From: Shailabh Nagar <nagar@watson.ibm.com>
To: Andrew Morton <akpm@osdl.org>
Cc: balbir@in.ibm.com, jlan@sgi.com, csturtiv@sgi.com,
linux-kernel@vger.kernel.org
Subject: Re: [Patch][RFC] Disabling per-tgid stats on task exit in taskstats
Date: Fri, 09 Jun 2006 17:56:12 -0400 [thread overview]
Message-ID: <4489EE7C.3080007@watson.ibm.com> (raw)
In-Reply-To: <20060609042129.ae97018c.akpm@osdl.org>
Andrew Morton wrote:
>
> You see the problem - if one userspace package wants the tgid-stats and
> another concurrently-running one does now, what do we do? Just leave it
> enabled and run a bit slower?
>
> If so, how much slower? Your changelog says some potential users don't
> need the tgid-stats, but so what? I assume this patch is a performance
> thing? If so, has it been quantified?
Here are some results from running a simple program (source below) that does
10 iterations of creating and then destroying 1000 threads. On the side, another utility
kept reading the pid (+tgid if present) stats from exiting tasks.
Yes No Ovhd
user 0.14 0.15 -6%
system 1.61 1.54 +5%
elapsed 2.01 1.94 +3%
Yes = tgid stats printed on exit
No = not printed
Ovhd = (Yes-No)/No * 100
So even in this extreme case where the per-tgid stats are indeed
half of the total data, the overhead is not very significant.
As pointed out earlier, more representative cases are
- single threaded apps (e.g. make -jX) where the current
taskstats interface already optimizes by not sending redundant per-tgid stats, or
- server-type multithreaded apps where the exits are going to be relatively infrequent (due to
reuse of thread pools) so the extra per-tgid output is not going to have much impact.
I'd suggest we drop the idea of including this patch until we have data showing that
the overhead is an issue.
--Shailabh
#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <pthread.h>
int n;
void *slow_exit(void *arg)
{
int i = (int) arg;
usleep((n-i)*2);
}
int main(int argc, char *argv[])
{
int i,rc, rep;
pthread_t *ppthread;
n = 5 ;
if (argc > 1)
n = atoi(argv[1]);
rep = 10;
if (argc > 2)
rep = atoi(argv[2]);
ppthread = malloc(n * sizeof(pthread_t));
if (ppthread == NULL) {
printf("Memory allocation failure\n");
exit(-1);
}
while (rep) {
for (i=0; i<n; i++) {
rc = pthread_create(&ppthread[i], NULL,
slow_exit, (void *)i);
if (rc) {
printf("Error creating thread %d\n", i);
exit(-1);
}
}
for (i=0; i<n; i++) {
rc = pthread_join(ppthread[i], NULL);
if (rc) {
printf("Error joining thread %d\n", i);
exit(-1);
}
}
rep--;
}
}
next prev parent reply other threads:[~2006-06-09 21:56 UTC|newest]
Thread overview: 134+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-06-09 7:41 [Patch][RFC] Disabling per-tgid stats on task exit in taskstats Shailabh Nagar
2006-06-09 8:00 ` Andrew Morton
2006-06-09 10:51 ` Balbir Singh
2006-06-09 11:21 ` Andrew Morton
2006-06-09 13:20 ` Shailabh Nagar
2006-06-09 18:25 ` Jay Lan
2006-06-09 19:12 ` Shailabh Nagar
2006-06-09 15:36 ` Balbir Singh
2006-06-09 18:35 ` Jay Lan
2006-06-09 19:31 ` Shailabh Nagar
2006-06-09 21:56 ` Shailabh Nagar [this message]
2006-06-09 22:42 ` Jay Lan
2006-06-09 23:22 ` Andrew Morton
2006-06-09 23:47 ` Jay Lan
2006-06-09 23:56 ` Andrew Morton
2006-06-10 12:21 ` Shailabh Nagar
2006-06-12 18:31 ` Jay Lan
2006-06-12 21:57 ` Shailabh Nagar
2006-06-10 13:05 ` Shailabh Nagar
2006-06-12 18:54 ` Jay Lan
2006-06-21 19:11 ` Jay Lan
2006-06-21 19:14 ` Jay Lan
2006-06-21 19:34 ` Shailabh Nagar
2006-06-21 23:35 ` Jay Lan
2006-06-21 23:45 ` Shailabh Nagar
2006-06-23 17:14 ` Shailabh Nagar
2006-06-23 18:19 ` Jay Lan
2006-06-23 18:53 ` Shailabh Nagar
2006-06-23 20:00 ` Jay Lan
2006-06-23 20:16 ` Shailabh Nagar
