RFR: 8343685: C2 SuperWord: refactor VPointer with MemPointer [v6]
Christian Hagedorn
chagedorn at openjdk.org
Thu Jan 16 15:48:05 UTC 2025
On Thu, 16 Jan 2025 06:25:43 GMT, Emanuel Peter <epeter at openjdk.org> wrote:
>> src/hotspot/share/opto/vectorization.hpp line 949:
>>
>>> 947:
>>> 948: // Vector element size statistics for loop vectorization with vector masks
>>> 949: class VectorElementSizeStats {
>>
>> Just noticed this when scrolling through the class in my IDE: All methods of this class could be made `const`. You could probably squeeze this in here as well as part of this bigger refactoring - probably not worth a separate RFE.
>
> Turns out it is dead code!
>
> emanuel at emanuel-oracle:/oracle-work/jdk-fork5/open$ grep VectorElementSizeStats src/hotspot/share/ -r
> src/hotspot/share/opto/vectorization.hpp:class VectorElementSizeStats {
> src/hotspot/share/opto/vectorization.hpp: VectorElementSizeStats(Arena* a) : _stats(NEW_ARENA_ARRAY(a, int, 4)) {
> Binary file src/hotspot/share/opto/.vectorization.hpp.swp matches
> emanuel at emanuel-oracle:/oracle-work/jdk-fork5/open$ git log -S VectorElementSizeStats
> commit 96781ba33d4717c8df2b9ba4cd5d66858cf5c2d1
> Author: Pengfei Li <pli at openjdk.org>
> Date: Fri Sep 15 01:02:44 2023 +0000
>
> 8312332: C2: Refactor SWPointer out from SuperWord
>
> Reviewed-by: epeter, kvn
>
> commit a38582e941c0234e76d1dbea60c731c83d2c9977
> Author: Pengfei Li <pli at openjdk.org>
> Date: Thu Jul 13 01:45:52 2023 +0000
>
> 8311691: C2: Remove legacy code related to PostLoopMultiversioning
>
> Reviewed-by: kvn, sviswanathan
>
> commit 741be46138c4a02f1d9661b3acffb533f50ba9cf
> Author: Pengfei Li <pli at openjdk.org>
> Date: Tue Apr 5 23:50:13 2022 +0000
>
> 8183390: Fix and re-enable post loop vectorization
>
> Reviewed-by: roland, thartmann, kvn
>
> The last occurance was removed with https://github.com/openjdk/jdk/pull/14824
>
> I'll just rip it out now.
Even better :-)
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/21926#discussion_r1918781582
More information about the hotspot-compiler-dev
mailing list