Subject: Re: [modeller_usage] Making many result model
From: Modeller Caretaker <>
Date: Wed, 16 May 2007 09:37:52 -0700
Cc:
서주현 wrote:
> It seems that modeller 9.1 takes more time to make one model which takes
> about 2 minutes additionally compared to modeller 8.2.
If I understand you correctly, you are saying that 9v1 is slower than
8v2. This depends on your platform, but on i386 9v1 is certainly a
little slower than 8v2 (maybe 10%) because 9v1 is dynamically linked
into a .so (8v2 is statically linked directly into the executable) and
on i386 at least you pay a performance penalty for this.
> I have 2 CPU in one main board. Threrfore, I think that my computer can
> tolerate two running script without the loss of computing speed.
>
> So if I want to make two models and run modeller with two running
> scripts like,
>
> "starting_model = 1
> ending_model = 1"
>
> and
>
> "starting_model = 2
> endin_model = 2"
>
> then, does modeller take half time and result the same models compared
> to time and result from the running script of
>
> "starting_model = 1
> ending_model = 2"
> ?
Yes, it will. However, both of your models will be the same by default,
since you'd be using the same random seed for each. So you would need to
specify different random seeds in each script when you create your
environ object.
Ben Webb, Modeller Caretaker
--
http://www.salilab.org/modeller/
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage