The ABINIT Users Mailing List ( CLOSED )

["Matteo Giantomassi" <Matteo.Giantomassi@uclouvain.be>]

>>
> other persons also suggested me leaving ecutwfn as ecut, but i found
> physical memory is exhausted with default value for ecutwfn when
> performing
> screening calculations, especially for q(0.00,0.00, 0.00).  Therefore i
> should do testings on ecutwfn and ecuteps to figure out the proper
> converged
> values. Do you have any tricks to tune physical memory load for screening
> calculation ?
>
>
  fftgw 11

  The dafault (fftgw 21) is more accurate (convolutions are
  treated  properly) but the price to pay is a significant
  increase of memory and CPU.

  In the forthcoming versions the default value of fftgw will
  be reset to 11.

  If fftgw 11 doesn't solve you problem, you might try gwmem.
 Usually gwmem 10 allows a significant memory saving
 at the price of a reasonable worsening of the CPU time.

 PS: I guess you are running in parallel over bands (gwpara=2)..
 Note that the k-point parallelism (gwpara = 1) doesn't help
 reduce the memory consumption!

 Regards
 Matteo Giantomassi



> Best wishes,
>
> Hai-Ping
>
>
>
>
>>
>>
>>  Hope it helps
>>  Regards
>>  Matteo Giantomassi
>>
>>
>>
>> >
>> >  completechi0_deltapart : WARNING  Number of G1-G2 pairs outside the
>> > G-sphere for Wfns =    72
>> >
>> >  completechi0_deltapart : WARNING  Number of G1-G2 pairs outside the
>> > G-sphere for Wfns =    72
>>
>>
>
>
> --
> Hai-Ping Lan
> Department of Electronics ,
> Peking University , Bejing, 100871
> lanhaiping@gmail.com, hplan@pku.edu.cn
>