[Mesa-users] memory usage net

[amber lauer]

OK! So if I understand what you're saying, calculating memory use is
essentially not a predictive process, but is done at run time.

And if I want to narrow it down within MESA, I can use Valgrind. Well,
thankfully I think I have a working model that fits within the parameters I
need, so I'm going to press on. Thanks.

On Sun, Feb 24, 2019 at 11:22 AM Rob Farmer <r.j.farmer at uva.nl> wrote:

> Hi
> >(isos+5)*nz*(3*isos+24)
>
> So thats the scaling for the matrix we solve (not the net). Note there is
> a isos*isos term in there, so it shouldn't scale linearly but quadratically
> in number of isotopes.
>
> The other big memory user for mesa is the EOS then probably the net/rates
> storage. One way to check what is being used (especially for the things
> that will be ~ constant with number of isotopes, like the eos) is to use
> massif:
>
> valgrind --tool=massif ./star
> then use massif-visualizer to view the output and it will show where the
> memory is being used in mesa.
>
> Rob
>
> http://valgrind.org/docs/manual/ms-manual.html
> https://github.com/KDE/massif-visualizer
>
>
>
>
>
>
>
> On Fri, 22 Feb 2019 at 22:59, amber lauer via Mesa-users <
> mesa-users at lists.mesastar.org> wrote:
>
>> Hey guys, I'm doing some tests to optimize a model to not go over a
>> certain memory usage, and referring to this thread:
>>
>>
>> *https://lists.mesastar.org/pipermail/mesa-users/2016-August/006859.html
>> <https://lists.mesastar.org/pipermail/mesa-users/2016-August/006859.html>*
>>
>> I am using the rough equation that memory usage for the net scales as
>>
>> (isos+5)*nz*(3*isos+24)
>>
>> But doing a quick test on my own, I get the following results:
>>
>> num isos   num zones  mem_calc  mem actual:
>> 53             ~750           8mb             1.06gb
>> 153           750             55mb            2.92gb
>> 305           750             200mb         8.96gb
>>
>> (now, I'm getting my memory usage numbers from top and multiplying the %
>> by my total)
>> and the last two columns don't seem to scale by a constant. I'm looking
>> for information on how to predict the overall usage of my model  to keep it
>> below a fixed limit, and though I'm happy to use brute force to achieve the
>> desired results, I'm wondering what I'm missing. What other factors go into
>> the memory usage? Thanks.
>>
>>
>>
>> --
>> Amber Lauer
>> Postdoctoral Researcher
>> Triangle Universities Nuclear Lab
>> Duke University
>> amber.lauer at tunl.duke.edu <amber.lauer at duke.edu>
>> _______________________________________________
>> mesa-users at lists.mesastar.org
>> https://lists.mesastar.org/mailman/listinfo/mesa-users
>>
>>

-- 
Amber Lauer
Postdoctoral Researcher
Triangle Universities Nuclear Lab
Duke University
amber.lauer at tunl.duke.edu <amber.lauer at duke.edu>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20190225/a3459947/attachment.html>