[Pm-utils] powertop suggestions

[Stefan Seyfried]

On Fri, Jun 29, 2007 at 10:23:03AM -0700, Arjan van de Ven wrote:
> Stefan Seyfried wrote:
> >Holy cow, NO!
> >Laptop mode kills performance badly, even the performance of a NFS mount.
> 
> do you have any real data on this?
> (and I mean on JUST kernel laptop mode; not any of the other stuff the laptop
> mode scripts poke at)

It was some kernels ago, when i wondered wy my copying from / to NFS was
only running at less than 5MB/second. I then did "echo 0 > laptop_mode"
(was set to 1 or 2) and it immediately increased to the usual >10MB/sec.

The dirty_ratio stuff etc. was kept unchanged.
I need to retry this with a recent kernel.

> >And with my measurements showing that disk spindown has more psychological
> >than real powersaving effects on modern laptop drives, i'd say we should
> >not do it unless the users asks for "kill my performance badly for a few
> >minutes more battery lifetime".
> 
> laptop mode does not spin down the disk.

But usually people use it to get their disk to actually spin down.

> laptop mode will save real power with ALPM; keeping the sata link idle longer
> helps; an idle sata link saves about 1W.

This might be true, i did not think about SATA. 1W from the controller is
twice as much than what you can get by spinning down the disk.

So if it can actually be useful, i need to reevaluate the performance
impact i had seen.

-- 
Stefan Seyfried
QA / R&D Team Mobile Devices        |              "Any ideas, John?"
SUSE LINUX Products GmbH, Nürnberg  | "Well, surrounding them's out." 

This footer brought to you by insane German lawmakers:
SUSE Linux Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)