RFR: 8291600: [vectorapi] vector cast op check is not always needed for vector mask cast [v3]

[Xiaohong Gong]

On Wed, 24 Aug 2022 01:45:03 GMT, Jie Fu <jiefu at openjdk.org> wrote:

> The performance improvement is impressive on x86 with avx2.
> 
> Can you clarify what tests did you do? Thanks.

Hi, thanks for looking at this PR! To test the x86 backend rules, I ran ` "test/jdk/jdk/incubator/vector,test/hotspot/jtreg/compiler/vectorapi"` group with `-XX:UseAVX=1, -XX:UseAVX=2, and -XX:UseAVX=3` separately on an AVX-512 machine. For the whole patch testing, I ran `hotspot::hotspot_all, jdk:tier1,tier2,tier3, langtools:tier1` on aarch64 and x86 machines. Thanks!

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

PR: https://git.openjdk.org/jdk/pull/9737