Changes between Version 5 and Version 6 of experiments


Ignore:
Timestamp:
10/01/08 09:50:04 (11 years ago)
Author:
daley
Comment:

email from klaus

Legend:

Unmodified
Added
Removed
Modified
  • experiments

    v5 v6  
    3939from the runs, that should give you more memory to play around in. 
    4040 
     41However, it seems for WD_def we do not vary lrefine_max.   
     42Chris needs to speak to his work colleague to understand this  
     43more thoroughly.  For the time being here is the email  
     44describing the weak scaling for WD_Def: 
     45{{{ 
     46On Wed, 1 Oct 2008, Chris Daley wrote: 
     47 
     48> Also can you please explain why we change: 
     49> r_match 
     50> refine_uni_radius 
     51> in each flash.par? 
     52 
     53THAT is how the size of the problem (i.e., muber of blocks) was 
     54changed.  Note the lrefine_min and lrefine_max do not change! 
     55 
     56> How do you know the correct value? 
     57 
     58It required quite a bit of work to come up with these values. 
     59 
     60They were chosen to give problem sizes of (roughly) 100 * 64 * 2^n 
     61total blocks, for integer n, n is the number in scaling<n> I believe. 
     62 
     63Klaus  
     64}}}