You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tapestry.apache.org by Adam Greene <ag...@professionalopensource.ca> on 2005/09/03 23:00:27 UTC
Hivemind Performance (in a Tapestry context)
Remembers these numbers are basically at an 8:1 ratio (so 8secs under profiler is only 1 second normally).
Base Time (total time across all calls of this particular method, subtracting time for submethods):
javassist.CtBehaviour.setBody -- 14.3sec
java.lang.forName -- 9.24sec
Highest average:
java.util.ResourceBundle -- 0.25sec
Cummulative (taking into account all of it's submethods):
org.apache.hivemind.impl.servicemodel.AbstractServiceModelImpl.constructCoreServiceImplementation -- 32.5secs across 87 calls.
Basically it comes down to half setup and half invocation (getService() takes up 17secs of the 32.5secs).
But all of this is basically setup. Once setup, runtime is basically 0.27sec cumulative, with most of that time spent in "forCurrentThread" calls (ie. ThreadedServiceModel stuff).
Re: Hivemind Performance (in a Tapestry context)
Posted by Adam Greene <ag...@iq-2000.com>.
I agree Howard. It is only 9 seconds on average for startup and then 1
second per call (and as I stated, those calls are not simple ones and my
guess that even most of that second is simple transfer between server and
browser as Shell indicated 0ms for render).
As to being faster than Tapestry 3.0, I think you are right. We use less
OGNL because of the new "tags", ie. service:, state:, etc. It would be
nice to see OGNL 3.0 released as that would speed up the places where OGNL
is used (but that is for a different group to deal with).
----- Original Message -----
From: "Howard Lewis Ship" <hl...@gmail.com>
To: "Tapestry users" <ta...@jakarta.apache.org>
Sent: Saturday, September 03, 2005 8:54 PM
Subject: Re: Hivemind Performance (in a Tapestry context)
I'm not thrilled with the slow startup of Tapestry 4, but it's a case
of "get it right, then get it fast". I'll be looking at ways to speed
up HiveMind for 1.2, then see if we can make use of that in Tapestry
4.1. I am pleased that runtime performance is good and, I suspect,
better than 3.0.
On 9/3/05, Adam Greene <ag...@professionalopensource.ca> wrote:
> Remembers these numbers are basically at an 8:1 ratio (so 8secs under
> profiler is only 1 second normally).
>
> Base Time (total time across all calls of this particular method,
> subtracting time for submethods):
>
> javassist.CtBehaviour.setBody -- 14.3sec
> java.lang.forName -- 9.24sec
>
> Highest average:
>
> java.util.ResourceBundle -- 0.25sec
>
> Cummulative (taking into account all of it's submethods):
>
> org.apache.hivemind.impl.servicemodel.AbstractServiceModelImpl.constructCoreServiceImplementation
> -- 32.5secs across 87 calls.
>
> Basically it comes down to half setup and half invocation (getService()
> takes up 17secs of the 32.5secs).
>
> But all of this is basically setup. Once setup, runtime is basically
> 0.27sec cumulative, with most of that time spent in "forCurrentThread"
> calls (ie. ThreadedServiceModel stuff).
>
--
Howard M. Lewis Ship
Independent J2EE / Open-Source Java Consultant
Creator, Jakarta Tapestry
Creator, Jakarta HiveMind
Professional Tapestry training, mentoring, support
and project work. http://howardlewisship.com
---------------------------------------------------------------------
To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tapestry-user-help@jakarta.apache.org
Re: Hivemind Performance (in a Tapestry context)
Posted by Howard Lewis Ship <hl...@gmail.com>.
I'm not thrilled with the slow startup of Tapestry 4, but it's a case
of "get it right, then get it fast". I'll be looking at ways to speed
up HiveMind for 1.2, then see if we can make use of that in Tapestry
4.1. I am pleased that runtime performance is good and, I suspect,
better than 3.0.
On 9/3/05, Adam Greene <ag...@professionalopensource.ca> wrote:
> Remembers these numbers are basically at an 8:1 ratio (so 8secs under profiler is only 1 second normally).
>
> Base Time (total time across all calls of this particular method, subtracting time for submethods):
>
> javassist.CtBehaviour.setBody -- 14.3sec
> java.lang.forName -- 9.24sec
>
> Highest average:
>
> java.util.ResourceBundle -- 0.25sec
>
> Cummulative (taking into account all of it's submethods):
>
> org.apache.hivemind.impl.servicemodel.AbstractServiceModelImpl.constructCoreServiceImplementation -- 32.5secs across 87 calls.
>
> Basically it comes down to half setup and half invocation (getService() takes up 17secs of the 32.5secs).
>
> But all of this is basically setup. Once setup, runtime is basically 0.27sec cumulative, with most of that time spent in "forCurrentThread" calls (ie. ThreadedServiceModel stuff).
>
--
Howard M. Lewis Ship
Independent J2EE / Open-Source Java Consultant
Creator, Jakarta Tapestry
Creator, Jakarta HiveMind
Professional Tapestry training, mentoring, support
and project work. http://howardlewisship.com
---------------------------------------------------------------------
To unsubscribe, e-mail: tapestry-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tapestry-user-help@jakarta.apache.org