[petsc-dev] Getting Petsc memory in use over time

[Junchao Zhang]

On Wed, Apr 1, 2020 at 12:00 PM Jacob Faibussowitsch <jacob.fai at gmail.com>
wrote:

>  but it does report the peak usage.
>
>
> On default -log_view? Or is it another option? I have always been using
> the below from -log_view, I always thought it was total memory usage.
>
Try  -log_view -log_view_memory created by Barry.


>
> Memory usage is given in bytes:
>
> Object Type          Creations   Destructions     Memory  Descendants' Mem.
> Reports information only for process 0.
>
> --- Event Stage 0: Main Stage
>
>          Matrix                    11             11        38060     0.
>          Vec Scatter             3              3         2400       0.
>          Vector                     6              6         9776       0.
>          Index Set              12             12        10776     0.
>          Star Forest Graph   3              3         3360       0.
>          Viewer                     4              3         2520       0.
>
>
> Best regards,
>
> Jacob Faibussowitsch
> (Jacob Fai - booss - oh - vitch)
> Cell: (312) 694-3391
>
> On Apr 1, 2020, at 11:56 AM, Jed Brown <jed at jedbrown.org> wrote:
>
> PETSc does not have this type of memory tracing, but it does report the
> peak usage.
>
> Jacob Faibussowitsch <jacob.fai at gmail.com> writes:
>
> Hello all,
>
> Is there a way to get memory usage in time from petsc? I know using
> -log_view gives you total amount of memory usage by the end of the program,
> but I'm interested in finding the point in time when petsc has the most
> _concurrently_ allocated amount of memory. It's fine if petsc allocated say
> 300 petabytes of memory over the program lifetime but it is useful to see
> when petsc is at its most memory intensive. Valgrind has the
> "--tool=massif" tool for this but I'm wondering if petsc has an in-house
> version.
>
> Best regards,
>
> Jacob Faibussowitsch
> (Jacob Fai - booss - oh - vitch)
> Cell: (312) 694-3391
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20200401/9cfebed8/attachment.html>