2006-06-23 20:36 ` Jay Lan
2006-06-23 21:19 ` Andrew Morton
2006-06-23 22:07 ` Jay Lan
2006-06-23 23:47 ` Andrew Morton
2006-06-24 2:59 ` Shailabh Nagar
2006-06-24 4:39 ` Andrew Morton
2006-06-24 5:59 ` Shailabh Nagar
2006-06-26 17:33 ` Jay Lan
2006-06-26 17:52 ` Shailabh Nagar
2006-06-26 17:55 ` Andrew Morton
2006-06-26 18:00 ` Shailabh Nagar
2006-06-26 18:12 ` Andrew Morton
2006-06-26 18:26 ` Jay Lan
2006-06-26 18:39 ` Andrew Morton
2006-06-26 18:49 ` Shailabh Nagar
2006-06-26 19:00 ` Jay Lan
2006-06-28 21:30 ` Jay Lan
2006-06-28 21:53 ` Andrew Morton
2006-06-28 22:02 ` Jay Lan
2006-06-29 8:40 ` Paul Jackson
2006-06-29 12:30 ` Valdis.Kletnieks
2006-06-29 16:44 ` Paul Jackson
2006-06-29 18:01 ` Andrew Morton
2006-06-29 18:07 ` Paul Jackson
2006-06-29 18:26 ` Paul Jackson
2006-06-29 19:15 ` Shailabh Nagar
2006-06-29 19:41 ` Paul Jackson
2006-06-29 21:42 ` Shailabh Nagar
2006-06-29 21:54 ` Jay Lan
2006-06-29 22:09 ` Shailabh Nagar
2006-06-29 22:23 ` Paul Jackson
2006-06-30 0:15 ` Shailabh Nagar
2006-06-30 0:40 ` Paul Jackson
2006-06-30 1:00 ` Shailabh Nagar
2006-06-30 1:05 ` Paul Jackson
[not found] ` <44A46C6C.1090405@watson.ibm.com>
2006-06-30 0:38 ` Paul Jackson
2006-06-30 2:21 ` Paul Jackson
2006-06-30 2:46 ` Shailabh Nagar
2006-06-30 2:54 ` Paul Jackson
2006-06-30 3:02 ` Paul Jackson
2006-06-29 19:22 ` Shailabh Nagar
2006-06-29 19:10 ` Shailabh Nagar
2006-06-29 19:23 ` Paul Jackson
2006-06-29 19:33 ` Andrew Morton
2006-06-29 19:43 ` Shailabh Nagar
2006-06-29 20:00 ` Andrew Morton
2006-06-29 22:13 ` Shailabh Nagar
2006-06-29 23:00 ` jamal
2006-06-29 20:01 ` Shailabh Nagar
2006-06-29 21:22 ` Paul Jackson
2006-06-29 22:54 ` jamal
2006-06-30 0:38 ` Shailabh Nagar
2006-06-30 1:05 ` Andrew Morton
2006-06-30 1:11 ` Shailabh Nagar
2006-06-30 1:30 ` jamal
2006-06-30 3:01 ` Shailabh Nagar
2006-06-30 12:45 ` jamal
2006-06-30 2:25 ` Paul Jackson
2006-06-30 2:35 ` Andrew Morton
2006-06-30 2:43 ` Paul Jackson
2006-06-29 19:33 ` Jay Lan
2006-06-30 18:53 ` Shailabh Nagar
2006-06-30 19:10 ` Shailabh Nagar
2006-06-30 19:19 ` Shailabh Nagar
2006-06-30 20:19 ` jamal
2006-06-30 22:50 ` Andrew Morton
2006-07-01 2:20 ` Shailabh Nagar
2006-07-01 2:43 ` Andrew Morton
2006-07-01 3:37 ` Shailabh Nagar
2006-07-01 3:51 ` Andrew Morton
2006-07-03 21:11 ` Shailabh Nagar
2006-07-03 21:41 ` Andrew Morton
2006-07-04 0:13 ` Shailabh Nagar
2006-07-04 0:38 ` Andrew Morton
2006-07-04 20:19 ` Paul Jackson
2006-07-04 20:22 ` Paul Jackson
2006-07-04 0:54 ` Shailabh Nagar
2006-07-04 1:01 ` Andrew Morton
2006-07-04 13:05 ` jamal
2006-07-04 15:18 ` Shailabh Nagar
2006-07-04 16:37 ` Shailabh Nagar
2006-07-04 19:24 ` jamal
2006-07-05 14:09 ` Shailabh Nagar
2006-07-05 20:25 ` Chris Sturtivant
2006-07-05 20:32 ` Shailabh Nagar
2006-07-03 4:53 ` Paul Jackson
2006-07-03 15:02 ` Shailabh Nagar
2006-07-03 15:55 ` Paul Jackson
2006-07-03 16:31 ` Paul Jackson
2006-07-04 0:09 ` Shailabh Nagar
2006-07-04 19:59 ` Paul Jackson
2006-07-05 17:20 ` Jay Lan
2006-07-05 18:18 ` Shailabh Nagar
2006-06-30 22:56 ` Andrew Morton
2006-06-29 18:05 ` Nick Piggin
2006-06-29 12:42 ` Shailabh Nagar
2006-06-24 3:08 ` Shailabh Nagar
2006-06-21 20:38 ` Andrew Morton
2006-06-21 21:31 ` Shailabh Nagar
2006-06-21 21:45 ` Jay Lan
2006-06-21 21:54 ` Andrew Morton
2006-06-21 22:19 ` Jay Lan
2006-06-21 21:59 ` Shailabh Nagar
2006-06-09 15:55 ` Chris Sturtivant
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=4489EE7C.3080007@watson.ibm.com \
--to=nagar@watson.ibm.com \
--cc=akpm@osdl.org \
--cc=balbir@in.ibm.com \
--cc=csturtiv@sgi.com \
--cc=jlan@sgi.com \
--cc=linux-kernel@vger.kernel.org \
/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