RFR: 8301958: Reduce Arrays.copyOf/-Range overheads [v7]

[Claes Redestad]

On Wed, 8 Feb 2023 01:10:59 GMT, David Schlosnagle <duke at openjdk.org> wrote:

>> Claes Redestad has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Minimize, force inline, generalize
>
> src/java.base/share/classes/java/util/Arrays.java line 3594:
> 
>> 3592:     public static int[] copyOf(int[] original, int newLength) {
>> 3593:         if (newLength == original.length) {
>> 3594:             return original.clone();
> 
> I am curious about the use of `clone` for some primitive array types (`short[]`, `int[]`, `long[]`, `char[]`, `float[]`) and `copyOf` using `System.arraycopy` in other types (`byte[]`, `double[]`). Do these types optimize differently or hit different intrinsics depending on primitive type? Is there difference in array zeroing?
> 
> From a quick [JMH benchmark](https://gist.github.com/schlosna/975e26965ec822ad42034b3ea2b08676) `System.arraycopy` seems slightly better.

I went back and forth on this and also saw a small win using `arraycopy`, but the PR ended up in an inconsistent state with some using one and some using the other. While this discrepancy seem like something we should treat as a bug, I've arranged to use `copyOf` helper consistently for now.

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

PR: https://git.openjdk.org/jdk/pull/12453