new exciting features in BFS (making desktop interactive at any load)

archman-cro · 2010-10-08 20:42:21

patching file include/linux/sched.h
Hunk #1 FAILED at 1192.
1 out of 1 hunk FAILED -- saving rejects to file include/linux/sched.h.rej
patching file kernel/sched_bfs.c
Hunk #2 FAILED at 644.
Hunk #3 FAILED at 948.
Hunk #4 FAILED at 968.
Hunk #5 FAILED at 979.
Hunk #6 succeeded at 1594 (offset 29 lines).
Hunk #7 succeeded at 2494 (offset 29 lines).
Hunk #8 FAILED at 2488.
Hunk #9 FAILED at 3443.
Hunk #10 FAILED at 3456.
7 out of 10 hunks FAILED -- saving rejects to file kernel/sched_bfs.c.rej
patching file kernel/sysctl.c
patching file include/linux/init_task.h
    Aborting...

...while applying "bfs357-test.patch" after "2.6.35.7-sched-bfs357+schedule_groups_as_entity-1.patch".
If I apply only the latest patch, it asks for 2.6.36...

Update 1: It works like this:

1. Apply "2.6.35.7-sched-bfs-357.patch"
2. Apply "bfs357-penalise_fork_depth_account_threads.patch"

Builds good like that on 2.6.35.

Update 2: Con has just released a "combo" of those two I wrote above. Testing it out.

Last edited by archman-cro (2010-10-08 21:57:02)

jarryson · 2010-10-09 08:45:25

Further investigation revealed this patch didn't work quite how I thought it did, but still offers significant advantages. See the newer posts on hierarchical tree based penalty.

so i think better to use "bfs357-penalise_fork_depth_account_threads.patch"

in /proc/sys/kernel/
group_thread_accounting - groups CPU accounting by threads
fork_depth_penalty - penalises according to depth of forking from init

seems this patch contain functions of "2.6.35.7-sched-bfs357+schedule_groups_as_entity-1.patch"

jarryson · 2010-10-09 13:10:08

i've tried both of them and i can't tell any difference.

the browser got the same speed no matter i was compiling something or not.

but i got some really wired problem with the these patches. such as gnome-panel always can not load all applets successfully, some applets can not loaded readomly.

it looks like panel itself do not wait applets to load. that's not acceptable. so i have to fallback to bfs357 without any of these two patches.

Last edited by jarryson (2010-10-09 13:11:05)

graysky · 2010-10-09 13:27:06

jarryson wrote:

i've tried both of them and i can't tell any difference.
the browser got the same speed no matter i was compiling something or not.
but i got some really wired problem with the these patches. such as gnome-panel always can not load all applets successfully, some applets can not loaded readomly.
it looks like panel itself do not wait applets to load. that's not acceptable. so i have to fallback to bfs357 without any of these two patches.

+1 - had the same problem with fusion-icon. Dropped by to ck1 + bfs357.

Lothium · 2010-10-09 13:38:07

I get the problem, that the desktop sometimes don't react to the right click of the mouse. For exampel when I try to open a link in chromium in a new tab, I click with the right mouse button on a link, but the menu don't appear or sometimes it appears after a few seconds. Very strange!

broch · 2010-10-09 17:43:19

maybe notify C.K. about the issue?

graysky · 2010-10-10 08:48:51

I just emailed him with the details...

jarryson · 2010-10-10 14:41:28

according to his lastest blog post( i didn't read it carefully), seems the problem i met must be fixed by the application itself. so i have to report to upstream, am i right?

Lothium · 2010-10-11 14:22:27

Mmh, strange. Two days ago I tried the patch with the zen kernel and I had a few problems (already talked about a few comments earlyer). Now I tried the newest pf-kernel with this patch and my system runs very snappy. Think I misconfigured anything with the zen-kernel.

akurei · 2010-10-20 16:16:56

SanskritFritz wrote:

akurei wrote:
Using kernel26zen-git.
What I'm doing now: Building kernel26zen-git (again), qt, kdelibs, kdebase-workspace, kdeplasma-applets-kdenetworkmanager via "yaourt -Sb". All simultaneously.
top - 16:27:15 up 38 min,  2 users,  load average: 8.17, 8.44, 7.42
Desktop is reacting as if there's no load at all. I love it!
Are you sure the new BFS 357 is included? I compiled the kernel from the AUR, but sched.h says
printk(KERN_INFO"BFS CPU scheduler v0.323 by Con Kolivas.\n");

Maybe you did not pull from git? Mine is:

printk(KERN_INFO"BFS CPU scheduler v0.357 by Con Kolivas.\n");

SanskritFritz · 2010-10-20 18:30:17

I did, but first I compiled the experimental branch, and later it turned out, it is in the stable branch... not logical, but hey.

Arch Linux

#26 2010-10-08 20:42:21

Re: new exciting features in BFS (making desktop interactive at any load)

#27 2010-10-09 08:45:25

Re: new exciting features in BFS (making desktop interactive at any load)

#28 2010-10-09 13:10:08

Re: new exciting features in BFS (making desktop interactive at any load)

#29 2010-10-09 13:27:06

Re: new exciting features in BFS (making desktop interactive at any load)

#30 2010-10-09 13:38:07

Re: new exciting features in BFS (making desktop interactive at any load)

#31 2010-10-09 17:43:19

Re: new exciting features in BFS (making desktop interactive at any load)

#32 2010-10-10 08:48:51

Re: new exciting features in BFS (making desktop interactive at any load)

#33 2010-10-10 14:41:28

Re: new exciting features in BFS (making desktop interactive at any load)

#34 2010-10-11 14:22:27

Re: new exciting features in BFS (making desktop interactive at any load)

#35 2010-10-20 16:16:56

Re: new exciting features in BFS (making desktop interactive at any load)

#36 2010-10-20 18:30:17

Re: new exciting features in BFS (making desktop interactive at any load)

Board footer