[Mesa-users] rsp run time

[amber lauer]

20 minutes is pretty good, in my opinion. However, there are a few more
things you can do.

Assuming you don't need the visualizations, make sure that you use these
settings:

in star_job:

pgstar_flag = .false.

in controls:

terminal_interval= some large number
photo_interval= some large number

and also take a look at the other controls for output:

http://mesa.sourceforge.net/controls_defaults.html#controls_for_output

Not sure what net you're using but you could also benchmark different size
nets and if you can show that a smaller net doesn't significantly alter the
model, you can shrink the net. I'm sure others will have some suggestions
too, those are the most obvious ones I know about. Good luck!
=================================
Dr. Amber Lauer
Postdoctoral Researcher
Triangle Universities Nuclear Lab
Duke University
amber.lauer at tunl.duke.edu <amber.lauer at duke.edu>


On Wed, Jan 22, 2020 at 10:10 PM Jeremy Sakstein <sakstein at hawaii.edu>
wrote:

> Hi everyone,
>
> I'm gearing up to run a large grid of RSP models and it's essential that I
> get the run time down as much as possible.
>
> I am only interested in the last cycle after it has converged (the history
> file only, no profiles), which is always by 800 cycles for the parameter
> space I am interested in) so my procedure so far has been the following:
>
> 1) Cut the history columns to as few as possible.
> 2) No profiles.
> 2) Run with 150 time steps per cycle (and no history) until cycle 800 then
> restart from the last saved photo for a few steps with 600 time steps per
> cycle and histroy_interval=1 to produce the model I need. (I checked that
> this is identical to running with 600 time steps per cycle from the outset.)
>
> This has gotten each run down to about 20 minutes (12 cores on a super
> computer) but ideally I would like better. I am just emailing on the off
> chance to see if anyone has any further suggestions?
>
> Cheers,
>
> Jeremy
> _______________________________________________
> mesa-users at lists.mesastar.org
>
> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.mesastar.org_mailman_listinfo_mesa-2Dusers&d=DwICAg&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=IZ3U2LUL6yhDK-Uf8F_eU9W85o61MM43boq-nVwHt38&m=EGRCxw5ue2oJxpqblqEhgQz4dPlff7HK8X34mbXmBM8&s=OfYlJMibP6DGE20zc4QJVLBhpG9Frldw3gdaLypG27o&e=
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20200123/9425c986/attachment.htm>