Re: [gentoo-user] [smp related] 2.6.13 is 10x slower than 2.6.12!

2005-10-05 Thread Iain Buchanan
On Tue, 2005-10-04 at 09:36 +0100, Richard Brown wrote:
 On 04/10/05, Iain Buchanan [EMAIL PROTECTED] wrote:
 
  Any pointers would be appreciated...
 
 Are you using gentoo-sources? This is the recommended routine here.
 
 http://dev.gentoo.org/~dsd/genpatches/bugs.htm
 
 The kernel does have a bugzilla, it's http://bugzilla.kernel.org

thanks.  Luckily I tried bugzilla.kernel.org before I tried the
suggestions on dev.gentoo.org.

For those following this thread, it appears to be a problem with 2.6.13
and C-states...  The original reporter also had a Dell Inspiron 9100, so
maybe its only that laptop.
More here: http://bugzilla.kernel.org/show_bug.cgi?id=5165

There are two patches on bugzilla above, which I'm going to try out.
Hopefully they'll make it into 2.6.13.x soon!
-- 
Iain Buchanan [EMAIL PROTECTED]

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] [smp related] 2.6.13 is 10x slower than 2.6.12!

2005-10-04 Thread Richard Brown
On 04/10/05, Iain Buchanan [EMAIL PROTECTED] wrote:

 Any pointers would be appreciated...


Are you using gentoo-sources? This is the recommended routine here.

http://dev.gentoo.org/~dsd/genpatches/bugs.htm

The kernel does have a bugzilla, it's http://bugzilla.kernel.org

Regards,

--
Richard Brown

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] [smp related] 2.6.13 is 10x slower than 2.6.12!

2005-10-04 Thread fire-eyes
Iain Buchanan wrote:
 Not long ago, I griped about 2.6.13 being _very_ slow to boot and run.
 My original email is at the bottom.
 
 I've since recompiled the kernel without SMP and Hyperthreading, and the
 system is _much_ faster.
 
 So, there is a problem with 2.6.13 and SMP or Hyperthreaded machines.
 Can anyone else confirm this?  What should I do?  Is there a kernel.org
 bugzilla?  Or perhaps I should put it in the gentoo bugzilla?
 
 Any pointers would be appreciated...
 
 thanks.
 
 On Tue, 2005-09-27 at 12:11 +0930, Iain Buchanan wrote:

Curious. I have used many kernels on many real (2 physical cpus) SMP
systems, and have never had problems. In fact they are all currently
running 2.6.13 or higher.

I haven't used anything on HT cpus though, so I don't know about that.
-- 
gentoo-user@gentoo.org mailing list



[gentoo-user] [smp related] 2.6.13 is 10x slower than 2.6.12!

2005-10-03 Thread Iain Buchanan
Not long ago, I griped about 2.6.13 being _very_ slow to boot and run.
My original email is at the bottom.

I've since recompiled the kernel without SMP and Hyperthreading, and the
system is _much_ faster.

So, there is a problem with 2.6.13 and SMP or Hyperthreaded machines.
Can anyone else confirm this?  What should I do?  Is there a kernel.org
bugzilla?  Or perhaps I should put it in the gentoo bugzilla?

Any pointers would be appreciated...

thanks.

On Tue, 2005-09-27 at 12:11 +0930, Iain Buchanan wrote:
 I just compiled 2.6.13-gentoo-r2 with mostly the same options as
 2.6.12-gentoo-r7 (my reference point).  The first time I booted, it took
 about 10-15 minutes on my 3GHz P4 laptop!  I watched the gdm background
 slowly display over about 30 seconds.
 
 I thought it might have something to do with the new default timer
 frequency, which I set to 250Hz instead of 1000Hz (which is the default
 for pre 2.6.13 kernels).  According to various online sites, 250Hz
 should still be just as usable, and may even use less power, increasing
 my battery performance...
 
 So I changed the timer frequency back to 1000Hz, recompiled and now it
 seems a little faster, but no where near 2.6.12.  It now boots in maybe
 5 minutes, and the gdm theme takes about 5s to display.
 
 glxgears (not maximised) gives me about 30 fps with a radeon 9700!
 
 I checked the frequency governor, and the default is still
 'performance'.  /proc/acpi/processor/CPU0/throttling shows the maximum
 frequency is active as per normal.
 
 Other kernel options I changed are:
 - support for software suspend added (could this be it? I definitely saw
 the message that no valid suspend was found in my swapfile,
 and /proc/swaps shows that I have swap enabled as per usual)
 - removed unnecessary SCSI drivers (I don't have anything SCSI)
 - removed SATA support (I definitely don't have SATA)
 
 Info:
 Dell Ispiron 9100, 3 GHz HT Pentium 4
 2.6.12 and 2.6.13: SMP support for 2 processors included
 
 I've seen 1 post online about the same thing - slow running 2.6.13, so
 surely there must be some people in the world running it ok!  Any help
 on this would be appreciated.
 
 TIA,

-- 
Iain Buchanan [EMAIL PROTECTED]

-- 
gentoo-user@gentoo.org mailing list