[Dock-fans] Speed-Optimization for your DOCK

Sheng-Hung Wang gentamicin5 at yahoo.com.tw
Mon Oct 31 09:10:05 PST 2005


Hi, Ricardo,

You are right, some optimization options should not be used.
It depended on different source code, hardware, and compiler.
I have tested these options for compiling DOCK_5.1.1.
And the docking energy is similar to -O2 option.
ex.
----------------------------------------------
 Anchors:              2
 Orientations Tried:    225
 Orientations Passed:   218
 Conformations:         50

 Energy Score:          -49.894741
          vdw:          -23.245426
           es:          -26.649315



1 Molecules Processed
Elapsed time:   25.51 seconds
-----------------------------------------------

The RMSD compare with original co-crystall ligand is always smaller
than 3 A.

So, to optimize program in "legal"/acceptable range will still be
beneficial.




--- Ricardo Nicoluci <nicoluci at gmail.com> 

> Hi,
> 
> Did you compare the results that you got ?!
> It is always interesnting optimize the code to the type of processor
> that you have, because you really use the core functions implemented
> on it. But, optimizations like -O3 or -ffast-math gives you results
> faster because it performs a faster calculation without worring about
> the quality of the number.
> Once I did some optimization with the programs of CCP4. We could
> reduce the calculation time by the half without losing the quality,
> but when we tried to to get even more faster, we started to get some
> results that was not compatible witht the expected one.
> 
> Thanks, Ricardo.
> 


___________________________________________________  ×îаæ Yahoo!ÆæĦ¼´•rͨӍ 7.0 beta£¬ÃâÙM¾W·ëŠÔ’ÈÎÄã´ò£¡  http://messenger.yahoo.com.tw/beta.html



More information about the Dock-fans mailing list