11 Oct
2004
11 Oct
'04
9:32 p.m.
One of the JACK developers reported this problem on LKML and the reply
was very interesting, this could explain a lot of the weird latency
issues that laptop users are seeing.
http://lkml.org/lkml/2004/10/11/182
The basic issue seems to be if your laptop has a broken BIOS which
implements ACPI using SMM you are out of luck. Here is some more
information on the problem (near the end of the page):
http://www.infoworld.com/cgi-bin/displayNew.pl?/glass/bg112596.htm
If ACPI causes you massive xrun/latency problems then you might have
such a broken BIOS. If you are having weird latency problems with a
laptop, even with ACPI disabled, try to see if it corresponds to the fan
turning on or changing speed. If so then you might be screwed :-(
There does not seem to be a lot of hope if you do have one of these
broken machines. But we can at least identify the problematic laptops,
complain to the manufacturers, and warn people not to buy them.
Lee
11 Oct
11 Oct
10:12 p.m.
Lee Revell wrote:
One of the JACK developers reported this problem on
LKML and the reply
was very interesting, this could explain a lot of the weird latency
issues that laptop users are seeing.
http://lkml.org/lkml/2004/10/11/182
The basic issue seems to be if your laptop has a broken BIOS which
implements ACPI using SMM you are out of luck. Here is some more
information on the problem (near the end of the page):
http://www.infoworld.com/cgi-bin/displayNew.pl?/glass/bg112596.htm
If ACPI causes you massive xrun/latency problems then you might have
such a broken BIOS. If you are having weird latency problems with a
laptop, even with ACPI disabled, try to see if it corresponds to the fan
turning on or changing speed. If so then you might be screwed :-(
There does not seem to be a lot of hope if you do have one of these
broken machines. But we can at least identify the problematic laptops,
complain to the manufacturers, and warn people not to buy them.
My laptop, a Compaq Presario 2516EA, probably suffers from this ACPI disease.
However, this has been mitigated by just disabling it; just booting with
acpi=off the xrun rate is almost negligible (e.g.running 2.6.9-rc3-mm3-T3
UP).
Don't know if the SMM issue is deadly hidden and commanded somewhere from
BIOS, then making life a pain, even though apci is off. How can I check
this?
For the record, I'm quite used for acpi being a real pita, but the lack of
it doesn't seem to be much important to me. OK, with acpi=off, I miss
battery status, temparature monitoring, whatever, just to name a few
goodies, but I buy that for cheap, just about when I get to do low-latency
audio.
For example, I can do along with jackd -p128 -n2 using the onboard audio
(ali5451) without much trouble, iif acpi=off, otherwise a xrun breakbeat
is featured ;)
Take care.
--
rncbc aka Rui Nuno Capela
rncbc(a)rncbc.org
11:09 p.m.
On Mon, 11 Oct 2004 21:02:03 -0400, Peter Lutek <plutek(a)infinity.net> wrote:
On Mon, 2004-10-11 at 19:01, Lee Revell wrote:
--snip--
Not yet. I'd also like to go a bit further in my understanding of the
problem before I participate in warning people about something turns
out to be a bit more nuanced.
I suppose my laptop (also a Compaq - an R3070us) could be considered a
'problematic' laptop also, but my thought is that I don't really care
about ACPI when I'm doing audio anyway. I'm not running on batteries
or doing any of that sort of stuff when I'm recording and mixing. I
want ACPI and every other distraction to the machine turned off. For
instance, I'm seeing far fewer xruns still with fluxbox. I see one or
two under Gnome. I see bunches of them under KDE. So, at least with
ACPI off, I think it's not so certain that ACPI is the *only* problem
here.
As long as I can make it work in some configuration then I'm happy and
the laptop is no longer problematic, at least for me...
My guess is that 'users' just using Alsa would not care so much about
latency and that they'd be happy with the machine.
Just my two cents.
- Mark
But we can at least identify the problematic
laptops,
complain to the manufacturers, and warn people not to buy them.
has anyone yet compiled such a list?
-p
11:27 p.m.
On Mon, 2004-10-11 at 20:51, Mark Knecht wrote:
On Mon, 11 Oct 2004 21:02:03 -0400, Peter Lutek
<plutek(a)infinity.net> wrote:
> On Mon, 2004-10-11 at 19:01, Lee Revell wrote:
> --snip--
> > But we can at least identify the problematic laptops,
> > complain to the manufacturers, and warn people not to buy them.
>
> has anyone yet compiled such a list?
