* Changing frequency rate to perf mem
@ 2013-10-23 15:43 Harald Servat
2013-10-24 5:12 ` Andi Kleen
0 siblings, 1 reply; 5+ messages in thread
From: Harald Servat @ 2013-10-23 15:43 UTC (permalink / raw)
To: linux-perf-users
Hello list,
I'm interested on changing the frequency rate to the perf mem tool. I
understand that this tool captures the data every N loads or stores,
however I don't see how to tune N.
I've looked into builtin-record.c approx line 980 and I found that
changing rec->opts.freq allows me capturing more or less data.
Initially, its value is 4000 and generates 4.5Mbytes for an application.
If I modify this value to 200, it generates 0.2Mbytes, therefore
200/4000 is closely 0.2/4.5 which seems the value I want to tune.
I'd like to know what is this rec->opts.freq with respect to perf
mem. I expected to setup the number of loads, therefore the higher N the
lesser number of samples generated, but it seems the other way around.
Where I can get any hint on what does this mean.
BTW, I tried to change this value by adding "-F" and "200" into the
argv that is passed to cmd_report by adding
rec_argv[i++] = strdup("-F");
rec_argv[i++] = strdup("200");
into line 42 of perf-mem.c - just after the strdup("record").
Unfortunately, this seems to break something and perf segfaults often.
Which is the most appropriate way to tune the user frequency of perf mem?
Thank you very much in advance!
--
WARNING / LEGAL TEXT: This message is intended only for the use of the
individual or entity to which it is addressed and may contain
information which is privileged, confidential, proprietary, or exempt
from disclosure under applicable law. If you are not the intended
recipient or the person responsible for delivering the message to the
intended recipient, you are strictly prohibited from disclosing,
distributing, copying, or in any way using this message. If you have
received this communication in error, please notify the sender and
destroy and delete any copies you may have received.
http://www.bsc.es/disclaimer
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Changing frequency rate to perf mem
2013-10-23 15:43 Changing frequency rate to perf mem Harald Servat
@ 2013-10-24 5:12 ` Andi Kleen
2013-10-24 7:48 ` Harald Servat
0 siblings, 1 reply; 5+ messages in thread
From: Andi Kleen @ 2013-10-24 5:12 UTC (permalink / raw)
To: Harald Servat; +Cc: linux-perf-users
Harald Servat <harald.servat@bsc.es> writes:
>
> BTW, I tried to change this value by adding "-F" and "200" into the
> argv that is passed to cmd_report by adding
>
> rec_argv[i++] = strdup("-F");
> rec_argv[i++] = strdup("200");
>
> into line 42 of perf-mem.c - just after the
> strdup("record").
You would need to increase the array size too.
> Unfortunately, this seems to break something and
> perf segfaults often. Which is the most appropriate way to tune the
> user frequency of perf mem?
Best would be probably to fix it to pass through unknown
options, then -F/-c could be just used. I'm not sure
if this is possible easily with the git option parser.
Alternatively you can just specify the command line perf mem record
would specify directly to perf record, plus -c/-F
-andi
--
ak@linux.intel.com -- Speaking for myself only
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Changing frequency rate to perf mem
2013-10-24 5:12 ` Andi Kleen
@ 2013-10-24 7:48 ` Harald Servat
2013-10-28 6:50 ` Namhyung Kim
0 siblings, 1 reply; 5+ messages in thread
From: Harald Servat @ 2013-10-24 7:48 UTC (permalink / raw)
To: Andi Kleen; +Cc: linux-perf-users
On 24/10/13 07:12, Andi Kleen wrote:
> Harald Servat <harald.servat@bsc.es> writes:
>>
>> BTW, I tried to change this value by adding "-F" and "200" into the
>> argv that is passed to cmd_report by adding
>>
>> rec_argv[i++] = strdup("-F");
>> rec_argv[i++] = strdup("200");
>>
>> into line 42 of perf-mem.c - just after the
>> strdup("record").
>
> You would need to increase the array size too.
Ouch! You're right! I forgot that!
>
>> Unfortunately, this seems to break something and
>> perf segfaults often. Which is the most appropriate way to tune the
>> user frequency of perf mem?
>
> Best would be probably to fix it to pass through unknown
> options, then -F/-c could be just used. I'm not sure
> if this is possible easily with the git option parser.
Sorry, I'm not sure what is the "git option parser" here.
I think that this could be implemented extending the mem_options[] in
file builtin-mem.c lines 205-221.
> Alternatively you can just specify the command line perf mem record
> would specify directly to perf record, plus -c/-F
>
> -andi
>
Thank you very much Andi.
--
WARNING / LEGAL TEXT: This message is intended only for the use of the
individual or entity to which it is addressed and may contain
information which is privileged, confidential, proprietary, or exempt
from disclosure under applicable law. If you are not the intended
recipient or the person responsible for delivering the message to the
intended recipient, you are strictly prohibited from disclosing,
distributing, copying, or in any way using this message. If you have
received this communication in error, please notify the sender and
destroy and delete any copies you may have received.
http://www.bsc.es/disclaimer
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Changing frequency rate to perf mem
2013-10-24 7:48 ` Harald Servat
@ 2013-10-28 6:50 ` Namhyung Kim
2013-10-28 8:15 ` Harald Servat
0 siblings, 1 reply; 5+ messages in thread
From: Namhyung Kim @ 2013-10-28 6:50 UTC (permalink / raw)
To: Harald Servat; +Cc: Andi Kleen, linux-perf-users
Hi Harald,
On Thu, 24 Oct 2013 09:48:09 +0200, Harald Servat wrote:
> On 24/10/13 07:12, Andi Kleen wrote:
>> Harald Servat <harald.servat@bsc.es> writes:
>>> Unfortunately, this seems to break something and
>>> perf segfaults often. Which is the most appropriate way to tune the
>>> user frequency of perf mem?
>>
>> Best would be probably to fix it to pass through unknown
>> options, then -F/-c could be just used. I'm not sure
>> if this is possible easily with the git option parser.
>
> Sorry, I'm not sure what is the "git option parser" here.
Just implementation details. The source code of the option parser came
from the git project.
> I think that this could be implemented extending the mem_options[] in
> file builtin-mem.c lines 205-221.
Or else, use 'perf record' directly as Andi suggested.
>
>> Alternatively you can just specify the command line perf mem record
>> would specify directly to perf record, plus -c/-F
$ perf record -Wd -e cpu/mem-loads/pp -F 200 -- ls
$ perf mem report
Thanks,
Namhyung
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Changing frequency rate to perf mem
2013-10-28 6:50 ` Namhyung Kim
@ 2013-10-28 8:15 ` Harald Servat
0 siblings, 0 replies; 5+ messages in thread
From: Harald Servat @ 2013-10-28 8:15 UTC (permalink / raw)
To: Namhyung Kim; +Cc: Andi Kleen, linux-perf-users
>> I think that this could be implemented extending the mem_options[] in
>> file builtin-mem.c lines 205-221.
>
> Or else, use 'perf record' directly as Andi suggested.
>
>>
>>> Alternatively you can just specify the command line perf mem record
>>> would specify directly to perf record, plus -c/-F
>
> $ perf record -Wd -e cpu/mem-loads/pp -F 200 -- ls
>
> $ perf mem report
>
Wow, thank you for such details! I'll take a look at this!
Best regards.
> Thanks,
> Namhyung
>
--
WARNING / LEGAL TEXT: This message is intended only for the use of the
individual or entity to which it is addressed and may contain
information which is privileged, confidential, proprietary, or exempt
from disclosure under applicable law. If you are not the intended
recipient or the person responsible for delivering the message to the
intended recipient, you are strictly prohibited from disclosing,
distributing, copying, or in any way using this message. If you have
received this communication in error, please notify the sender and
destroy and delete any copies you may have received.
http://www.bsc.es/disclaimer
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2013-10-28 8:15 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-10-23 15:43 Changing frequency rate to perf mem Harald Servat
2013-10-24 5:12 ` Andi Kleen
2013-10-24 7:48 ` Harald Servat
2013-10-28 6:50 ` Namhyung Kim
2013-10-28 8:15 ` Harald Servat
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).