From: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
To: "Hernandez Lopez, Fabiola" <fabiola.hernandez.lopez@intel.com>,
"Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: "linux-pm@vger.kernel.org" <linux-pm@vger.kernel.org>
Subject: Re: Frequency not returning to fixed value after AVX workload
Date: Mon, 14 Oct 2019 14:23:15 -0700 [thread overview]
Message-ID: <e7e89cd85fb5d367338491651b36cc912967edab.camel@linux.intel.com> (raw)
In-Reply-To: <67737C20FC7E4D44B44817ABFA5B1DCE6EC020E9@CRSMSX104.amr.corp.intel.com>
On Mon, 2019-10-14 at 21:11 +0000, Hernandez Lopez, Fabiola wrote:
> How do you measure the frequency?
Do whatever you are doing to get to your problem state and while doing
the test run in another window:
#turbostat
Attach the output.
Thanks,
Srinivas
> With lscpu.
>
> What exactly do you do to get the expected frequency back?
> A script is used that sets the frequency with
> /sys/devices/system/cpu/cpu*/cpufreq/scaling_setspeed
>
> Is the target frequency in the turbo range?
> No, the target frequency is the same as the base processor frequency:
> 2.4GHz. The max turbo frequency for this CPU si 3.9 GHz.
> This has been tested with turbo boost enabled and disabled.
>
> This behavior is very sporadic: < 10%
>
> Thanks,
> Fabiola
>
> -----Original Message-----
> From: Rafael J. Wysocki [mailto:rjw@rjwysocki.net]
> Sent: Monday, October 14, 2019 4:45 AM
> To: Hernandez Lopez, Fabiola <fabiola.hernandez.lopez@intel.com>
> Cc: linux-pm@vger.kernel.org; Srinivas Pandruvada <
> srinivas.pandruvada@linux.intel.com>
> Subject: Re: Frequency not returning to fixed value after AVX
> workload
>
> On Friday, October 11, 2019 7:45:55 PM CEST Hernandez Lopez, Fabiola
> wrote:
> > Hi,
> >
> > We are seeing an unexpected behavior after applying AVX workloads.
> > After setting the CPU frequency to a fixed value with CPU Freq -
> > userspace governor and applying heavy AVX workloads, the CPU
> > frequency
> > is decreased (as expected) but it never returns to the previously
> > established value.
>
> How do you measure the frequency?
>
> > This does not happen on all cores, only on a single core. The only
> > way
> > to return to the desired frequency is by setting it through the
> > command line again.
>
> What exactly do you do to get the expected frequency back?
>
> > We are wondering how can this change in frequency happen.
>
> Is the target frequency in the turbo range?
>
>
>
next prev parent reply other threads:[~2019-10-14 21:23 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-11 17:45 Frequency not returning to fixed value after AVX workload Hernandez Lopez, Fabiola
2019-10-14 9:44 ` Rafael J. Wysocki
2019-10-14 21:11 ` Hernandez Lopez, Fabiola
2019-10-14 21:23 ` Srinivas Pandruvada [this message]
2019-10-23 18:26 ` Hernandez Lopez, Fabiola
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e7e89cd85fb5d367338491651b36cc912967edab.camel@linux.intel.com \
--to=srinivas.pandruvada@linux.intel.com \
--cc=fabiola.hernandez.lopez@intel.com \
--cc=linux-pm@vger.kernel.org \
--cc=rjw@rjwysocki.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).