unofficial linux-ck repo with cpu optimized packages is ONLINE

Dethredic · 2011-10-22 14:41:24

nvm, figure it out

Last edited by Dethredic (2011-10-22 15:04:26)

berz_ · 2011-10-22 16:08:44

I want to try out the BFS, but am unsure about which architectures I have.

$ cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 23
model name	: Pentium(R) Dual-Core CPU       T4200  @ 2.00GHz

It is listed as an "Intel® Pentium® Mobile Processor" on intel's site.
After searching I still do not know whether I should install ck-pentm, ck-p4 or ck-core2.
I do know that I can compile with -march=native, but prefer to just install from a repo.

Last edited by berz_ (2011-10-22 16:09:16)

graysky · 2011-10-22 16:16:48

@berz_ - you can't go wrong the generic packages. (ck-generic)

Or just see what the following returns and use that package.

$ gcc -c -Q -march=native --help=target|grep march

ijanos · 2011-10-26 11:06:17

Hi,

the latest upgrade fails because of broadcom-wl-ck-core2 package in the ck repo contains a file (/etc/modprobe.d/broadcom-wl.conf) that the other broadcom-wl package (in AUR, for the -arch kernel) also provides.

The obvious solution is to add broadcom-wl to confilcts=() array in the PKGBUILD but then wireless connetion will only be available with the -ck kernel and not with the stock kernel, which is not really ideal.

I suggest renaming the file in the package to broadcom-wl-ck.conf, I think two files with the same blacklist won't cause any trouble.

Any better ideas?

Zom · 2011-10-26 11:45:49

ijanos wrote:

Hi,
the latest upgrade fails because of broadcom-wl-ck-core2 package in the ck repo contains a file (/etc/modprobe.d/broadcom-wl.conf) that the other broadcom-wl package (in AUR, for the -arch kernel) also provides.
The obvious solution is to add broadcom-wl to confilcts=() array in the PKGBUILD but then wireless connetion will only be available with the -ck kernel and not with the stock kernel, which is not really ideal.
I suggest renaming the file in the package to broadcom-wl-ck.conf, I think two files with the same blacklist won't cause any trouble.

Any better ideas?

This is the exact approach taken with the nvidia-ck(-*)-packages, so no, it should be just fine I think.

graysky · 2011-10-26 12:17:13

@ijanos - Great idea and thanks for pointing it out. Fixed in 5.100.82.112-1 which just posted to the AUR and repo.

Last edited by graysky (2011-10-26 12:19:23)

ijanos · 2011-10-26 12:40:32

graysky wrote:

@ijanos - Great idea and thanks for pointing it out. Fixed in 5.100.82.112-1 which just posted to the AUR and repo.

Thank You for the quick response.

JazzplayerL9 · 2011-10-26 16:49:14

Had a quick question for anyone else using the linux-ck-pentm package from graysky's repo.

Every...maybe 5 min or so I get a couple of seconds of system hang. It's really noticeable when playing music because it just stops for a couple of seconds and then continues. Is anyone else having this problem? (linux-ck-pentm 3.0.7-1) And...more importantly, does anyone know how to fix it?

Thanks!

graysky · 2011-10-26 16:51:37

@jazzp - does this happen with the ck-generic package?

JazzplayerL9 · 2011-10-26 16:57:58

@graysky,

Haven't tried that yet...I'll install it and see what happens. Btw, seriously quick response!

UPDATE: Yeah, it does the same thing with the generic package.

Last edited by JazzplayerL9 (2011-10-26 17:08:22)

graysky · 2011-10-26 17:44:45

If you build it from the AUR, does it do the same thing? If so, and if the Arch kernel does not cause this effect, I would think there is some regression within the ck1 patchset and your specific hardware. In that case, I'd post it on CK's blog to see what he thinks.

JazzplayerL9 · 2011-10-26 18:20:05

I'll try building from the AUR...I'll let you know how it works in a few hours. There definitely might be a regression concerning my hardware...it's a Sony VAIO...they're a real pain. Thanks for your help graysky!

UPDATE: Just compiled and installed linux-ck from the AUR...same problem. Time to do some debugging!

Last edited by JazzplayerL9 (2011-10-26 21:16:21)

Markus00000 · 2011-10-30 15:39:00

(I posted this somewhere else some time ago, but had not noticed this thread.)

Does the ck package has anything to do with bash's history search and yank-last-arg?

I have this in my .inputrc:

$if mode=vi
    set keymap vi-command
    # these are for vi-command mode
    "\ek": history-search-backward
    "\ej": history-search-forward
    "\e.": yank-last-arg
    set keymap vi-insert
    # these are for vi-insert mode
    "\ek": history-search-backward
    "\ej": history-search-forward
    "\e.": yank-last-arg
$endif

This has been working for ages and still works when I boot the standard 3.0.x kernel. But when I boot the 3.0.x ck kernel, alt+j and alt+k become previous-history and next-history respectively, while alt+. becomes something totally weird (seems like it is pasting parts of the current line).

