Parallel GC and array object layout: way off the base and laid out in reverse?

Aleksey Shipilev aleksey.shipilev at oracle.com
Wed Sep 25 06:27:12 PDT 2013


Hi Jon,

On 09/21/2013 03:37 AM, Jon Masamitsu wrote:
> On 9/20/2013 12:59 PM, Aleksey Shipilev wrote:
>> On 09/17/2013 03:17 PM, Aleksey Shipilev wrote:
>>> Ok, here is the current fix:
>>>    http://cr.openjdk.java.net/~shade/8024394/webrev.00/
>>>
>>> I haven't tested the performance on large scale yet, because wanted to
>>> cross-check if this is a viable solution code-wise.
>> This fix brings statistically insignificant differences on
>>    {SPECjbb2005, SPECjbb2013, SPECjvm2008, SPECjvm98} x
>>      x {Linux x86/x86_64,  Solaris x86/x86_64/SPARCv9}
> 
> Were you comparing GC pause times for those benchmarks?

I was comparing the benchmark scores, and the scores are within the
measurement error. I peeked into the GC logs for SPECjbb2005, and the GC
time was the same.

-Aleksey.


More information about the hotspot-gc-dev mailing list