The stats in the profiler was the first thing that come to my mind too.
But we are talking hundreds of mega-bytes here. I do not think our poor little
histograms are that ugly.
Thanks.
--
Nauman
Quoting Mike Marty <mikem@xxxxxxxxxxx>:
>
> Yes, I've seen the same behavior.
>
> We don't know of any memory leaks in GEMS. I've run Valgrind/Purify on
> Ruby before and didn't find any leaks.
>
> It may be due to Ruby's profiler that continues to gather statistics and
> build histograms. We have not had time to study how these structures grow
> with simulation time. I think the best way to measure this is to add code
> to periodically check the sizes of the various data structures. Its been
> on my long-term TODO list.
>
> I suspect that Simics resident memory usage grows as more of the target
> memory and disk is accessed. Simics contains its own memory manager and
> usually caps the amount of resident memory used however.
>
> Not sure if any of this helps you, but you are not alone in seeing this
> behavior. It hasn't been problematic enough for me to personally look
> into.
>
> --Mike
>
> > My simulation with gems keep on increasing their memory usage with time.
> The
> > final memory usage is sometimes double or even more than double of the
> initial
> > memory usage.
> > It becomes a real problem with TPC, specially if I run it longer.
> > Is it a common behavior?
> > Is this a memory leak or some other phenomenon is going on?
> >
> > Thanks.
> > --
> > Nauman
> >
> >
> > _______________________________________________
> > Gems-users mailing list
> > Gems-users@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> >
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
>
|