RFR: 8338967: Improve performance for MemorySegment::fill [v10]

[Maurizio Cimadamore]

On Mon, 2 Sep 2024 08:53:57 GMT, Per Minborg <pminborg at openjdk.org> wrote:

> Here is the performance improvement for Linux a64. Looks like a significant improvement.

This looks good. Again, the goal of this PR is not to squeeze every nanosecond out - but, rather, to achieve a performance model that is "sensible" - whereby, if you work on smallish segments you get more or less the same degree of performance you get when operating on small arrays. These changes seem to achieve that goal.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/20712#issuecomment-2324280589