RFR: 8310190: C2 SuperWord: AlignVector is broken, generates misaligned packs

[Gui Cao]

On Mon, 9 Oct 2023 08:10:49 GMT, Fei Gao <fgao at openjdk.org> wrote:

>> @fg1417 Can you please see if this patch fixes the bug, and if the tests run fine on ARM32, and maybe some other platforms you think it is relevant for?
>
>> @fg1417 Can you please see if this patch fixes the bug, and if the tests run fine on ARM32, and maybe some other platforms you think it is relevant for?
> 
> Hi @eme64, the build on ARM32 is broken on mainline currently but can be fixed by https://github.com/openjdk/jdk/pull/15645. Well, if it does make sense to you, I'll do the testing with Voitylov's fix. Suppose there won't be any cross impact.

> @fg1417 @reinrich @zifeihan
> 
> I am getting more confident with the updated version. I reran testing, and the first performance testing results look promising (i.e. no regression).
> 
> Would you mind re-running testing on your platforms? (I ran on x64 and aarch64 without SVE)
> 
> @fg1417 I don't know aarch64 very well. Would you want to create a matching-rule for aarch64 for `VerifyVectorAlignment`, similar to what I did in `x86.ad`? It is not necessary, but could be nice.

Hi, I have performed tier1-3 test on linux-riscv64 using QEMU with -XX:+UseRVV. Still good.

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

PR Comment: https://git.openjdk.org/jdk/pull/14785#issuecomment-1807550648