-v2 or v3 which has SSE4.2 and AVX2.
I can not quite picture this, would you mind sharing a few examples or patches?
--
Regards,
Nazir Bilal Yavuz
Microsoft
