Code review comment for lp:~maddevelopers/mg5amcnlo/3.0.2

Valentin Hirschi (valentin-hirschi) wrote :

Very naively I would expect the source of large memory consumption as to be
stemming from either a memory leak of PolyFit unnecessarily keeping all
training sample fed in memory.
Rik, do you know which one it is? If the latter, isn't there any easy way
of truncating the internal "history" of PolyFit so as to effectively cap
its memory footprint?

On Wed, May 13, 2020 at 5:45 PM marco zaro <email address hidden> wrote:

> Hi Rikkert,
> thanks for your work on this. As I told you, I had crashes on the louvain
> cluster, due to the large memory and space usage. I guess that if you have
> few jobs that can be run on multicore, then you do get improvements, but i
> think that on a cluster it can create problems…
>
> cheers,
>
> marco
>
>
> > On 13 May 2020, at 17:42, Rikkert Frederix <email address hidden>
> wrote:
> >
> > Hi,
> >
> > It's kind of a pity that this PolyVirtual doesn't work for you. Let's
> keep it turned off by default as Marco suggests -- at least for this
> release.
> >
> > Best,
> > Rikkert
> >
> > --
> > https://code.launchpad.net/~maddevelopers/mg5amcnlo/3.0.2/+merge/381317
> > Your team MadDevelopers is subscribed to branch
> lp:~maddevelopers/mg5amcnlo/3.0.2.
>
>
> --
> https://code.launchpad.net/~maddevelopers/mg5amcnlo/3.0.2/+merge/381317
> Your team MadDevelopers is subscribed to branch
> lp:~maddevelopers/mg5amcnlo/3.0.2.
>

--
Valentin

« Back to merge proposal