Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Talos Improvement: Ts, MAX Dirty Profile decrease 13.98% on Leopard Firefox3.6

0 views
Skip to first unread message

Nobody

unread,
Nov 19, 2009, 12:31:49 AM11/19/09
to dev-tree-...@lists.mozilla.org

Improvement: Ts, MAX Dirty Profile decrease 13.98% on Leopard Firefox3.6

Previous results: 1119.47 from build 20091118071333 of revision 488074886732 at 2009-11-18 09:11:00 on talos-rev2-leopard03 run # 0

New results: 962.947 from build 20091118080727 of revision a0c8439c621d at 2009-11-18 08:46:00 on talos-rev2-leopard13 run # 0

Suspected checkin range: from revision 488074886732 to revision a0c8439c621d

dietrich ayala

unread,
Nov 19, 2009, 1:17:17 AM11/19/09
to Blake Kaplan, Robert Sayre, Graydon Hoare
This is nice number. However, the range doesn't have anything that
stands out as being responsible. Cc'ing folks in range.

Looks like there was also a big win on Linux for this same test, and
some level of win registering on the Medium test as well.

Windows seems flat, other than a remarkable removal of noise on Vista:
http://tinyurl.com/2vr3pz.

On 11/18/2009 9:31 PM, Nobody wrote:
> Improvement: Ts, MAX Dirty Profile decrease 13.98%

> <http://graphs.mozilla.org/graph.html#tests=[{"machine":171,"test":54,"branch":10},{"machine":178,"test":54,"branch":10},{"machine":180,"test":54,"branch":10},{"machine":174,"test":54,"branch":10},{"machine":175,"test":54,"branch":10},{"machine":182,"test":54,"branch":10},{"machine":181,"test":54,"branch":10},{"machine":170,"test":54,"branch":10},{"machine":177,"test":54,"branch":10},{"machine":176,"test":54,"branch":10},{"machine":179,"test":54,"branch":10},{"machine":173,"test":54,"branch":10},{"machine":172,"test":54,"branch":10},{"machine":169,"test":54,"branch":10}]&sel=1258476360,1258649160>


> on Leopard Firefox3.6
>
> Previous results: 1119.47 from build 20091118071333 of revision
> 488074886732 at 2009-11-18 09:11:00 on talos-rev2-leopard03 run # 0
>
> New results: 962.947 from build 20091118080727 of revision a0c8439c621d
> at 2009-11-18 08:46:00 on talos-rev2-leopard13 run # 0
>
> Suspected checkin range: from revision 488074886732 to revision
> a0c8439c621d <

> http://hg.mozilla.org/releases/mozilla-1.9.2/pushloghtml?fromchange=488074886732&tochange=a0c8439c621d>
>

Taras Glek

unread,
Nov 26, 2009, 3:32:49 PM11/26/09
to dietrich ayala, Blake Kaplan, Robert Sayre, Graydon Hoare
On 11/18/2009 10:17 PM, dietrich ayala wrote:
> This is nice number. However, the range doesn't have anything that
> stands out as being responsible. Cc'ing folks in range.
>
> Looks like there was also a big win on Linux for this same test, and
> some level of win registering on the Medium test as well.
>
> Windows seems flat, other than a remarkable removal of noise on Vista:
> http://tinyurl.com/2vr3pz.

If cold startup is changing on 2 platforms and warm isn't, my guess is
that this is due to binary layout changing resulting in less disk seeks.
OSX and Linux are both GCC platforms, so resulting binaries may be
remotely similar.
This may be "proof" that we need to reorder the way our binaries are
laid out in release builds.

Ie something like https://bugzilla.mozilla.org/show_bug.cgi?id=525013#c20

0 new messages