You are not logged in.

#1 2009-09-03 20:27:33

mangus
Member
From: Bologna, Italy
Registered: 2007-04-07
Posts: 288

omg! Con Kolivas is back!

http://ck.kolivas.org/patches/bfs/

Con is on the move again! long time linux user surely remember his desktop patchsets and his fights with the mainline kernel people.
As long time Con's patches user my dream would be an official arch kernel patched for desktops with this new scheduler (bfs , brain fuck scheduler big_smile).
Dreams costs nothing....

Offline

#2 2009-09-03 21:18:28

mollison
Member
Registered: 2009-01-07
Posts: 15

Re: omg! Con Kolivas is back!

I'd definitely be interested in seeing a PKGBUILD to install the kernel patched with the BFS scheduler. :-)    Maybe I will make one but probably will not have time.

Offline

#3 2009-09-03 21:35:29

tomk
Forum Fellow
From: Ireland
Registered: 2004-07-21
Posts: 9,838

Re: omg! Con Kolivas is back!

It will never be in the official Arch kernel, because it will never be in the mainline kernel. Just keep an eye on the AUR - I'd say someone will submit a build any day now. smile

Offline

#4 2009-09-03 21:37:47

tomk
Forum Fellow
From: Ireland
Registered: 2004-07-21
Posts: 9,838

Re: omg! Con Kolivas is back!

hehe... it's already there - although the comment is not encouraging.

Offline

#5 2009-09-03 21:56:48

broch
Banned
From: L.A. California
Registered: 2006-11-13
Posts: 975

Re: omg! Con Kolivas is back!

I am using BFS for few days (zen kernel -rc7/8-zen1 - not AUR though but git), better system responsiveness under load than CFS,
zen has some extra perks (Sched_iso X -  for X server) and one can select both BFS and BFQ for nice/experimantal CPU and I/O scheduling.

This is still experimental so while .31-rc7-zen1 works without any issues with .31-rc8-zen1, I lost suspend to RAM. Still this is a small price for improved responsiveness.

I hope that Con Kolivas will stay and we will have again good responsive desktop linux kernel.

Last edited by broch (2009-12-12 15:33:27)

Offline

#6 2009-09-04 12:04:00

alecmg
Member
Registered: 2008-12-21
Posts: 84

Re: omg! Con Kolivas is back!

I hope the collision course for kernel to become fatter and slower and more bloated at an alarming rate will be reversed


Xyne wrote:
"We've got Pacman. Wacka wacka, bitches!"

Offline

#7 2009-09-04 13:07:47

.:B:.
Forum Fellow
Registered: 2006-11-26
Posts: 5,819

Re: omg! Con Kolivas is back!

<3 Con Kolivas

So glad to see he picked up his game smile. I am gonna wreck my kernel as soon as I can tongue.


Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy

Offline

#8 2009-09-04 13:24:06

shining
Pacman Developer
Registered: 2006-05-10
Posts: 2,043

Re: omg! Con Kolivas is back!

alecmg wrote:

I hope the collision course for kernel to become fatter and slower and more bloated at an alarming rate will be reversed

Can you even prove your points or is this just random nonsense ?

I can remember benchmarks comparing different kernel versions, it definitely didn't show slower and slower kernels.
Nor did I ever notice that in my own experience.

And of course, the kernel grows, and thanks god it grows, with all the hardware there is to support out there. But this does not affect a given system at all.

And I don't see why/how Con's work would reverse anything at all. I don't think you even get Con's main point : there is not one scheduler to rule them all
because focusing on one usage (e.g. common desktop usage), and one kind of machines (dual/quad cores but not 4096 cores) allows better performance


pacman roulette : pacman -S $(pacman -Slq | LANG=C sort -R | head -n $((RANDOM % 10)))

Offline

#9 2009-09-06 13:21:16

.:B:.
Forum Fellow
Registered: 2006-11-26
Posts: 5,819

Re: omg! Con Kolivas is back!

Anyone tried it? I'm going to soon.


Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy

Offline

#10 2009-09-06 13:41:41

bangkok_manouel
Member
From: indicates a starting point
Registered: 2005-02-07
Posts: 1,554

Re: omg! Con Kolivas is back!

B wrote:

Anyone tried it? I'm going to soon.

Using it at the moment...

