[vectorIntrinsics] RFR: Use Integer/Long compress and expand. [v2]

[Sandhya Viswanathan]

On Mon, 18 Apr 2022 21:50:27 GMT, Paul Sandoz <psandoz at openjdk.org> wrote:

>> Use the Integer/Long compress and expand implementations.
>> 
>> Fix compilation of perf tests for constant shift operations (brought in via merge that does not have perf tests).
>
> Paul Sandoz has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Remove debug print

I have only two minor comments. Rest of the patch looks good to me.

test/jdk/jdk/incubator/vector/templates/Perf-Scalar-header.template line 44:

> 42: 
> 43:     private static final $type$ CONST_SHIFT = $Boxtype$.SIZE / 2;
> 44: 

This is only needed for integral types and not for float/double.

test/jdk/jdk/incubator/vector/templates/Perf-header.template line 81:

> 79: 
> 80:     private static final $type$ CONST_SHIFT = $Boxtype$.SIZE / 2;
> 81: 

This is only needed for integral types and not for float/double.

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

PR: https://git.openjdk.java.net/panama-vector/pull/194