[vectorIntrinsics+compress] RFR: 8274837: Intel backend support for compress VectorAPI

[Sandhya Viswanathan]

On Wed, 6 Oct 2021 13:02:53 GMT, Joshua Zhu <jzhu at openjdk.org> wrote:

> I separate my implementation of "compress" API into several patches for easy review.
> This patch is the change in backend used to support vector-to-vector compress API.
> ByteVector and ShortVector are supported in AVX512vbmi2.

I have only one input on match_rule_supported_vector. Also please bring the evcompress_reg instruct into this PR from 145. Rest of the patch looks good to me.

src/hotspot/cpu/x86/x86.ad line 1818:

> 1816:         return false;
> 1817:       }
> 1818:       if (size_in_bits < 512 && !VM_Version::supports_avx512vl()) {

It would be better to add a separate case for Op_CompressV instead of adding it as a fallthrough to gather/scatter.

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

PR: https://git.openjdk.java.net/panama-vector/pull/144