.30 kernel with bfs

Using 1987461 loops per ms, running every load for 30 seconds
Benchmarking kernel 2.6.30-bfs at datestamp 200909052130

--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.007 +/- 0.00785    0.013             100            100
Video     0.006 +/- 0.00652    0.013             100            100
X         0.006 +/- 0.00686    0.012             100            100
Burn      0.006 +/- 0.00652    0.009             100            100
Write      0.01 +/- 0.0109     0.027             100            100
Read      0.011 +/- 0.0118     0.023             100            100
Compile    0.01 +/- 0.0181     0.266             100            100
Memload   0.012 +/- 0.0196     0.356             100            100

--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.007 +/- 0.00751    0.013             100            100
X         0.006 +/- 0.00676    0.025             100            100
Burn       11.4 +/- 13.8        16.7             100           31.4
Write     0.019 +/- 0.394       16.7             100           99.9
Read      0.008 +/- 0.00863    0.024             100            100
Compile    11.7 +/- 14          33.3            99.6           30.4
Memload   0.009 +/- 0.0115     0.226             100            100

--- Benchmarking simulated cpu of X in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.006 +/- 0.0817         1             100           99.3
Video     0.006 +/- 0.0817         1             100           99.3
Burn       52.5 +/- 73.3         164            19.6           8.89
Write     0.006 +/- 0.0817         1             100           99.3
Read      0.006 +/- 0.0817         1             100           99.3
Compile    56.2 +/- 78.2         182            19.2            8.5
Memload   0.013 +/- 0.141          2             100             99

--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU
None      0.037 +/- 0.0624     0.422             100
Video     0.053 +/- 0.204       1.76            99.9
X         0.075 +/- 0.352       3.01            99.9
Burn        150 +/- 151          168              40
Write     0.825 +/- 2.4         23.4            99.2
Read      0.181 +/- 0.196      0.652            99.8
Compile     161 +/- 163          214            38.3
Memload   0.695 +/- 1.73        13.3            99.3

latest -mmotm

Using 1987461 loops per ms, running every load for 30 seconds
Benchmarking kernel 2.6.31-rc8-mm1 at datestamp 200909052201

--- Benchmarking simulated cpu of Audio in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.008 +/- 0.0108     0.177             100            100
Video     0.013 +/- 0.0132     0.036             100            100
X         0.014 +/- 0.0148     0.022             100            100
Burn      0.006 +/- 0.0199     0.467             100            100
Write     0.184 +/- 1.79        28.6             100            100
Read      0.012 +/- 0.0161     0.138             100            100
Compile    0.01 +/- 0.0264     0.596             100            100
Memload   0.034 +/- 0.266       4.44             100            100

--- Benchmarking simulated cpu of Video in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.021 +/- 0.0213     0.083             100            100
X         0.006 +/- 0.00699    0.015             100            100
Burn        7.2 +/- 12.4        34.6            99.3           81.9
Write     0.791 +/- 5.11          50            98.3             97
Read      0.009 +/- 0.0133      0.18             100            100
Compile    21.7 +/- 42.3         236            60.1           53.2
Memload   0.043 +/- 0.175       2.61             100            100

--- Benchmarking simulated cpu of X in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU  % Deadlines Met
None      0.006 +/- 0.0817         1             100           99.3
Video     0.006 +/- 0.0817         1             100           99.3
Burn       49.3 +/- 74.1         200            15.8           7.79
Write      3.41 +/- 13.4          72            90.9           87.5
Read      0.006 +/- 0.0817         1             100           99.3
Compile      60 +/- 90           232            16.9           8.66
Memload   0.063 +/- 0.473          5            97.1           96.1

--- Benchmarking simulated cpu of Gaming in the presence of simulated ---
Load    Latency +/- SD (ms)  Max Latency   % Desired CPU
None      0.164 +/- 0.191       1.31            99.8
Video     0.213 +/- 0.229       1.15            99.8
X         0.148 +/- 0.24        2.22            99.9
Burn        150 +/- 153          218              40
Write      6.98 +/- 19.9         177            93.5
Read      0.326 +/- 0.389        2.5            99.7
Compile     211 +/- 215          421            32.2
Memload   0.617 +/- 1.16        7.59            99.4

All design goals must be phrased in such a way that it is hard to use them as slogans to justify stupidity.

Offline

#11 2009-09-06 13:52:50

wonder
Developer
From: Bucharest, Romania
Registered: 2006-07-05
Posts: 5,937
Website

Re: omg! Con Kolivas is back!

@bangkok_manouel how did you make those benchmarks?


Give what you have. To someone, it may be better than you dare to think.
Blog

Offline

#12 2009-09-06 13:58:51

bangkok_manouel
Member
From: indicates a starting point
Registered: 2005-02-07
Posts: 1,554

Re: omg! Con Kolivas is back!

wonder wrote:

@bangkok_manouel how did you make those benchmarks?

a tool called interbench, please see:
http://users.on.net/~ckolivas/interbench/


All design goals must be phrased in such a way that it is hard to use them as slogans to justify stupidity.

Offline

#13 2009-09-06 13:58:58

flamelab
Member
From: Athens, Hellas (Greece)
Registered: 2007-12-26
Posts: 2,160

Re: omg! Con Kolivas is back!

