commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Maurizio Cucchiara <mcucchi...@apache.org>
Subject Re: [OGNL] Performance analysis
Date Sat, 22 Oct 2011 16:28:54 GMT
This is the result of the performance comparison:
Apache Commons OGNL http://s.apache.org/H1u
Legacy OGNL implementation http://s.apache.org/ih

There is a huge difference. Furthermore, looks like I need to refine
declarative method cache.

Twitter     :http://www.twitter.com/m_cucchiara
G+          :https://plus.google.com/107903711540963855921
Linkedin    :http://www.linkedin.com/in/mauriziocucchiara

Maurizio Cucchiara



On 22 October 2011 18:23, Maurizio Cucchiara <mcucchiara@apache.org> wrote:
> For the record, starting from now, OGNL Runtime has a new setCacheFactory
> method which allows the user to choose his preferred implementation.
>
>
> Twitter     :http://www.twitter.com/m_cucchiara
> G+          :https://plus.google.com/107903711540963855921
> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>
> Maurizio Cucchiara
>
>
> On 22 October 2011 12:27, Maurizio Cucchiara <mcucchiara@apache.org> wrote:
>>
>> Apropos the idea of the maven profile is very good.
>> +1 and thank you.
>>
>> Twitter     :http://www.twitter.com/m_cucchiara
>> G+          :https://plus.google.com/107903711540963855921
>> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>>
>> Maurizio Cucchiara
>>
>>
>> On 22 October 2011 12:24, Maurizio Cucchiara <mcucchiara@apache.org>
>> wrote:
>>>
>>> Sure you can, before that I should merge new branch with the current
>>> trunk.
>>> Further, FYI I have just submitted a patch for a small
>>> improvement http://issues.carrot2.org/browse/JUNITBENCH-40
>>> Twitter     :http://www.twitter.com/m_cucchiara
>>> G+          :https://plus.google.com/107903711540963855921
>>> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>>>
>>> Maurizio Cucchiara
>>>
>>>
>>> On 22 October 2011 12:17, Simone Tripodi <simonetripodi@apache.org>
>>> wrote:
>>>>
>>>> Hi Mau,
>>>> that for the explanation, that really helps on clarifying the graph!!!
>>>>
>>>> I have an idea about merging the tests in trunk with a profile
>>>> approach that I already submitted for the Disruptor project[1] - they
>>>> have performance/benchmark tests too - if it is fine for you I can
>>>> work on it -  not today that's Rugby day, maybe tomorrow :)
>>>>
>>>> All the best and have a nice WE,
>>>> Simo
>>>>
>>>> [1] http://code.google.com/p/disruptor/issues/detail?id=2
>>>>
>>>> http://people.apache.org/~simonetripodi/
>>>> http://simonetripodi.livejournal.com/
>>>> http://twitter.com/simonetripodi
>>>> http://www.99soft.org/
>>>>
>>>>
>>>>
>>>> On Sat, Oct 22, 2011 at 10:00 AM, Maurizio Cucchiara
>>>> <mcucchiara@apache.org> wrote:
>>>> > I almost forgot there are even old vs commons comparison. Stay tuned
>>>> > :)
>>>> >
>>>> > Twitter     :http://www.twitter.com/m_cucchiara
>>>> > G+          :https://plus.google.com/107903711540963855921
>>>> > Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>>>> >
>>>> > Maurizio Cucchiara
>>>> >
>>>> >
>>>> >
>>>> > On 22 October 2011 09:58, Maurizio Cucchiara <mcucchiara@apache.org>
>>>> > wrote:
>>>> >> Thank you Simo,
>>>> >> Ok, I realized after that the graph needs at least a short
>>>> >> explanation:
>>>> >> In that test I have tried to test every caches which I re-engineered.
>>>> >> In the graph you mentioned, there are depicted 3 different kind
of
>>>> >> cache implementations:
>>>> >> 1. Concurrent HashMap (CHM)
>>>> >> 2. Thread-safe HashMap (HM)
>>>> >> 3. Reentrant Read Write Lock (RRWL)
>>>> >>
>>>> >> As you will notice, there are some cases where RRWL is faster, other
>>>> >> where CHM and so on.
>>>> >> So there is no yet a real winner approach, though, at very quick
>>>> >> look,
>>>> >> CHM seems to be the slower one (my guess is that it's the only
>>>> >> not-fully thread safe).
>>>> >>
>>>> >> To answer your question about projects merging, they surely could
>>>> >> work
>>>> >> under the same project, my only concern is about the execution time.
>>>> >> At the moment I'm writing the whole execution time is near 50 seconds
>>>> >> (currently the non-benchmark side take more or less 20 seconds).
>>>> >> Furthermore the benchmark tests don't give you an answer in term
of
>>>> >> correctness (aside from the concurrency issues),  ATM the only
>>>> >> motivation behind them is performance measurement.
>>>> >> Next time I could check the hit/miss ratio.
>>>> >> Twitter     :http://www.twitter.com/m_cucchiara
>>>> >> G+          :https://plus.google.com/107903711540963855921
>>>> >> Linkedin    :http://www.linkedin.com/in/mauriziocucchiara
>>>> >>
>>>> >> Maurizio Cucchiara
>>>> >>
>>>> >
>>>>
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
>>>> For additional commands, e-mail: dev-help@commons.apache.org
>>>>
>>>
>>
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message