Re: [PATCH 0/2] prefer wakers -v2

[Mike Galbraith]

On Wed, 2009-01-14 at 12:39 +0100, Peter Zijlstra wrote:
> OK, so lets try this again...

I took tip out for a little test-drive.  Mysql/pgsql+oltp showed no
effect vs 29.git beyond noise level, but cache hot make -j8 kbuild times
in an nfs4 mount show marked improvement.

2.6.29.git
real    6m29.291s
user    10m20.355s
sys     2m1.288s

2.6.29.tip
real    4m30.160s
user    10m26.115s
sys     2m3.088s

Wire saturates at ~4m, so there's still some room for improvement.

On the interactivity front, tip feels snappier, particularly when IO is
going on.  One place that this is quite noticeable is reading lkml with
evolution while building a new kernel.  I generally keep thread count
low when using evolution because IO makes it sluggish when deleting and
moving on to the next message.  With tip, the evolution experience is
much improved.

	-Mike

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/