>
I suppose my laptop (also a Compaq - an R3070us) could
be considered a
'problematic' laptop also, but my thought is that I don't really care
about ACPI when I'm doing audio anyway. I'm not running on batteries
or doing any of that sort of stuff when I'm recording and mixing. I
want ACPI and every other distraction to the machine turned off. For
instance, I'm seeing far fewer xruns still with fluxbox. I see one or
two under Gnome. I see bunches of them under KDE. So, at least with
ACPI off, I think it's not so certain that ACPI is the *only* problem
here.
The LKML poster's main point was that if your laptop implements ACPI
with SMM you WILL get xruns because SMM disables interrupts. Even
disabling ACPI in the kernel will not eliminate the problem 100% because
the hardware sensors will cause the machine to go to SMM mode to enable
the fan for example.
So this _is_ definitely a problem IF your laptop uses SMM to implement
ACPI. It remains to be seen how widespread the issue is, and whether
there is a way to tell whether a given machine has the problem other
than trial and error.
Lee
15 Oct
15 Oct
8:48 a.m.
On Mon, Oct 11, 2004 at 09:06:23PM -0400, Lee Revell wrote:
On Mon, 2004-10-11 at 20:51, Mark Knecht wrote:
APM is implemented with SMM. turning off ACPI will give you problems.
(we are talking about the sub milisecond latency range here)
i still see some hope to fix this stuff by disabling APM (which is done
by the ACPI code) but not using ACPI.
will experiment with this stuff during the weekend....
--
torben Hohn
http://galan.sourceforge.net -- The graphical Audio language
On Mon, 11 Oct 2004 21:02:03 -0400, Peter Lutek
<plutek(a)infinity.net> wrote:
> On Mon, 2004-10-11 at 19:01, Lee Revell wrote:
> --snip--
> > But we can at least identify the problematic laptops,
> > complain to the manufacturers, and warn people not to buy them.
>
> has anyone yet compiled such a list?
>
I suppose my laptop (also a Compaq - an R3070us)
could be considered a
'problematic' laptop also, but my thought is that I don't really care
about ACPI when I'm doing audio anyway. I'm not running on batteries
or doing any of that sort of stuff when I'm recording and mixing. I
want ACPI and every other distraction to the machine turned off. For
instance, I'm seeing far fewer xruns still with fluxbox. I see one or
two under Gnome. I see bunches of them under KDE. So, at least with
ACPI off, I think it's not so certain that ACPI is the *only* problem
here.
The LKML poster's main point was that if your laptop implements ACPI
with SMM you WILL get xruns because SMM disables interrupts. Even
disabling ACPI in the kernel will not eliminate the problem 100% because
the hardware sensors will cause the machine to go to SMM mode to enable
the fan for example.
So this _is_ definitely a problem IF your laptop uses SMM to implement
ACPI. It remains to be seen how widespread the issue is, and whether
there is a way to tell whether a given machine has the problem other
than trial and error.
11 Oct
11 Oct
11:15 p.m.
On Mon, 2004-10-11 at 21:02, Peter Lutek wrote:
On Mon, 2004-10-11 at 19:01, Lee Revell wrote:
--snip--
No, because the problem was only very recently identified. We are not
sure which laptops are affected at this point.
Lee
But we can at least identify the problematic
laptops,
complain to the manufacturers, and warn people not to buy them.
has anyone yet compiled such a list?
15 Oct
15 Oct
8:44 a.m.
On Mon, Oct 11, 2004 at 07:01:27PM -0400, Lee Revell wrote:
One of the JACK developers reported this problem on
LKML and the reply
was very interesting, this could explain a lot of the weird latency
issues that laptop users are seeing.
since when am i a jack developer ?
anyway.
with ACPI enabled i am getting a lot of xruns.
and the soundcard IRQ is different.
i dont know where this comes from.
according to the posted link ACPI was designed to overcome the SMM
problem. with ACPI enabled my machine does not change the fan speed all
the time. so i guess there is still hope somewhere....
--
torben Hohn
http://galan.sourceforge.net -- The graphical Audio language
2:57 p.m.
On Fri, 2004-10-15 at 06:31, torbenh(a)gmx.de wrote:
On Mon, Oct 11, 2004 at 07:01:27PM -0400, Lee Revell
wrote:
I could have sworn I saw your name on it somewhere. Guess I am on too
many mailing lists...
Lee
One of the JACK developers reported this problem
on LKML and the reply
was very interesting, this could explain a lot of the weird latency
issues that laptop users are seeing.
since when am i a jack developer ?
7391
days inactive
7395
days old
linux-audio-user@lists.linuxaudio.org
8 comments
5 participants
participants (5)
-
Lee Revell -
Mark Knecht -
Peter Lutek -
Rui Nuno Capela -
torbenh＠gmx.de