GC benchmarks

Tony Printezis Antonios.Printezis at sun.com
Mon Jun 8 17:05:54 UTC 2009


Hi all,

The reality these days is that, with a bit of effort in tuning the GC, 
GC overhead in applications is really very low (single digit percentage, 
sometimes even as low as 1% or 2%). The actual overhead / pause times / 
etc. are very application dependent. So, if you come up with a say 
synthetic benchmark that does mostly GC, I don't know whether you'll 
learn anything by comparing how our GCs perform on it. We have a few 
such benchmarks, but they are mainly used for stress testing, not 
performance testing.

Tony

Paul Hohensee wrote:
> Actually, specjbb2005 stresses gc more than jbb2000.  The latter can 
> be gamed to
> avoid gc entirely during the timed intervals of the run, whereas the 
> former cannot.
> Also, the timed intervals of the run in jbb2005 are 4 minutes long 
> compared with
> 2 minutes in jbb2000, which pretty much guarantees at least a young 
> gen collection
> during a timed interval.
>
> Paul
>
> Dan Hicks wrote:
>> The old SPECjbb2000 benchmark was pretty much a pure test of GC 
>> peformance (with a little bit of pure CPU overhead thrown in).  The 
>> (current) SPECjbb2005 benchmark was modified to throw in more system 
>> complexity (though I don't recall the details) and is less of a GC 
>> benchmark.
>>
>> http://www.spec.org/jbb2000/results/
>>
>> Unfortunately, the results aren't real current, and aren't organized 
>> in a way to allow you to compare GC algorithms very readily.
>>>
>>> Message: 1
>>> Date: Tue, 2 Jun 2009 14:01:57 -0400
>>> From: Clemens Eisserer <linuxhippy at gmail.com>
>>> Subject: GC benchmarks
>>> To: hotspot-gc-dev at openjdk.java.net
>>> Message-ID:
>>>     <194f62550906021101j4a3dacfbx3219a344bb91ed at mail.gmail.com>
>>> Content-Type: text/plain; charset=UTF-8
>>>
>>> Hi,
>>>
>>> Just for fun, does anybody know some benchmarks comparing the
>>> different GCs available with some real-world load?
>>> I would be interested in results for small servers (4-16P), like
>>> memory overhead, pause times, throughput impact.
>>>
>>> I know basically how the different GCs work, however I hadn't much
>>> luck finding hard numbers ;)
>>>
>>> Thank you in advance, Clemens
>>>
>>>
>>> End of hotspot-gc-dev Digest, Vol 24, Issue 1
>>> *********************************************
>>>
>>>   
>>

-- 
---------------------------------------------------------------------
| Tony Printezis, Staff Engineer   | Sun Microsystems Inc.          |
|                                  | MS UBUR02-311                  |
| e-mail: tony.printezis at sun.com   | 35 Network Drive               |
| office: +1 781 442 0998 (x20998) | Burlington, MA 01803-2756, USA |
---------------------------------------------------------------------
e-mail client: Thunderbird (Linux)





More information about the hotspot-gc-dev mailing list