[Nek5000-users] Load_fld() without restart

[nek5000-users at lists.mcs.anl.gov]

Hi Holly,

For certain, you would want to use the .f00000 format for this,
with MPI-IO turned on via the flag in makenek.

What you propose would definitely be i/o intensive and I expect
more time would be spent reading than computing.  Are you certain
you need to re-load at every timestep?

Paul

On Mon, 20 Jan 2014, nek5000-users at lists.mcs.anl.gov wrote:

> Hi Holly,
>
> I'm not quite sure how to estimate how much time is needed to load the .fld
> files, it would certainly depend on how many processors you are using,
> whether you are using MPI_IO, etc.. (maybe the main developers could give
> some sort of leading order complexity here?)
>
> As for the variables, the main variables (vx, vy, vz, pr, t) are
> overwritten each time step, although I think there are some lag arrays
> available (such as vxlag).  However, I'm not familiar with how these are
> used, so you may want to do some digging.
>
> Josh
>
>
> On Mon, Jan 20, 2014 at 2:51 AM, <nek5000-users at lists.mcs.anl.gov> wrote:
>
>>  Hi Josh,
>>
>> Thank you for your answer. We had already considered using this approach
>> but we have a few queries. We would like to load a .fld file at each time
>> step, which means the process might become very costly. Do you have any
>> idea of the cost (in time) of such a method? Also, what becomes of the
>> variables calculated during the previous time steps?
>>
>> Thanks again,
>>
>> Holly
>>
>> Le 17/01/2014 14:21, nek5000-users at lists.mcs.anl.gov a écrit :
>>
>> Hi Holly,
>>
>>  If you wanted to load a .fld file without copying over your variables,
>> you could do something like
>>
>>  ! Create temporary storage
>> real tmp1(lx1*ly1*lz1*lelv), tmp2(lx1*ly1*lz1*lelv), ...
>>
>>  ! Copy simulation fields to temporary variables
>> call copy(tmp1,vx,nx1*ny1*nz1*nelv)
>> ...
>>
>>  ! Load field file and do whatever work you want to do with it
>>
>>  ! Copy simulation fields back to nek variables
>> call copy(vx,tmp1,nx1*ny1*nz1*nelv)
>>
>>
>>  You should be able to use some of the Nek scratch arrays to store your
>> temporary variables (instead of taking up more memory with the explicit
>> declaration), but I'm not sure at the moment which scratch arrays
>> load_fld() uses.
>>
>>  - Josh
>>
>>
>>
>> On Fri, Jan 17, 2014 at 4:56 AM, <nek5000-users at lists.mcs.anl.gov> wrote:
>>
>>> Hi Neks !
>>>
>>> Is there a way to load a .fld file during a simulation (in the same way
>>> load_fld() does) but without doing a complete restart of all the variables ?
>>>
>>> Thanks for any help !
>>>
>>> Holly
>>> _______________________________________________
>>> Nek5000-users mailing list
>>> Nek5000-users at lists.mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>
>>
>>
>>
>>  --
>> Josh Camp
>>
>> "All that is necessary for the triumph of evil is that good men do
>> nothing" -- Edmund Burke
>>
>>
>> _______________________________________________
>> Nek5000-users mailing listNek5000-users at lists.mcs.anl.govhttps://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>
>>
>> _______________________________________________
>> Nek5000-users mailing list
>> Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>
>>
>
>
> -- 
> Josh Camp
>
> "All that is necessary for the triumph of evil is that good men do nothing"
> -- Edmund Burke
>