Re: [PATCH] [RFC] timekeeping: Rework frequency adjustments to workbetter w/ nohz

From: John Stultz
Date: Wed Jan 29 2014 - 12:56:29 EST


On Tue, Jan 28, 2014 at 9:58 AM, Richard Cochran
<richardcochran@xxxxxxxxx> wrote:
> I tested for a regression using the patched kernel with the nohz=off
> command line option...
>
> On Mon, Jan 13, 2014 at 10:15:18AM -0800, John Stultz wrote:
>> On 01/13/2014 09:51 AM, Richard Cochran wrote:
>> >
>> > - Linux 3.12.7-nohz-plain-20140106 nohz-plain.log
>> > - Linux 3.12.7-nohz-plain-20140106 NOHZ=OFF periodic-plain.log
>> > - Linux 3.12.7-nohz-fix-20140106-00001-gd753140 nohz-fix.log
>
> Regression check:
>
> - Linux homeboy 3.12.7-nohz-fix-20140106-00001-gd753140 NOHZ=OFF
> nohz-fix-periodic.log
>
>> > The performance in the log files as reflected in the clock offset is
>> > summarized in this table. The values are in nanoseconds.
>> >
>> > | | periodic-plain | nohz-fix | nohz-plain |
>> > |---------+----------------+---------------+---------------|
>> > | minimum | -1.599000e+03 | -1.051000e+03 | -5.373700e+04 |
>> > | maximum | +1.311000e+03 | +1.048000e+03 | +6.389500e+04 |
>> > | mean | +9.880240e-02 | -7.747305e+01 | +1.597904e+01 |
>> > | stddev | +4.610021e+02 | +3.960978e+02 | +1.491263e+04 |
>
> Comparing the nohz=off case with and without the patch, the three hour
> test looks like this.
>
> | | periodic-plain | nohz-fix-periodic |
> |---------+----------------+-------------------+
> | minimum | -1.599000e+03 | -1.427000e+03 |
> | maximum | +1.311000e+03 | +1.279000e+03 |
> | mean | +9.880240e-02 | -2.710778e+01 |
> | stddev | +4.610021e+02 | +3.974372e+02 |
>
>> > http://linuxptp.sourceforge.net/nohz-fix/current_nohz.png
>> > http://linuxptp.sourceforge.net/nohz-fix/periodic_vs_fix.png
>
> I also made a third graph showing before and after the patch.
>
> http://linuxptp.sourceforge.net/nohz-fix/nohz_regression.png
>
>> If you do get a chance to look again, I'd also be interested if running
>> with nohz=off w/ the fix doesn't show any regression compared to the
>> unmodified nohz=off case.
>
> Looks like there is no regression.

That's great to hear! Looks like we might be able to queue it for 3.15...

Thanks so much again for the detailed testing! I really appreciate it!

thanks
-john
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/