preload

[Casey Dahlin]

Mark wrote:
> 2008/1/23, Jakub 'Livio' Rusinek <jakub.rusinek at gmail.com>:
>   
>> Dnia 23-01-2008, śro o godzinie 18:05 +0100, drago01 pisze:
>>     
>>> Marc Wiriadisastra wrote:
>>>       
>>>> Those people running preload can we get some benchmarks to see whether
>>>> there is a benefit and if so how much of a benefit there is.
>>>>
>>>> I ask this because I got a post on the from Rahul asking whether there
>>>> was any significant benefit running preload. I personally don't have any
>>>> numbers so I'm going to try and start to benchmark the information.
>>>>
>>>>         
>>> Well I tryed it and benchmarked it using bootchart and its even _worse_
>>> then readahead that we decided to disable.
>>>
>>> No preload: 57 sec
>>> With preload: 1m 25sec
>>>
>>> ie.  ~50% slowdown
>>>       
>> Strange... Preload shouldn't slow system down, but make it load
>> faster...
>>     
>
> I'm getting crazy of your reply's..
> And it's right that preload is making things slower. it needs to load
> them in the memory sometime. that time is in the boot progress. if
> your preload "database" is big than you will notice the slowdown in
> the boot but you will also notice the speedup in the applications
> (like firefox).
>
> I personally rather have a slow boot and for the rest fast responding
> applications than just a "slow" and slow applications.
>
> Btw i'm currently busy with benching and that really is irritating...
> nearly all the apps that i want to bench can't be benched with "time
> appname"!
>
> O well.. expect some results soon
>
>   

I just blogged a partial solution to your benching problem:

http://screwyouenterpriseedition.blogspot.com/2008/01/timing-application-startup.html

Tell me what you think. (Code format is due to blogger lacking some 
things I'd rather it had).

--CJD