[vectorIntrinsics+mask] RFR: 8273371: [Vector API] Crash in Assembler::evpermd when code-gen for masked rearrange operation on AVX512

[Jie Fu]

On Tue, 7 Sep 2021 04:51:40 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>> Hi all,
>> 
>> I'd like to fix the crash in Assembler::evpermd when code-gen for masked int/float rearrange operation on AVX512.
>> This is because evpermd can be only used for vector-size-in-bits > 128 [1].
>> The fix is disabling the code-gen for int/float if vector-size-in-bits < 256.
>> 
>> Thanks.
>> Best regards,
>> Jie
>> 
>> [1] https://www.felixcloutier.com/x86/vpermd:vpermw
>
> LGTM

Thanks @jatin-bhateja

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

PR: https://git.openjdk.java.net/panama-vector/pull/121