Better or advised configuration approach for Optimus based systems
ajax at redhat.com
Wed Nov 23 00:08:19 UTC 2011
On 11/22/11 7:29 AM, Gianluca Cecchi wrote:
> I have an Asus U36SD laptop, with Optimus technology.
> I'm not sure if my current approach is the correct/more safe one.
> I think more and more devices are going to have this technology, so
> this could be a cue to address future F17 potential users problems
I think they're not, to be entirely honest. The gap between Intel's
performance and everybody else's power consumption is narrowing, which
is the whole thing Optimus is trying to address.
But if you _want_ to buy a computer that's more expensive, more
complicated, and more difficult to use, I guess that's a thing you can do.
> So I'm using his acpi-call_240611.orig.tar.gz from ironhide (I have to
> check new versions), putting in /etc/rc.d/rc.local
> echo "Disabling Nvidia video adapter..." | tee -a /var/log/nvida_disabled.log
> /sbin/modprobe acpi_call
> echo '\_SB.PCI0.PEG0.GFX0.DOFF'> /proc/acpi/call
I don't expect we're ever going to see acpi_call merged in the kernel.
/proc/acpi is already a deprecated interface (because apparently it's
okay to remove useful functionality as long as it's because you don't
like it; I'm not bitter).
It would be more appropriate to get the nouveau driver to power the
device off when not active.
> What about i915_enable_rc6 not being the default...? Does still put
> stability problems? Does it make sense to enable it or not by default
> based on different adapters/chipsets?
> In my case it gives me about 1hour and a half of battery life more....
Yes, it has stability problems. It'll be the default when upstream
thinks it works (and if I don't see regressions as a result).
> BTW: is the Intel driver in Fedora 16 currently built with
> "--enable-sna" option or planned to be ?
Not at the moment, and I'm not sure when or if I'll change that. I'm
tempted to switch it on in rawhide for a while and see what complaints I
get, but I don't really have the time to tend to the fallout from it in
the next... month or so?
Part of this is I don't think it's possible to build the driver with
both uxa and sna, which I would really really want for debugging while
the default is changed.
More information about the test