* Jiri Olsa <[email protected]> wrote:
> On Tue, Oct 24, 2017 at 02:59:44PM +0200, Ingo Molnar wrote:
> >
> > * Jiri Olsa <[email protected]> wrote:
> >
> > > I recently made some changes on threaded record, which are based
> > > on Namhyungs time* API, which is needed to read/sort the data afterwards
> > >
> > > but I wasn't able to get any substantial and constant reduce of LOST
> > > events
> > > and then I got sidetracked and did not finish, but it's in here:
> >
> > So, in the context of system-wide profiling, the way that would work best I
> > think
> > is the following:
> >
> > thread #0 binds itself to CPU#0 (via sched_setaffinity) and creates a
> > per-CPU event on CPU#0
> > thread #1 binds itself to CPU#1 (via sched_setaffinity) and creates a
> > per-CPU event on CPU#1
> > thread #2 binds itself to CPU#2 (via sched_setaffinity) and creates a
> > per-CPU event on CPU#2
> >
> > etc.
> >
> > Is this how you implemented it?
>
> in a way ;-) but I made it more generic and let record create just
> few threads and let them share cpu subset.. and so there was no binding
>
> >
> > If the threads in the thread pool are just free-running then the scheduler
> > might
> > not migrate it to the 'right' CPU that is streaming the perf events and
> > there will
> > be a lot of cross-talking between CPUs.
>
> ok it's easy to add binding now and 1:1 thread:cpu mapping.. I'll retry
Please Cc: me - this is a really interesting aspect of perf scalability!
Thanks,
Ingo
