RFR: 8205051: Poor Performance with UseNUMA when cpu and memory nodes are misaligned [v5]

[David Holmes]

On Thu, 2 Jan 2025 19:32:39 GMT, Derek White <drwhite at openjdk.org> wrote:

>> Swati Sharma has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   8205051: Minor comment change
>
> Hi Alan, I think the general practice is if the command line asked for something (UseAVX=3, -Xdebug, etc), and the JVM can't do it, then print some kind of warning. 
> 
> The UseNUMA code didn't print a warning previously if UseNUMA was disabled by the JVM, but curiously _did_ print a warning if the JVM disabled UseAdaptiveSizePolicy, due to UseNUMA and UseLargePages being enabled. I think the new code is an improvement, but we're happy to update if there's consensus.
> 
> Additionally @jaikiran asked in [JDK-8346834](https://bugs.openjdk.org/browse/JDK-8346834) about the specific format of the warning - old warning or unified logging warning, which is probably a bigger question.
> 
> Sorry for the unexpected failures!

@dwhite-intel  The definition of `UseNUMA` is:

  product(bool, UseNUMA, false,                                             \
          "Use NUMA if available")  

Therefore it is wrong to give a warning if you set the flag but NUMA is not available!  IMO this part of the change should be reverted as it not only affects tests but may affect end users as well.

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

PR Comment: https://git.openjdk.org/jdk/pull/22395#issuecomment-2571497767