From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: perf record and intel_cqm/llc_occupancy Date: Fri, 24 Jun 2016 09:43:50 -0700 Message-ID: <20160624164350.GZ13997@two.firstfloor.org> References: <8760szocdp.fsf@tassilo.jf.intel.com> <8760szocdp.fsf@tassilo.jf.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from one.firstfloor.org ([193.170.194.197]:47457 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751721AbcFXQnx (ORCPT ); Fri, 24 Jun 2016 12:43:53 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-perf-users-owner@vger.kernel.org List-ID: To: Omar Awile Cc: "linux-perf-users@vger.kernel.org" , Andi Kleen , Aram Santogidis , kan.liang@intel.com On Fri, Jun 24, 2016 at 12:27:27PM +0000, Omar Awile wrote: > Hi Andi, >=20 > Thanks, I see your point. I=E2=80=99ve seen tried perf stat for these= events, which is quite nice. However, we would like to be able to corr= elate the cache occupancy with the current call stack. I don=E2=80=99t = see how this could be done with perf stat. Is there a way to do this? perf stat cannot collect call graphs. >=20 > Alternatively, since perf record already keeps track of the call stac= k, is it feasible to extend it to allow recording cache occupancy, or i= s this inherently impossible? perf can collect counts while sampling another event (called leader sampling using :S). Unfortunately this does not work with events from different PMUs, which would be needed here. Kan looked into this some time ago, but there were too many problems with it. >=20 > Finally, if none of the two alternatives work, do you have an idea ho= w else this could be achieved? =46or now the only way would be to sample and perf stat record in paral= lel and then merge the data based on the time stamps. It likely will not be very accurate. -Andi