From: Nick Piggin <piggin@cyberone.com.au>
To: Erwin Telser <erwin.telser@pop.agri.ch>
Cc: linux-kernel@vger.kernel.org
Subject: Re: Responsiveness of 2.6.0-Test9
Date: Tue, 04 Nov 2003 13:32:58 +1100 [thread overview]
Message-ID: <3FA70FDA.2000502@cyberone.com.au> (raw)
In-Reply-To: <200311032353.16570.erwin.telser@pop.agri.ch>
Erwin Telser wrote:
>>Erwin Telser wrote:
>>
>>>Are there things to observe when switching from 2.4 to 2.6, if I want the
>>>same responsiveness? I'm asking because I' ve made the following
>>>observation. (I always had the feeling, 2.4 seems to faster, but this is
>>>the first time it' s very obvious)
>>>
>>>I' ve connected two monitors on a Matrox G550. One runs with DRI the other
>>>one doesn't (not possible with the current driver). The Monitor with DRI
>>>I'm using as a TV Set, watching movies with xawtv. (With a bttv 878 tuner
>>>card).
>>>
>>>Now with the 2.4.22 kernel (preemptible patch aplied) I can play the
>>>little game kbounce on the other monitor, without to notice any slowdown,
>>>no matter, whether xawtv is running or not.
>>>
>>>But with the 2.6 Kernel (compiled with preemptible option) the bouncing
>>>balls slow down considerably, as soon as I move the mouse.
>>>
>>>I know the whole thing is a little foolish. But anyway, are there some
>>>tricks to get the same responsiveness?
>>>
>>Hi,
>>Your report is not foolish at all ;)
>>Please make sure your X process is at default static priority, 0.
>>Report back if you are still having problems. What sort of processor
>>do you have? Is your 2.6 video driver running with the same sort of
>>acceleration as your 2.4 one (do you get the same frame rate in glxgears)?
>>
>
>I installed X from scratch and didn't edit any start script. so I assume, that
>it runs with priority 0. The processor is an Intel P4 2000MHz in an ASUS
>P4B266 Board. If I compare the frames per second with glxgears there is a
>slight advantage for 2.4
>
renice 0 `pidof X`
>
>2.4:
>with DRI 358 - 364
>without DRI 181 - 198
>
>2.6
>with DRI 350 - 362
>without DRI 180 - 187
>
>An even better way to show the differenze between 2.4 and 2.6 is to change the
>size of the xawtv window. If I do that, the slowdown is quite significant.
>But it happens only with xawtv. If I do the same with konqueror or MPlayer,
>nothing happens. A very strange thing are the values I get from ksysguard
>
>Kernel 2.4, glxgears
>user load: 3% - 4%
>system load: 95%
>IRQ per second: ~100
>
>Kernel 2.6, glxgears
>user load: 15%
>system load: 85%
>IRQ per second: ~1100
>
>Kernel 2.4, xawtv
>user load: 2%
>system load: 0%
>IRQ per second: ~100
>
>Kernel 2.6, xawtv
>user load: 9%
>system load: 0%
>IRQ per second: ~1060
>
>Kernel 2.4, MPlayer
>user load: 35%
>system load: 2%
>IRQ per second: ~1170
>
>Kernel 2.6, MPlayer
>user load: 42%
>system load: 2%
>IRQ per second: ~2130
>
>Something is causing a high IRQ load with kernel 2.6. Probably an APIC
>problem?
>
It is the timer interrupt. Now happens 1000 per second instead of 100.
This might just be the reason why you are seeing a different balance, the
collection of cpu statistics is more precise.
prev parent reply other threads:[~2003-11-04 2:33 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-11-02 23:45 Responsiveness of 2.6.0-Test9 Erwin Telser
2003-11-03 4:43 ` Nick Piggin
2003-11-03 23:53 ` Erwin Telser
2003-11-04 2:32 ` Nick Piggin [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3FA70FDA.2000502@cyberone.com.au \
--to=piggin@cyberone.com.au \
--cc=erwin.telser@pop.agri.ch \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox