How long unit the "~200 Line Linux Kernel Patch That Does Wonders" is applied to us?


second exodous

Advanced Member
Joined
Sep 27, 2005
Messages
2,974
Location
Utah, USA
When /. reports on kernel patches it is either something really good or really bad, but this one looks really good!


I was wondering if this would apply to us at all as it has to do with better multi-tasking. I hope it will help my tiny Atom PC I'm using to save power, thing slows to a crawl when I have too much open, have to close my browser to watch video through VLC or it is all pixeley.
 
There's a thread on the other forum about this which discusses this also - I'm just dropping a link here for reference, as its got some interesting posts in it.
 
Last edited by a moderator:
Long story short: it helps keep your desktop experience smooth even when the CPU is under heavy load. For example, if you are compiling something in the background but still want to be able to surf the web.


I don't think it will benefit the Pandora at all but may be useful, at some point in the future, as a kind of parachute just in case you do something that spins the CPU to 100% and you're having trouble navigating the desktop to kill it.
 
About the patch, it's quite obvious good for fixing the problem of heavy CPU background tasks killing the desktop performance. if it will help much for other use cases? prolly not if you hear the comments.


also interesting command:


nice -1 <pid of whatever runs fullscreen / requires your attention>


should be enough to hint the scheduler that VLC performance is what you want


For my own atom PC's, flashblock is the best software you could install


not doing file IO also helps alot on Atom CPU's (i suspect the same of the omap3)


a kernel tuned for speed, not for data reliability could be nice (more windows 95 like, less server OS) but this might sound like heresy to some (remember how fast staging ext4 was before people found out it could kill your data if you sneezed wrong? i say, don't sneeze, i want speed)
 
The patch, and also the alternative shell-only version, only affects processes with a tty. Which doesn't include e.g. graphical desktop tools. To quote one person on the kernel list:


"The kernel patch discussed here *has* *no* *relevance* for normal users.





The kernel patch discussed here is only relevant for people which start



mplayer from one terminal, and "make -j" from another."






End quote.
 
The patch, and also the alternative shell-only version, only affects processes with a tty. Which doesn't include e.g. graphical desktop tools. To quote one person on the kernel list:


"The kernel patch discussed here *has* *no* *relevance* for normal users.





The kernel patch discussed here is only relevant for people which start



mplayer from one terminal, and "make -j" from another."






End quote.
Yeah that's pretty accurate. It seems quite arbitrary to group processed by tty. I don't have anything against hierarchical scheduling but the example videos seem a bit contrived.


I would be kinda neat though if you could group processes manually for scheduling purposes. But I suppose this would rapidly start to resemble priority based scheduling instead.
 
Back
Top