RFR: 8267968: [PPC64] Use prefixed load and addi instructions for better performance in POWER10 [v3]
Kazunori Ogata
ogatak at openjdk.java.net
Thu Jun 10 07:39:19 UTC 2021
On Wed, 9 Jun 2021 14:07:11 GMT, Martin Doerr <mdoerr at openjdk.org> wrote:
>> Kazunori Ogata has updated the pull request incrementally with two additional commits since the last revision:
>>
>> - Revert changes for pusing nodes in loadConLNodesTuple and add comments about the node _last points to
>> - Fix grammatical errors in comments
>
> Sorry for my late response. I was busy with other things. I've looked at this change for some time and I wonder if such a complex change should be done at all. I like the idea, but does it really improve performance for any real applications or benchmarks? At least those parts which only increase complexity should not get done.
@TheRealMDoerr Thank you for your comment. I agree with it. Honestly, we can't measure performance in reliable manner because we only have development chips/machines, where some of the features are not optimal or disabled for development purpose. So I'll close this pull request. When we get measurement data, I'll revisit this change and check if I can pick up effective parts of this change with reducing complexity.
-------------
PR: https://git.openjdk.java.net/jdk/pull/4267
More information about the hotspot-dev
mailing list