Messages in this thread | | | Date | Tue, 29 Jan 2013 04:29:52 +0100 | Subject | Re: [GIT PULL] cputime: Full dynticks task/cputime accounting v7 | From | Frederic Weisbecker <> |
| |
2013/1/29 Steven Rostedt <rostedt@goodmis.org>: > On Tue, 2013-01-29 at 03:39 +0100, Frederic Weisbecker wrote: > >> Please send me your kernelspin so I can test as well. > > http://rostedt.homelinux.com/private/kernelspin.c
I can't reproduce your issue. I'm getting almost the same results with vtime + force context tracking and tick based cputime.
3 runs with tick based cputime:
real 0m10.005s user 0m0.210s sys 0m9.780s
real 0m10.005s user 0m0.170s sys 0m9.820s
real 0m10.005s user 0m0.130s sys 0m9.860s
3 runs with vtime + context tracking forced:
real 0m10.005s user 0m0.807s sys 0m9.197s
real 0m10.005s user 0m0.832s sys 0m9.172s
real 0m10.005s user 0m0.787s sys 0m9.218s
> http://rostedt.homelinux.com/private/userspin.c > > -- Steve > >
| |