[Intel-gfx] [PATCH 1/2] drm/i915: Boost RPS frequency for CPU stalls

[Meng, Mengmeng]

Hi,

It's Power results on IVB mobile. With the patches, workloads would cost more power.
mobile IVB

workloads

git-53749d3

git-53749d3 with four patches

Patches VS no Patch

Average watt

Watt-hour

Time

Average watt

Watt-hour

Time

Average watt

Watt-hour

idle

25.6

3.7

9′

25.7

3.8

9′

100.39%

102.70%

video

30.7

2.7

5'30"

30.9

2.7

5'30"

100.65%

100.00%

game

33.5

6.4

11'43"

33.9

6.5

11'41"

101.19%

101.56%




Thanks,

Meng





-----Original Message-----
From: Chris Wilson [mailto:chris at chris-wilson.co.uk]
Sent: Tuesday, August 27, 2013 6:49 PM
To: Meng, Mengmeng
Cc: Kenneth Graunke; Stéphane Marchesin; Owen Taylor; Zhuang, Lena; Zhang, Ouping; Jesse Barnes; intel-gfx at lists.freedesktop.org
Subject: Re: [PATCH 1/2] drm/i915: Boost RPS frequency for CPU stalls



On Tue, Aug 27, 2013 at 10:12:29AM +0000, Meng, Mengmeng wrote:

>    Hi,

>

>    We test performance of this patch set compared to the current mainline

>    without these patches applied.

>

>    With patches v.s. the current mainline, most of performance change

>    little(<2%) except GL benchmark 2.7.0(↑5%). You may see results attached.



That looks like the range I found locally, and nice that there is no shocking regression. Can you schedule some power testing as well?



Thanks,

-Chris



--

Chris Wilson, Intel Open Source Technology Centre
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20130829/7a0585a2/attachment.html>