[vectorIntrinsics+fp16] RFR: 8295992: Add support for iotaShuffle and addIndex APIs for Halffloat vector type. [v2]

[Jatin Bhateja]

On Thu, 3 Nov 2022 17:39:15 GMT, Swati Sharma <duke at openjdk.org> wrote:

>> Hi All,
>> 
>> The patch contains the below changes:
>> - Java vector API changes to support iotaShuffle and addIndex operations for Halffloat vectors.
>> - Re enable generation of test points for these APIs in vector API test automation script.
>> 
>> Please review and provide your feedback.
>> 
>> Thanks,
>> Swati
>
> Swati Sharma has updated the pull request incrementally with one additional commit since the last revision:
> 
>   8295992: Resolved comments.

src/jdk.incubator.vector/share/classes/jdk/incubator/vector/X-Vector.java.template line 3384:

> 3382:             case VECTOR_OP_MIN: return (v, m) ->
> 3383:                     toBits(v.rOp(MAX_OR_INF, m, (i, a, b) -> Halffloat.valueOf(Math.min(Halffloat.valueOf(a).floatValue(), Halffloat.valueOf(b).floatValue()))));
> 3384:             case VECTOR_OP_MAX: return (v, m) ->

TODO item unrelated to this patch:
We need to optimize fall back implementation to directly use newly added Float.float16ToFloat API,  currently Halffloat.valueOf(a).floatValue() is introducing redundancies by first creating a Halffloat instance followed by extracting a floating point value,  'a' is already a short value encoding of a FP16 value.

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

PR: https://git.openjdk.org/panama-vector/pull/207