You are not logged in.
Just as an FYI to fellow bfs junkies: CK has released a BFS patch for v0.363 that is optimized for per-cpu frequency scaling. See CK's blog entry:
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
I tought that he left kernel developement?,no?
Anyway, so far I haven't tried BFS aldough I heard that it gives better preformance on desktops...but is that true even after 2.6.38...not sure.
Offline
Left kernel dev? No, just was focusing on his compression util temporarily. If you read his blogs, CK doesn't think 'the patch that does wonders' does shit under sane computing conditions... who runs make -j100 and why optimize scheduling for insane circumstances anyway?! Quad core = make -j4; dual core = make -j2, etc.
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
Awesome, everything I need is up to 2.6.38 now: BFQ + CK + BFS + LOOP-AES. Some patches were holding me back from updating.. Had to wait on BFQ and got impatient and stopped checking for updates. This post spurred me to check and thankfully it's there, also glad I waited for this BFS test as hopefully it will help on my quad-core AMD. Thanks for posting -- time to make a permanent upgrade from .36, finally (.37 was awful).
btw, a rule of thumb I always went by as far as `make -j' is cpu count + 1, as demonstrated by this ugly bash line:
# echo /sys/devices/system/cpu/cpu[^^] + | wc -w
Offline
I take it this will find it's way into kernel26-ck 2.6.38? *Fingers Crossed*
Offline
Awesome, everything I need is up to 2.6.38 now: BFQ + CK + BFS + LOOP-AES. Some patches were holding me back from updating.. Had to wait on BFQ and got impatient and stopped checking for updates. This post spurred me to check and thankfully it's there, also glad I waited for this BFS test as hopefully it will help on my quad-core AMD. Thanks for posting -- time to make a permanent upgrade from .36, finally (.37 was awful).
btw, a rule of thumb I always went by as far as `make -j' is cpu count + 1, as demonstrated by this ugly bash line:
# echo /sys/devices/system/cpu/cpu[^^] + | wc -w
OT: If you want bfq, I fixed it for 2.6.38.
And uhh...
echo $(( $(nproc) + 1 ))
Last edited by falconindy (2011-03-26 21:14:25)
Offline
what in fact your patch fixes?
Arch64/DWM || My Dropbox referral link
Offline
[off-topic] (apologies)
OT: If you want bfq, I fixed it for 2.6.38.
I didn't realize bfq was broken for 2.6.38?
And uhh...
echo $(( $(nproc) + 1 ))
Sad to say I never knew about nproc, which I guess is considered more reliable than /sys files(?). Hah, at least they both use the same amount of commands.
Another OT: (edited): Turns out my problem with wifi is related to 'wext' driver in .38 where it worked fine in .36. Now I'm using 'nl80211' and it works great.
[/off-topic]
By the way, system seems quite responsive as compared to my other kernel with older bfs and ck (2.6.36 series). Compiled two kernels at once while browsing the web, playing a 720p video and editing/opened a good handful of documents without so much as a flicker in screen redraws when panning excessively between workspaces. So far so good..
Last edited by milomouse (2011-03-27 15:28:07)
Offline
Had to wait on BFQ and got impatient and stopped checking for updates
I didn't realize bfq was broken for 2.6.38?
There's a conflict here, and there was a conflict in applying the old bfq patchset to 2.6.38. I just moved around some code.
Sad to say I never knew about nproc, which I guess is considered more reliable than /sys files(?). Hah, at least they both use the same amount of commands.
Probably the only time I've ever used nproc, as well.
Offline
I take it this will find it's way into kernel26-ck 2.6.38? *Fingers Crossed*
Did you read the release notes or changelog? It made it in release -2.
http://aur.archlinux.org/packages.php?ID=32877
OT: If you want bfq, I fixed it for 2.6.38.
What are you talking about? It applies natively without error, no?
==> Patching source with BFQ patches
patching file block/Kconfig.iosched
patching file block/blk-ioc.c
patching file block/cfq-iosched.c
patching file fs/ioprio.c
patching file include/linux/iocontext.h
patching file block/Makefile
patching file include/linux/cgroup_subsys.h
patching file block/bfq-cgroup.c
patching file block/bfq-ioc.c
patching file block/bfq-iosched.c
patching file block/bfq-sched.c
patching file block/bfq.h
Last edited by graysky (2011-03-26 22:33:48)
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
The sources I have must be out of date then... it was an impossible patch without fixing a few lines.
Ah! He didn't update his 'source' page at: http://algo.ing.unimo.it/people/paolo/d … ources.php
[DIR] Parent Directory -
[DIR] 2.6.32/ 26-Aug-2010 07:34 -
[DIR] 2.6.33/ 26-Aug-2010 07:34 -
[DIR] 2.6.34-zen1/ 26-Aug-2010 07:35 -
[DIR] 2.6.34/ 26-Aug-2010 07:31 -
[DIR] 2.6.35/ 26-Aug-2010 07:28 -
[DIR] 2.6.36/ 25-Oct-2010 08:00 -
[DIR] 2.6.37-v2-r1/ 25-Mar-2011 17:15 -
[DIR] 2.6.37-v2/ 25-Mar-2011 17:06 -
If you go to the following, they are there: http://algo.ing.unimo.it/people/paolo/d … es/2.6.38/
Last edited by graysky (2011-03-26 23:08:09)
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
@graysky: Highlight and scroll down, the links are on source page too. It's an inline frame, it appears, with scrollbar.
Last edited by milomouse (2011-03-26 23:24:32)
Offline