On Fri, 13 Mar 2026 07:57:12 GMT, Jatin Bhateja <[email protected]> wrote:

>> src/hotspot/cpu/aarch64/aarch64.ad line 8312:
>> 
>>> 8310:   ins_pipe(pipe_slow);
>>> 8311: %}
>>> 8312: 
>> 
>> How is this change related to the Float16 changes?
>
> Hi @eme64 , This change was introduced by @Bhavana-Kilambi to fix AARCH64 
> failure, she was not able to reproduce it outside this patch, it was 
> communicated to you earlier.
> https://github.com/openjdk/jdk/pull/28002#issuecomment-3789507594

Ah ok. That sounds ok.

>> src/jdk.incubator.vector/share/classes/jdk/incubator/vector/ByteVector128.java
>>  line 95:
>> 
>>> 93:     @ForceInline
>>> 94:     @Override
>>> 95:     public final Class<Byte> elementType() { return ETYPE; }
>> 
>> This looks like a refactoring that could be done separately, right?
>
> I will revert this change from the patch in next version, will wait for other 
> comments from @PaulSandoz.

Ok, great!

>> test/jdk/jdk/incubator/vector/ShortVectorMaxTests.java line 1731:
>> 
>>> 1729:         assertEquals(asFloatingSpecies.vectorShape(), 
>>> SPECIES.vectorShape());
>>> 1730:         assertEquals(asFloatingSpecies.length(), SPECIES.length());
>>> 1731:         assertEquals(asFloating.viewAsIntegralLanes().species(), 
>>> SPECIES);
>> 
>> These changes seem unrelated to Float16, right? Can they be done separately?
>
> This is specific to Float16,  Float16Vector returns ShortVector with 
> viewAsIntegralLanes and viceversa for viewAsFloatingPoint.

Ah, ok. Makes sense!

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

PR Review Comment: https://git.openjdk.org/jdk/pull/28002#discussion_r2929922860
PR Review Comment: https://git.openjdk.org/jdk/pull/28002#discussion_r2929923504
PR Review Comment: https://git.openjdk.org/jdk/pull/28002#discussion_r2929924049

Reply via email to