[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20161024185321.GC5136@kernel.org>
Date: Mon, 24 Oct 2016 15:53:21 -0300
From: Arnaldo Carvalho de Melo <acme@...nel.org>
To: Davidlohr Bueso <dbueso@...e.de>
Cc: Ingo Molnar <mingo@...nel.org>, linux-kernel@...r.kernel.org,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
Peter Zijlstra <peterz@...radead.org>,
Arnaldo Carvalho de Melo <acme@...hat.com>
Subject: Re: [PATCH 33/37] perf bench futex: Cache align the worker struct
Em Mon, Oct 24, 2016 at 11:50:50AM -0700, Davidlohr Bueso escreveu:
> On 2016-10-24 09:20, Arnaldo Carvalho de Melo wrote:
> > From: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
> >
> > It popped up in perf testing that the worker consumes some amount of
> > CPU. It boils down to the increment of `ops` which causes cache line
> > bouncing between the individual threads.
> >
> > This patch aligns the struct by 256 bytes to ensure that not a cache
> > line is shared among CPUs. 128 byte is the x86 worst case and grep says
> > that L1_CACHE_SHIFT is set to 8 on s390.
>
> Hmm so this one should have been replaced with this:
>
> https://marc.info/?l=linux-kernel&m=147690008801082
Humm, sorry, can you please send a patch fixing this?
- Arnaldo
Powered by blists - more mailing lists