Messages in this thread | | | Date | Thu, 25 Apr 2013 14:23:36 -0700 | From | "Paul E. McKenney" <> | Subject | Re: [PATCH documentation 2/2] kthread: Document ways of reducing OS jitter due to per-CPU kthreads |
| |
On Thu, Apr 25, 2013 at 10:59:05PM +0200, Thomas Gleixner wrote: > On Thu, 25 Apr 2013, Paul E. McKenney wrote: > > On Thu, Apr 25, 2013 at 12:23:12PM +0200, Borislav Petkov wrote: > > > On Mon, Apr 22, 2013 at 09:03:29PM -0700, Paul E. McKenney wrote: > > > > > > +Name: ehca_comp/%u > > > > > > +Purpose: Periodically process Infiniband-related work. > > > > > > +To reduce corresponding OS jitter, do any of the following: > > > > > > +1. Don't use EHCA Infiniband hardware. This will prevent these > > > > > > > > > > Sounds like this particular hardware is slow and its IRQ handler/softirq > > > > > needs a lot of time. Yes, no? > > > > > > > > > > Can we have a reason why people shouldn't use that hw. > > > > > > > > Because it has per-CPU kthreads that can cause OS jitter. ;-) > > > > > > Yeah, I stumbled over this specific brand of Infiniband hw. It looks > > > like this particular Infiniband driver uses per-CPU kthreads and the > > > others in drivers/infiniband/hw/ don't? > > > > > > I hope this explains my head-scratching moment here... > > > > Ah! I rewrote the first sentence to read: > > > > Don't use eHCA Infiniband hardware, instead choosing hardware > > that does not require per-CPU kthreads. > > Another option would be to teach that eHCA driver to be configurable > on which cpus kthreads are desired and on which not. I can't see a > reason (aside of throughput) why that hardware can't cope with a > single thread.
Good point! I have added a third item to the eHCA list:
Rework the eHCA driver so that its per-CPU kthreads are provisioned only on selected CPUs.
Thanx, Paul
| |