linear scaling calculation doesn't work

tao liu liut... at gmail.com
Thu Jul 16 07:54:47 CEST 2015


Dear Samuel,

Thanks for your suggestion.

I tried 1152 atoms model to compare the time, but ls_scf is much slower, 
almost 2 times slower than normal OT.

with ls_scf    

#     Step Nr.          Time[fs]        Kin.[a.u.]          
Temp[K]            Pot.[a.u.]        Cons Qty[a.u.]        UsedTime[s]
         0            0.000000        21.889027653      4000.000000000    
-45629.275229507    -45585.497174201         0.000000000
         1            1.000000        21.963579939      4013.623681591    
-45629.348031908    -45585.492189722       580.978587588
         2            2.000000        21.980309366      4016.680816479    
-45629.265005412    -45585.381146897       187.372567314
         3            3.000000        21.919515136      4005.571281372    
-45629.280761605    -45585.440858208       356.206347420
         4            4.000000        21.769580539      3978.172239420    
-45629.152421726    -45585.441282007       349.159549365
         5            5.000000        21.582894296      3944.057203240    
-45628.871926502    -45585.323330731       195.776054914

without ls_scf

#     Step Nr.          Time[fs]        Kin.[a.u.]          
Temp[K]            Pot.[a.u.]        Cons Qty[a.u.]        UsedTime[s]
         0            0.000000        21.889027653      4000.000000000    
-45629.227858504    -45585.449803199         0.000000000
         1            1.000000        21.960057035      4012.979906328    
-45629.354620148    -45585.502305413      1649.856096870
         2            2.000000        21.948354669      4010.841416379    
-45629.357943653    -45585.506069241       129.048230915
         3            3.000000        21.857157417      3994.176034457    
-45629.286968991    -45585.509514837       234.400758075
         4            4.000000        21.705799206      3966.516841323    
-45629.158680009    -45585.511526881       200.970832917
         5            5.000000        21.522476406      3933.016440408    
-45629.001383141    -45585.513562281       108.934935516


Is there anything wrong? Thanks.

Regards,
Tao

On Wednesday, July 15, 2015 at 12:07:37 PM UTC+1, Samuel Andermatt wrote:
>
> Linear scaling sets in at systems at least larger than 1000 atoms.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150715/fc46578a/attachment.html>


More information about the CP2K-user mailing list