Is kernel26-bfs more responsive under heavy load than the regular kernel26 ? Or are there other differences ?

Offline

#14 2009-09-06 14:05:46

bangkok_manouel
Member
From: indicates a starting point
Registered: 2005-02-07
Posts: 1,554

Re: omg! Con Kolivas is back!

http://ck.kolivas.org/patches/bfs/bfs-faq.txt should answer most questions.


All design goals must be phrased in such a way that it is hard to use them as slogans to justify stupidity.

Offline

#15 2009-09-06 15:45:06

mangus
Member
From: Bologna, Italy
Registered: 2007-04-07
Posts: 288

Re: omg! Con Kolivas is back!

I had some hard freeze after some usage (OOM like) with rev 208 , using X and KDE4. I got an intel  dualcore
Anyone else? It's annoying and weird hmm

Offline

#16 2009-09-06 15:53:18

agapito
Member
From: Spain
Registered: 2008-11-13
Posts: 495

Re: omg! Con Kolivas is back!

This is funny

supported_features.png


Excuse my poor English.

Offline

#17 2009-09-06 16:08:11

skottish
Forum Fellow
From: Here
Registered: 2006-06-16
Posts: 7,931

Re: omg! Con Kolivas is back!

CK wrote:

Is this stable?

Probably not. I use it on half a dozen machines but the code only booted for
the first time successfully on the 25th August 2009 so you work out how new
it is.

Sweet! B, test it out for me.

Offline

#18 2009-09-06 16:11:32

flamelab
Member
From: Athens, Hellas (Greece)
Registered: 2007-12-26
Posts: 2,160

Re: omg! Con Kolivas is back!

Ι installed the kernel with .209 patch.

It seems OK for now.

Offline

#19 2009-09-06 16:19:00

.:B:.
Forum Fellow
Registered: 2006-11-26
Posts: 5,819

Re: omg! Con Kolivas is back!

skottish wrote:
CK wrote:

Is this stable?

Probably not. I use it on half a dozen machines but the code only booted for
the first time successfully on the 25th August 2009 so you work out how new
it is.

Sweet! B, test it out for me.

Hey! I'm not your guinea pig tongue.

* B goes off to build a 2.6.30 kernel

Laptop is still on 2.6.29, time I move on big_smile.


Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy

Offline

#20 2009-09-06 16:39:13

bangkok_manouel
Member
From: indicates a starting point
Registered: 2005-02-07
Posts: 1,554

Re: omg! Con Kolivas is back!

mangus wrote:

I had some hard freeze after some usage (OOM like) with rev 208 , using X and KDE4. I got an intel  dualcore
Anyone else? It's annoying and weird hmm

no hard freeze but the "unkillable tasks" problem. with 209+this patch (which is known to help but not to completely solve this issue) http://ck.kolivas.org/patches/bfs/clone … read.patch , i didn't get any.


All design goals must be phrased in such a way that it is hard to use them as slogans to justify stupidity.

Offline

#21 2009-09-06 17:48:53

.:B:.
Forum Fellow
Registered: 2006-11-26
Posts: 5,819

Re: omg! Con Kolivas is back!

I just built my kernel with the patch mentioned above, setting the clock till when I see my first system instability tongue.

Edit: launched Transmission and within seconds it seems to have locked up. It has its own process but no window visible, and I can't kill it (SIGTERM or SIGKILL), not even as root...

I think i'm going back to the non-CK kernel scheduler for the time being tongue.


Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy

Offline

#22 2009-09-06 18:05:19

Army
Member
Registered: 2007-12-07
Posts: 1,784

Re: omg! Con Kolivas is back!

I read the faq, but there's one question left, which is important for me. Does it make sense to switch to bfs with a single core cpu?

Offline

#23 2009-09-06 18:41:10

.:B:.
Forum Fellow
Registered: 2006-11-26
Posts: 5,819

Re: omg! Con Kolivas is back!

Well given Con seems to have written this thing taking old systems into account, I'd say yes...


Got Leenucks? :: Arch: Power in simplicity :: Get Counted! Registered Linux User #392717 :: Blog thingy

Offline

#24 2009-09-06 20:41:48

kakalaky
Member
Registered: 2006-12-28
Posts: 10

Re: omg! Con Kolivas is back!

Army wrote:

I read the faq, but there's one question left, which is important for me. Does it make sense to switch to bfs with a single core cpu?

I'm using bfs on my old Pentium M laptop.  Responsiveness is MUCH better.

Offline

#25 2009-09-06 21:00:48

Army
Member
Registered: 2007-12-07
Posts: 1,784

Re: omg! Con Kolivas is back!

Ok, I have a pentium m as well, I think I'll give it a shot. B, you are right, that might be where this information was hidden for my eyes ;-)

Offline

Board footer

Powered by FluxBB