I see this happening on both of my two machines (ck-corei7-avx and ck-pentm).

graysky · 2011-10-30 19:45:16

@JazzplayerL9 - I thought that would be the case as the ck-generic packages are theoretically identical to the one you would build from the AUR.
@Markus0000- That's an interesting observation. The PKGBUILDs are nearly identical to the one in the AUR. There should be *no* modifications to files not owned by the PKG itself. What happens if you build the kernel from the AUR tarball?

b9anders · 2011-10-30 22:42:12

JazzplayerL9 wrote:

Had a quick question for anyone else using the linux-ck-pentm package from graysky's repo.
Every...maybe 5 min or so I get a couple of seconds of system hang. It's really noticeable when playing music because it just stops for a couple of seconds and then continues. Is anyone else having this problem? (linux-ck-pentm 3.0.7-1) And...more importantly, does anyone know how to fix it?
Thanks!

I get it sometimes when I run terminal tasks, such as compiling or pacman upgrades.

I am using the linux-ck-corei7.

In fact it happened just now when I ran "pacman -Qs linux-ck" to check which package I use. Didn't when I ran it again. It's been consistent ever since the Brain Fuck Scheduler started working for me (I first installed and used it just after linux 3 kernel came out and the patchset hadn't been updated).

So it's definitely something to do with scheduler.

student975 · 2011-10-31 02:46:21

With ck I have rather sudden feature When the system is heavy loaded by multithread applications,

- top shows, say, 260-280% of CPU load (and it is normal for some used benchmarking test and 4-core CPU using, i5 760 at my case),
- qps and KSysGuard (process table) can show anything starting from, say, 10% till 100% of expected load,
- KSysGuard CPU load graph doesn't show any 'system' load, just - sometimes - 'user' only (but must as far as it is the same test I use under official kernel when all load indicators are consistent).

I have linux-ck-corei7 from the repo-ck. The issue doesn't depend on I/O scheduler in use (and it is expected).

Where to dig in further?
Can anybody reproduce the issue?

Last edited by student975 (2011-10-31 02:47:18)

Markus00000 · 2011-10-31 05:03:01

graysky wrote:

@Markus0000- That's an interesting observation. The PKGBUILDs are nearly identical to the one in the AUR. There should be *no* modifications to files not owned by the PKG itself. What happens if you build the kernel from the AUR tarball?

Building from AUR does break these bindings, too.

graysky · 2011-10-31 09:51:56

@student - no idea what you're talking about What does htop show under load?
@Markus - that suggests whatever is going on to be caused by the upstream patches ... somehow. Suggest you contact ck either on his blog or by email.

student975 · 2011-10-31 10:08:35

@graysky

'top' shows ~300% CPU load as expected for four cores.
'htop' shows jumping values in ~25-100% range.

Last edited by student975 (2011-10-31 10:10:55)

graysky · 2011-10-31 10:23:29

student975 wrote:

'top' shows ~300% CPU load as expected for four cores.
'htop' shows jumping values in ~25-100% range.

So things seem normal then, no?

student975 · 2011-10-31 10:31:04

graysky wrote:

So things seem normal then, no?

No. 'top' is the only showing something related to reality. 'htop' is also affected:

student975 wrote:

'htop' shows jumping values in ~25-100% range.

graysky · 2011-10-31 10:50:40

@student - I never noticed this before. When I am building a package for example, and look at htop, I don't see cc taking up the cpu cycles.... is this what you are describing?

student975 · 2011-10-31 11:12:38

@graysky

OK, I'll try to prepare reproducible form official repo test case. I mean at constant durable multithreaded CPU loading test case htop (as well as qps and KSysGuard) shows almost arbitrary jumping values.

//---

offtopic: Have you a big avatar picture (and is it free)? My younger son (he is nine) want it to play with

graysky · 2011-10-31 11:25:03

@student - can you open a new thread to keep this one on topic: related to packages and repo. The issue you're describing is interesting, but not truly related. To your other question about the picture, just do a google image search. The avatar appeared on the covers of two of Disturbed's albums, Ten Thousand Fists and Indestructible.

student975 · 2011-10-31 16:11:51

graysky wrote:

@student - can you open a new thread to keep this one on topic: related to packages and repo. The issue you're describing is interesting, but not truly related. To your other question about the picture, just do a google image search. The avatar appeared on the covers of two of Disturbed's albums, Ten Thousand Fists and Indestructible.

Sorry, done, thanks! https://bbs.archlinux.org/viewtopic.php?id=129410

Arch Linux

#501 2011-10-22 14:41:24

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#502 2011-10-22 16:08:44

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#503 2011-10-22 16:16:48

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#504 2011-10-26 11:06:17

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#505 2011-10-26 11:45:49

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#506 2011-10-26 12:17:13

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#507 2011-10-26 12:40:32

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#508 2011-10-26 16:49:14

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#509 2011-10-26 16:51:37

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#510 2011-10-26 16:57:58

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#511 2011-10-26 17:44:45

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#512 2011-10-26 18:20:05

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#513 2011-10-30 15:39:00

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#514 2011-10-30 19:45:16

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#515 2011-10-30 22:42:12

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#516 2011-10-31 02:46:21

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#517 2011-10-31 05:03:01

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#518 2011-10-31 09:51:56

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#519 2011-10-31 10:08:35

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#520 2011-10-31 10:23:29

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#521 2011-10-31 10:31:04

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#522 2011-10-31 10:50:40

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#523 2011-10-31 11:12:38

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#524 2011-10-31 11:25:03

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

#525 2011-10-31 16:11:51

Re: unofficial linux-ck repo with cpu optimized packages is ONLINE

Board footer