xfs
[Top] [All Lists]

Re: [RFC PATCH-tip 6/6] xfs: Enable reader optimistic spinning for DAX i

To: Waiman Long <Waiman.Long@xxxxxxx>
Subject: Re: [RFC PATCH-tip 6/6] xfs: Enable reader optimistic spinning for DAX inodes
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Tue, 14 Jun 2016 11:24:23 -0700
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>, Ingo Molnar <mingo@xxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxx, x86@xxxxxxxxxx, linux-alpha@xxxxxxxxxxxxxxx, linux-ia64@xxxxxxxxxxxxxxx, linux-s390@xxxxxxxxxxxxxxx, linux-arch@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx, Davidlohr Bueso <dave@xxxxxxxxxxxx>, Jason Low <jason.low2@xxxxxx>, Dave Chinner <david@xxxxxxxxxxxxx>, Scott J Norton <scott.norton@xxxxxxx>, Douglas Hatch <doug.hatch@xxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <1465927959-39719-7-git-send-email-Waiman.Long@xxxxxxx>
References: <1465927959-39719-1-git-send-email-Waiman.Long@xxxxxxx> <1465927959-39719-7-git-send-email-Waiman.Long@xxxxxxx>
User-agent: Mutt/1.5.24 (2015-08-30)
On Tue, Jun 14, 2016 at 02:12:39PM -0400, Waiman Long wrote:
> This patch enables reader optimistic spinning for inodes that are
> under a DAX-based mount point.
> 
> On a 4-socket Haswell machine running on a 4.7-rc1 tip-based kernel,
> the fio test with multithreaded randrw and randwrite tests on the
> same file on a XFS partition on top of a NVDIMM with DAX were run,
> the aggregated bandwidths before and after the patch were as follows:

And why is this specific to DAX?  Many I/O operations already never
got out to disk, and ilock is mostly held for operations that have
nothing to do with disk I/O.

<Prev in Thread] Current Thread [Next in Thread>