ok so I did some more browser testing, midori writes almost as bad as firefox and kazehakesa wasn't usable(kept segfaulting). But dillo2 was quit useable for general browsing(no flash or cookies) and it hardly ever wrote to disk. So my recommendation for laptop browsing is dillo2, it's not in the repos yet but if you go here http://misc.andi.de1.cc/dillo/ you can use the debian lenny deb. I'm also going to look into if the script runs on a distro like slitaz that loads entirely to ram to see what kind of powersaving you can get by not using the disk at all.
Through all this i have negected to mention that i am using reiser and not ext3 ... i think all the HDD settings for cache and writeback only effect ext3. Can you confirm that? If so I will be going back to it.
I'm using reiser as well(hate the filesystem checks on ext3) and they work for reiser. The dirty writeback and similar settings are not filesystem dependent, only thing that is filesystem dependent is journal commit and that does work on reiser.
Last edited by iggykoopa; December 3rd, 2008 at 03:03 PM.
ok great then .. i always hated that about ext3 .. plus reiser is just so much faster for moving large amounts of data around.
Btw:
python log.py
mail,news.none -/var/log/messages
@iggykoopa
python log.py
mail,news.none -/var/log/messages
you wrote wrong the acpi thermal polling code, it is THRM, instead of THM! =)
Ah thanks for the kernel tip, i dropped more .7 W with pcie_aspm module. Although I need to use pcie_aspm=force in kernel line.
Just a suggestion, but is there anyway to, instead of using that checkbox to say if it is detected or not, turn the checkbox inactive and the whole line becomes gray, non active and non editable?
Last edited by felipefoz; December 4th, 2008 at 02:05 AM. Reason: added suggestion
I might be totally off base but the settings for Dirty_background_ratio is a % from 0 -100 that tells the system how much memory it will fill before it writes the cache to disk .. in your setup you have 1 for power saving and 5 for performance.. wouldnt the exact oposite apply if you wished to keep the hdd as much as possible?
dirty ratio is what your thinking of for the percentage, I'll have to look into what the dirty background ratio is but the settings I have are what everyone else is using in their scripts. Also with the battery thermal management it's thm on my laptop, I thought you made the typo I guess I'll have to make it detect which is ok for your laptop then. thanks for pointing it out. The reason I didn't want to grey out a line if it isn't detected is incase any of my autodetection code didn't function correctly people could still override it easily. But if people would prefer it that way it wouldn't be too hard to change it. You would still be able to override the autodetection manually in the config file.
ok I've done some more testing and it seems like the script in it's current form is about as low as I'm going to be able to get on my machine, I'll still need to add the stuff on the last couple pages and any other stuff for other hardware. Anyway I've tested openbox instead of gnome, ext2 so theres no journal and a couple other things...it seems the lowest power usage I can eek out of a 1420 is about 9.8 watts, not bad but I was hoping I could do better. So now that I'm done with my testing for the moment I'll get back to coding, sorry for the delay on implementing the stuff on the last couple pages, just wanted to make sure there weren't any other recommendations I could make on configurations outside of the scope of the script. Also it doesn't seem like the effort of getting an os to run from ram is really worth it, seems like the powersaving you'll get from it is pretty minimal compared to the effort it takes. I may do a little more testing in that area once I'm caught up on the bugs you guys have given me so far though. again thanks for the help testing this, couldn't do it without you.
Bookmarks