[MMTests] IO metadata on XFS

[Mel Gorman]

On Tue, Jul 03, 2012 at 02:31:19PM +0200, Daniel Vetter wrote:
> On Tue, Jul 03, 2012 at 11:59:51AM +0100, Mel Gorman wrote:
> > On Tue, Jul 03, 2012 at 10:19:28AM +1000, Dave Chinner wrote:
> > > On Mon, Jul 02, 2012 at 08:35:16PM +0100, Mel Gorman wrote:
> > > > Adding dri-devel and a few others because an i915 patch contributed to
> > > > the regression.
> > > > 
> > > > On Mon, Jul 02, 2012 at 03:32:15PM +0100, Mel Gorman wrote:
> > > > > On Mon, Jul 02, 2012 at 02:32:26AM -0400, Christoph Hellwig wrote:
> > > > > > > It increases the CPU overhead (dirty_inode can be called up to 4
> > > > > > > times per write(2) call, IIRC), so with limited numbers of
> > > > > > > threads/limited CPU power it will result in lower performance. Where
> > > > > > > you have lots of CPU power, there will be little difference in
> > > > > > > performance...
> > > > > > 
> > > > > > When I checked it it could only be called twice, and we'd already
> > > > > > optimize away the second call.  I'd defintively like to track down where
> > > > > > the performance changes happend, at least to a major version but even
> > > > > > better to a -rc or git commit.
> > > > > > 
> > > > > 
> > > > > By all means feel free to run the test yourself and run the bisection :)
> > > > > 
> > > > > It's rare but on this occasion the test machine is idle so I started an
> > > > > automated git bisection. As you know the milage with an automated bisect
> > > > > varies so it may or may not find the right commit. Test machine is sandy so
> > > > > http://www.csn.ul.ie/~mel/postings/mmtests-20120424/global-dhp__io-metadata-xfs/sandy/comparison.html
> > > > > is the report of interest. The script is doing a full search between v3.3 and
> > > > > v3.4 for a point where average files/sec for fsmark-single drops below 25000.
> > > > > I did not limit the search to fs/xfs on the off-chance that it is an
> > > > > apparently unrelated patch that caused the problem.
> > > > > 
> > > > 
> > > > It was obvious very quickly that there were two distinct regression so I
> > > > ran two bisections. One led to a XFS and the other led to an i915 patch
> > > > that enables RC6 to reduce power usage.
> > > > 
> > > > [aa464191: drm/i915: enable plain RC6 on Sandy Bridge by default]
> > > 
> > > Doesn't seem to be the major cause of the regression. By itself, it
> > > has impact, but the majority comes from the XFS change...
> > > 
> > 
> > The fact it has an impact at all is weird but lets see what the DRI
> > folks think about it.
> 
> Well, presuming I understand things correctly the cpu die only goes into
> the lowest sleep state (which iirc switches off l3 caches and
> interconnects) when both the cpu and gpu are in the lowest sleep state.

I made a mistake in my previous mail. gdm and X were were *not* running.
Once the screen blanked I would guess the GPU is in a low sleep state
the majority of the time.

> rc6 is that deep-sleep state for the gpu, so without that enabled your
> system won't go into these deep-sleep states.
> 
> I guess the slight changes in wakeup latency, power consumption (cuts
> about 10W on an idle desktop snb with resulting big effect on what turbo
> boost can sustain for short amounts of time) and all the follow-on effects
> are good enough to massively change timing-critical things.
> 

Maybe. How aggressively is the lowest sleep state entered and how long
does it take to exit?

> So this having an effect isn't too weird.
> 
> Obviously, if you also have X running while doing these tests there's the
> chance that the gpu dies because of an issue when waking up from rc6
> (we've known a few of these), but if no drm client is up, that shouldn't
> be possible. So please retest without X running if that hasn't been done
> already.
> 

Again, sorry for the confusion but the posted results are without X running.

-- 
Mel Gorman
SUSE Labs