Hi Hemant,
----- Original Message -----
> On 07/21/2015 04:52 AM, Nathan Scott wrote:
> > Yep, thats the idea. No hard-coding, all representation through the config
> > file.
>
> Makes sense. Thanks for the explanation.
>
> Few questions below :
>
> > Just had a thought that might make this all a whole lot simpler - it may be
> > possible here to combine use of pmdaperfevent (for direct hw counter
> > access)
> > and the existing pmdasummary (for the server-side derivation calculation).
> >
> > For example, lets pretend those memory metrics you're interested in are now
> > exported via perfevent.conf as, say, "perfevent.hwcounters.MEM.value". You
> > might be able to create a pmdasummary configuration file like
> >
> > summary.bandwidth.count = sum_inst(perfevent.hwcounters.MEM.value) /
> > hinv.nnode;
> > summary.bandwidth.max = hinv.memory.max_bandwidth;
>
> Here, the above perfevent.hwcounters.MEM.value represents the consolidated
> value (aggregated and multiplied with the scale) or a single counter's
> value?
I think pmdasummary (i.e. pmie) could do the aggregation and scaling for
you.
I don't have the hardware to try it, but give the default perfevent.conf
that Joe's supplied with pmdaperfevent a go - it seems to be listing all
of the events that you were using in pmdabandwidth. So, it may work out
of the box in terms of exporting the raw values you were looking for,
then you should be able to coerce pmdasummary into doing any additional
aggregation/scaling needed.
> And, this configuration file for pmdasummary shall have to be in the
> client side, right?
pmdasummary is all server side.
cheers.
--
Nathan
|