Re: [PATCH v1 1/2] perf header: Support HYBRID_TOPOLOGY feature
From: Jiri Olsa
Date: Wed May 05 2021 - 16:28:57 EST
On Wed, May 05, 2021 at 10:49:23AM -0300, Arnaldo Carvalho de Melo wrote:
> Em Tue, May 04, 2021 at 09:37:34PM +0200, Jiri Olsa escreveu:
> > On Tue, May 04, 2021 at 04:28:33PM -0300, Arnaldo Carvalho de Melo wrote:
> > > Em Tue, May 04, 2021 at 04:56:44PM +0200, Jiri Olsa escreveu:
> > > > On Fri, Apr 30, 2021 at 03:46:01PM +0800, Jin Yao wrote:
> > > > > It would be useful to let user know the hybrid topology.
> > > > > Adding HYBRID_TOPOLOGY feature in header to indicate the
> > > > > core cpus and the atom cpus.
> > >
> > > > > With this patch,
> > >
> > > > > For the perf.data generated on hybrid platform,
> > > > > reports the hybrid cpu list.
> > >
> > > > > root@otcpl-adl-s-2:~# perf report --header-only -I
> > > > > ...
> > > > > # cpu_core cpu list : 0-15
> > > > > # cpu_atom cpu list : 16-23
> > >
> > > > hum, should we print 'hybrid:' or something to make
> > > > sure its not confused with something else? like
> > >
> > > > # hybrid cpu_core cpu list : 0-15
> > > > # hybrid cpu_atom cpu list : 16-23
> > >
> > > But this _core/_atom already got to be enough? I disagreed with that
> > > naming, but neverthless having one or the other present in an output is
> > > a clear mark of this hybrid topology.
> > >
> > > I.e having that extra hybrid string that wouldn't add information to the
> > > output.
> >
> > sure when you know that cpu_core/cpu_atom are hybrid pmus ;-)
> > and I guess other arch will come with other names
>
> Yeah, its too Intel centric, I thought they would come up with
> cpu_big/cpu_little and map it to core/atom on Intel and whatever other
> BIG/little arches come up with.
>
> Perhaps:
>
> root@otcpl-adl-s-2:~# perf report --header-only -I
> ...
> # hybrid cpu system:
> # cpu_core cpu list : 0-15
> # cpu_atom cpu list : 16-23
>
> ?
'hybrid pmus' would sounds better to me,
but as long as there's hybrid in there I'm good ;-)
thanks,
jirka
>
> - Arnaldo
>