I see. I am modifying self_basic.c to monitor those events, where I fork
a process to execute the command line arguments and use perf_event_open
to track the corresponding process id.
Binh
On 12/12/13 10:23 AM, Stephane Eranian wrote:
On Mon, Dec 9, 2013 at 7:21 PM, Binh Q. Pham <phambinh1...@gmail.com
<mailto:phambinh1...@gmail.com>> wrote:
Sure, they are:
1. INSTRUCTION_RETIRED
2. CYCLES
3. DTLB_LOAD_MISSES:MISS_CAUSES_A_WALK
4. DTLB_LOAD_MISSES:WALK_DURATION
5. DTLB_LOAD_MISSES:STLB_HIT
6. DTLB_STORE_MISSES:MISS_CAUSES_A_WALK
If you don't have the NMI watchdog enabled, then 1-6 can go together.
Otherwise, it's 1-5.
7. DTLB_STORE_MISSES:WALK_DURATION
8. DTLB_STORE_MISSES:STLB_HIT
Please let me know if you need the hexa values of these events as
welll.
And what tool or example or you using with those events?
You can pass more events than there are counters, but in that case the
kernel will
multiplex them for you and the tool takes care of scaling the counts
for you.
Thank you,
Binh
On 12/06/2013 07:40 PM, Stephane Eranian wrote:
Hi,
Please provide the full list of events you are using.
Thanks.
On Dec 3, 2013 1:58 PM, "Binh Q. Pham" <phambinh1...@gmail.com
<mailto:phambinh1...@gmail.com>> wrote:
I am having a problem while using libpfm-4.4.0, which I think
related to
this question. What I see is that the number of instructions
retired
drops significantly if I add a few more events to the list of
monitored
events in my program. I suspect that the added events
conflict with the
existing ones, hence brings the actual count down. Do you
have the same
thoughts?
Also, could you point me to where I can find information
about these
conflicting events in order to configure them correctly? How
can I know
how many performance counters that my CPU has?
Thanks,
Binh
On 07/23/2013 02:22 AM, Stephane Eranian wrote:
> Hi,
>
> On Tue, Jul 23, 2013 at 2:25 AM, Leonardo Piga
<leonardo.p...@gmail.com <mailto:leonardo.p...@gmail.com>> wrote:
>> Hello,
>>
>> I am trying to measure some performance counters on Ivy
Bridge, which has 4
>> PCM. I would like to know if there is a way to group up
counters so that I
>> can measure events on the same time.
>>
>> For example, I need to measure 8 counters (e.g.,
A,B,C,D,E,F,G,H). And I
>> would like to sample A,B,C,D together and E,F,G,H together.
>>
>> How can I do this using libpfm? I saw that task.c on
perf_examples has a way
>> to group up counters, but I couldn't use it.
>>
> task -e A,B,C,D -e E,F,G,H .....
>
> Of course in each group, events have to be compatible with
each other.
> That means they cannot conflict on a counter. If so then,
try moving the
> conflicting events to separate groups.
>
> Also a group cannot have more events than there are counters.
> Note that by default the NMI watchdog uses one counter and runs
> all the time unless you disable it via
/proc/sys/kernel/nmi_watchdog.
>
>> Thank you for any help,
>>
>> --
>> Leonardo
>>
>>
------------------------------------------------------------------------------
>> See everything from the browser to the database with
AppDynamics
>> Get end-to-end visibility with application monitoring from
AppDynamics
>> Isolate bottlenecks and diagnose root cause in seconds.
>> Start your free trial of AppDynamics Pro today!
>>
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
>> _______________________________________________
>> perfmon2-devel mailing list
>> perfmon2-devel@lists.sourceforge.net
<mailto:perfmon2-devel@lists.sourceforge.net>
>> https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
>>
>
------------------------------------------------------------------------------
> See everything from the browser to the database with
AppDynamics
> Get end-to-end visibility with application monitoring from
AppDynamics
> Isolate bottlenecks and diagnose root cause in seconds.
> Start your free trial of AppDynamics Pro today!
>
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
> _______________________________________________
> perfmon2-devel mailing list
> perfmon2-devel@lists.sourceforge.net
<mailto:perfmon2-devel@lists.sourceforge.net>
> https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
--
-Binh
------------------------------------------------------------------------------
Sponsored by Intel(R) XDK
Develop, test and display web and hybrid apps with a single
code base.
Download it for free now!
http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
<mailto:perfmon2-devel@lists.sourceforge.net>
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
--
-Binh
--
------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
