Does perf randomize the sampling period?

Does perf randomize the sampling period?

[Milian Wolff]

[-- Attachment #1: Type: text/plain, Size: 727 bytes --]

Hey all,

does perf randomize its sampling intervals/period to prevent pattern effects? 
I found https://lkml.org/lkml/2010/3/4/51 but neither the man page on perf 
record says anything on that matter, nor does the perf wiki say it's 
supported:

https://perf.wiki.kernel.org/index.php/Tutorial#Period_and_rate

> There is currently no support for sampling period randomization.

So is it supported, or not? If it is supported, how do I enable it? Or is this 
enabled by default?

If it is not supported, why so? Is it not important to have?

Thanks
-- 
Milian Wolff | milian.wolff@kdab.com | Software Engineer
KDAB (Deutschland) GmbH&Co KG, a KDAB Group company
Tel: +49-30-521325470
KDAB - The Qt Experts

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5903 bytes --]

Re: Does perf randomize the sampling period?

[Manuel Selva]

Regarding perf mem record on Intel platforms, I think that the sampling 
randomization is ensured by the hardware.

----
Manu

On 09/09/2015 05:11 PM, Milian Wolff wrote:
> Hey all,
>
> does perf randomize its sampling intervals/period to prevent pattern effects?
> I found https://lkml.org/lkml/2010/3/4/51 but neither the man page on perf
> record says anything on that matter, nor does the perf wiki say it's
> supported:
>
> https://perf.wiki.kernel.org/index.php/Tutorial#Period_and_rate
>
>> There is currently no support for sampling period randomization.
>
> So is it supported, or not? If it is supported, how do I enable it? Or is this
> enabled by default?
>
> If it is not supported, why so? Is it not important to have?
>
> Thanks
>

Re: Does perf randomize the sampling period?

[Andreas Hollmann]

According to this paper

Establishing a Base of Trust with Performance
Counters for Enterprise Workloads

software randomization is not available in perf.

"We program the PMU to sample with prime and
non-prime periods. Due to perf limitations, software-
based period randomization was unavailable, but the
hardware randomizes the 4 least significant bits."


2015-09-09 17:16 GMT+02:00 Manuel Selva <selva.manuel@gmail.com>:
> Regarding perf mem record on Intel platforms, I think that the sampling
> randomization is ensured by the hardware.
>
> ----
> Manu
>
> On 09/09/2015 05:11 PM, Milian Wolff wrote:
>>
>> Hey all,
>>
>> does perf randomize its sampling intervals/period to prevent pattern
>> effects?
>> I found https://lkml.org/lkml/2010/3/4/51 but neither the man page on perf
>> record says anything on that matter, nor does the perf wiki say it's
>> supported:
>>
>> https://perf.wiki.kernel.org/index.php/Tutorial#Period_and_rate
>>
>>> There is currently no support for sampling period randomization.
>>
>>
>> So is it supported, or not? If it is supported, how do I enable it? Or is
>> this
>> enabled by default?
>>
>> If it is not supported, why so? Is it not important to have?
>>
>> Thanks
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-perf-users"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Does perf randomize the sampling period?

[Andi Kleen]

Milian Wolff <milian.wolff@kdab.com> writes:

> Hey all,
>
> does perf randomize its sampling intervals/period to prevent pattern
> effects?

It does not.

Usually you can get most of the benefits of randomization by using a
prime period with -c.

-Andi
-- 
ak@linux.intel.com -- Speaking for myself only