apm vs. acpi

[Matthew Saltzman]

On Wed, 29 Dec 2004, Ed Hill wrote:

> On Wed, 2004-12-29 at 08:48 -1000, Amy M wrote:
> > All things considered, this fedora-list has become perhaps one of the
> > best forums to learn Linux.  Our sincere thanks to those who are
> > gracious enough to donate their time answering questions that sometimes
> > may seem ridiculous.
> >
> > My question now is: From what I have read, it appears that if I want to
> > do suspend to ram, I have to forgo acpi and use apm.  Performance-wise,
> > would anyone care to comment on the pros and cons of acpi vis-a-vis
> > apm?  Thanks again.

Definitely system-dependent.  Try www.linux-on-laptops.com to see what
others have experienced.

>
> Hi Amy,
>
> I can only relate my FC kernel-2.6 experiences for one laptop, a
> ThinkPad A22p (PIII-900) and they are:
>
>  - with very recent kernels (eg. 2.6.9-1.681_FC3) both APM
>    and ACPI suspend-to-RAM work
>  - with both I often need to unload and then re-load the
>    sound kernel modules
>  - APM:
>    - uses *very* little power when suspended to RAM (lasts
>      for many days starting from a full charge)
>    - can occasionally have problems with pcmcia (even when
>      all cards are removed) so I usually use:
>        "/etc/init.d/pcmcia stop ; apm -s"
>      and then restart pcmcia after wake-up with:
>        "/etc/init.d/pcmcia start"

     - Cycles my miniPCI wireless card, so that resuming puts me on the
       network where I am, even if I've changed location.

>    - results in lockups about once every 40--50 suspend-
>      resume cycles
>  - ACPI:
>    - has no apparent problems with pcmcia
>    - experienced no lockups (in about ~60 cycles)
>    - uses a *LOT* (perhaps as much as 10X) more battery
>      power while suspended to RAM

Known bug affecing Thinkpads.  See Red Hat Bugzilla 142928, which points
to http://bugzilla.kernel.org/show_bug.cgi?id=3022.  Other systems may not
be affected to this extent.

>    - suspends and resumes very quickly
>    - routinely gives a kernel error on wakeup saying
>      something about interruptable_sleep() but they seem
>      to be harmless

     - Does not restart my miniPCI network card.

>
> Having done many hundreds of suspend-resume cycles using both APM and
> ACPI, I've decided to stick with APM to reduce the battery usage.

Me too (Thinkpad T41), but I'm hoping to see the fix sometime soon.  The
suspend/resume cycle is much faster with ACPI.

>
> Ed
>
>

-- 
		Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs