On Mon, Jun 15, 2020 at 12:55:22PM -0700, Paul E. McKenney wrote:
> On Mon, Jun 15, 2020 at 09:11:58PM +0200, Peter Zijlstra wrote:
> > On Mon, Jun 15, 2020 at 10:21:49AM -0700, Paul E. McKenney wrote:
> > > On Mon, Jun 15, 2020 at 06:40:48PM +0200, Peter Zijlstra wrote:
> >
> > > > Thanks! I've got 16*TREE03 running since this morning, so far so
> > > > nothing :/
> > > > (FWIW that's 16/9 times overcommit, idle time fluctuates around 10%).
> > >
> > > My large system as large remote memory latencies, as in the better part
> > > of a microsecond. My small system is old (Haswell). So, just to grasp
> > > at the obvious straw, do you have access to a multisocket Haswell system?
> >
> > I've been running this on a 4 socket haswell ex.
>
> OK, I officially have no idea, then!
Ah, are you overcommitting? None of my Haswell runs did any overcommit.
Also, any failures in the meantime?
Any debug/tracing/whatever that I could add to my runs?
Thanx, Paul
