[Intel-gfx] Framebuffer compression on GM45 disabled with KMS

[Pedro Ribeiro]

On 3 May 2010 14:53, Pedro Ribeiro <pedrib at gmail.com> wrote:
> On 3 May 2010 08:18, Tino Keitel <tino.keitel+xorg at tikei.de> wrote:
>> On Mon, May 03, 2010 at 00:35:34 +0100, Pedro Ribeiro wrote:
>>
>> [...]
>>
>>> Hi Tino,
>>>
>>> this gains have been mostly with the computer sitting idle and me not
>>> touching it. The biggest gain (over 1w) is when I let the monitor go
>>> off into powersaving mode. There is a clear difference here. But maybe
>>> its just for my chipset...
>>
>> I only test with the LCD switched on, but set to lowest brightness.
>>
>> In the previous test, I had wrong kernel options. The i915. prefix was
>> missing.  Now I tested again and got down to 7.0W:
>>
>> Wakeups-from-idle per second :  8.0     interval: 30.0s
>> Power usage (ACPI estimate): 7.0W (3.9 hours) (long term: 7.8W,/3.4h)
>>
>> I also changed the background image to solid color for better FBC. What
>> I noticed during my tests is that I only get such low numbers after a
>> resume from suspend to RAM.  Power consumption is ~0.5W higher after a
>> resume from suspend to disk using TuxOnIce.  IMHO this should not
>> happen.  So, overall I get lower values with the current configuration,
>> but I don't know what change gives the power saving, and the suspend
>> issue makes it hard to compare.  It could also be a combination of some
>> of the configuration changes.
>>
>> Regards,
>> Tino
>> _______________________________________________
>> Intel-gfx mailing list
>> Intel-gfx at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>>
>
> That's strange... I do not notice any difference if I hibernate with
> TuxOnIce (which I do all the time).
> Sometimes there is a 2w increase after a hibernate/resume cycle, but
> that is due to the flash plugin, which behaves badly with hibernate
> (its a piece of crap anyway). Killing the flash process in makes the
> power consumption go down again.
>
> Regards,
> Pedro
>

Haven't had any problems at all with this patch reversion.

Jesse, did you receive any bug reports back when render reclock was
enabled relating to the GM45 platform? If not, it might not be a bad
idea to enable it again. Its been a bliss for me.

Regards,
Pedro