Request for review: JDK-8009928 PSR:PERF Increase default string table size

[David Holmes]

On 15/04/2013 11:32 PM, harold seigel wrote:
> Hi David,
>
> I updated the webrev to use LP64_ONLY and NOT_LP64.  Please see
> http://cr.openjdk.java.net/~hseigel/bug_8009928_2/
> <http://cr.openjdk.java.net/%7Ehseigel/bug_8009928_2/>

Thanks - looks good!

>  >>I'm not clear why you needed to add the minimumStringTableSize - is
> 1009 really a hard minimum on the table size?
>
> The existing code uses defaultStringTableSize as the minimum StringTable
> table size.  Since defaultStringTableSize is now 60013 on LP64, I needed
> a new variable to preserve the 1009 minimum.

Ah I see. It isn't clear if the old default truly is an absolute 
minimum, but the logic was forcing it to act that way.

Thanks,
David

> Thanks!  Harold
>
>
> On 4/14/2013 10:26 PM, David Holmes wrote:
>> Hi Harold,
>>
>> On 12/04/2013 10:49 PM, harold seigel wrote:
>>> Hi,
>>>
>>> Please review the following fix for bug: JDK-8009928 PSR:PERF Increase
>>> default string table size
>>>
>>> Summary:   To improve performance, increase the default size of the
>>> StringTable to 60013 for 64-bit platforms.  This will increase the
>>> default size of the StringTable by about 944K.  Please see the
>>> discussion in the bug report about why this change is need.
>>>
>>> Open webrev at http://cr.openjdk.java.net/~hseigel/bug_8009928/
>>> <http://cr.openjdk.java.net/%7Ehseigel/bug_8009928/>
>>
>> You can do this more succinctly using the LP64_ONLY and NOT_LP64 macros.
>>
>> I'm not clear why you needed to add the minimumStringTableSize - is
>> 1009 really a hard minimum on the table size?
>>
>> Thanks,
>> David
>>
>>> Bug link at http://bugs.sun.com/view_bug.do?bug_id=8009928
>>>
>>> The change was tested by hand to ensure that the right string table
>>> sizes were being set for both 32 and 64 bit platforms. Regression
>>> testing was done using JCK Lang and VM, JTREG tests, and ute
>>> vm.quick.testlist tests.
>>>
>>> Thank you!
>>> Harold
>