[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180725143850.GA2886@bombadil.infradead.org>
Date: Wed, 25 Jul 2018 07:38:50 -0700
From: Matthew Wilcox <willy@...radead.org>
To: "Elliott, Robert (Persistent Memory)" <elliott@....com>
Cc: Cannon Matthews <cannonmatthews@...gle.com>,
Michal Hocko <mhocko@...nel.org>,
Mike Kravetz <mike.kravetz@...cle.com>,
Andrew Morton <akpm@...ux-foundation.org>,
"Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
"linux-mm@...ck.org" <linux-mm@...ck.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Andres Lagar-Cavilla <andreslc@...gle.com>,
Salman Qazi <sqazi@...gle.com>, Paul Turner <pjt@...gle.com>,
David Matlack <dmatlack@...gle.com>,
Peter Feiner <pfeiner@...gle.com>,
Alain Trinh <nullptr@...gle.com>
Subject: Re: [PATCH v2] RFC: clear 1G pages with streaming stores on x86
On Wed, Jul 25, 2018 at 05:02:46AM +0000, Elliott, Robert (Persistent Memory) wrote:
> Even with that, one CPU core won't saturate the memory bus; multiple
> CPU cores (preferably on the same NUMA node as the memory) need to
> share the work.
It's probably OK to not saturate the memory bus; it'd be nice if other
cores were allowed to get work done. If your workload is single-threaded,
you're right, of course, but who has a single-threaded workload these
days?!
Powered by blists - more mailing lists