Making my laptop quiet
Published on Wednesday, the 8th of September, 2010 and last updated on the 24th of November, 2011
A few days ago I talked about how noisy my new Dell Studio 15 was but I can now report back with the beginnings of a solution to that problem, and it doesn't appear to be ACPI related.The first clue I needed was that if I switch to the proprietary fglrx drivers for the Radeon the fan quite quickly drops off to a much more reasonable level. It seems the fglrx drivers have issues, however, in particular I get big black patches on my screen. This video corruption happens especially in Firefox, but sometimes in other applications as well. They also appear to screw up my suspend/resume, which is probably even more annoying to me.The second clue that I needed was that Radeon power management support has only just made it into recent kernels. Thanks to Michael Kirkland for providing me with both of those clues :-)Looking under /sys/class/drm/ I find a whole bunch of stuff, but in particular there are /sys/class/drm/card0/device/power_method and /sys/class/drm/card0/device/power_profile.Looking through the kernel source code I can see that power_profile can be set to low, mid, high, auto and default, while power_method can be set to either dynpm or profile.Trying out all of these values, it seems I get the quietest result with the profile method and either the low or mid profile. The dynpm method is nearly as good, and I would think it should really be the default for a 'Mobility' chipset. From the detailed benchmarking that Phoronix did I wonder if it shouldn't be the default for everyone.For myself, I see some small 'tearing' artifacts occasionally when running with the low profile. These disappear when I run with the mid profile, and since that seems to have pretty much the same temperature (and noise) results I'll go with that one. Though the laptop often does still make more noise than I would prefer it to, it is no longer annoying everyone in the room.Not unexpectedly this seems to have a huge impact on power use, too. It appears that the laptop should now give me around 4.5 hours when I do everything I can think of to lower the power use, whereas before it was more like 2.5 hours.Now I guess I can get back to hacking on DAViCal...
A few days ago I talked about how noisy my new Dell Studio 15 was but I can now report back with the beginnings of a solution to that problem, and it doesn't appear to be ACPI related.
The first clue I needed was that if I switch to the proprietary fglrx drivers for the Radeon the fan quite quickly drops off to a much more reasonable level. It seems the fglrx drivers have issues, however, in particular I get big black patches on my screen. This video corruption happens especially in Firefox, but sometimes in other applications as well. They also appear to screw up my suspend/resume, which is probably even more annoying to me.
The second clue that I needed was that Radeon power management support has only just made it into recent kernels. Thanks to Michael Kirkland for providing me with both of those clues :-)
Looking under /sys/class/drm/ I find a whole bunch of stuff, but in particular there are /sys/class/drm/card0/device/power_method and /sys/class/drm/card0/device/power_profile.
Looking through the kernel source code I can see that power_profile can be set to low, mid, high, auto and default, while power_method can be set to either dynpm or profile.
Trying out all of these values, it seems I get the quietest result with the profile method and either the low or mid profile. The dynpm method is nearly as good, and I would think it should really be the default for a 'Mobility' chipset. From the detailed benchmarking that Phoronix did I wonder if it shouldn't be the default for everyone.
For myself, I see some small 'tearing' artifacts occasionally when running with the low profile. These disappear when I run with the mid profile, and since that seems to have pretty much the same temperature (and noise) results I'll go with that one. Though the laptop often does still make more noise than I would prefer it to, it is no longer annoying everyone in the room.
Not unexpectedly this seems to have a huge impact on power use, too. It appears that the laptop should now give me around 4.5 hours when I do everything I can think of to lower the power use, whereas before it was more like 2.5 hours.
Now I guess I can get back to hacking on DAViCal...
Flattr
![[D]](/images/tag_delicious.png "Submit to Del.icio.us"){kind=small}
![[Digg]](/images/tag_digg.png "Submit to Digg"){kind=small}
![[FB]](/images/tag_facebook.png "Set status on Facebook"){kind=small}
![[R]](/images/tag_reddit.png "Submit to Reddit"){kind=small}
![[SU]](/images/tag_stumbleupon.png "Submit to Stumble Upon"){kind=small}
![[Tweet]](/images/tag_twitter.png "Set status on Twitter"){kind=small}
I played with fglrx some time
I played with fglrx some time ago (now I use open drivers, as AMD dropped support for hardware was only bought only 2 years earlier...), here is my enhanced script for fglrx http://rozie.jogger.pl/2008/11/21/linux-sterowanie-zarzadzaniem-acpi-i-fglrx/ (use Google translate, please) - original had some bugs with resuming after opening lid. Don't know if it still is helpful, but just in case...
Support from AMD for fglrx drivers
I think that AMD probably want to get rid of the fglrx drivers entirely, and dropping support for the r500 chipset is likely a part of that. The work that they are currently putting into making the open drivers work properly is probably another part.
I think it's great to see AMD putting increasing effort into the open drivers and will be only too happy if the fglrx ones disappear entirely.
Thanks for the link for people wanting a script to help suspend/resume with them though :-)
Cheers,
Andrew.