[jdk17u-dev] RFR: 8302736: Major performance regression in Math.log on aarch64

[Goetz Lindenmaier]

On Thu, 20 Jul 2023 12:47:36 GMT, Andrew Haley <aph at openjdk.org> wrote:

>> I backport this for parity with 17.0.9-oracle.
>> 
>> I resolved runtime.cpp because the comment in the context 
>> differs.
>> I omitted the change to continuationFreezeThaw.cpp.
>> This file is not in 17. It contains loom code, loom is not in 17.
>
>> /integrate
> 
> It seems my comment was not addressed. Perhaps an oversight?

@theRealAph 
Sorry, yes, I oversaw your comment. There is just too much boilerplate stuff on this page...

So you propose to craft a different change for 17?
One that only omits the WXWrite for Math.log?
Or should we just live with the performace issue, after all it is not a regression. 
If so I would back out this.

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

PR Comment: https://git.openjdk.org/jdk17u-dev/pull/1588#issuecomment-1643979840