12 May
2005
12 May
'05
6:57 p.m.
hi all..
i'm trying to sleep for some very low time slices ... about 100 to 1000
us ...
but i can't get below about 1 ms ... are there any workarounds to sleep
for very small time slices?
i tested nanosleep(), usleep() and select(0, 0, 0, 0, &timout) ...
thanks .... tim
--
mailto:TimBlechmann@gmx.de ICQ: 96771783
http://www.mokabar.tk
latest mp3: kMW.mp3
http://mattin.org/mp3.html
latest cd: Goh Lee Kwang & Tim Blechmann: Drone
http://www.geocities.com/gohleekwangtimblechmannduo/
After one look at this planet any visitor from outer space
would say "I want to see the manager."
William S. Burroughs
12 May
12 May
7:49 p.m.
On Thu, 2005-05-12 at 22:58 +0200, Tim Blechmann wrote:
hi all..
i'm trying to sleep for some very low time slices ... about 100 to 1000
us ...
but i can't get below about 1 ms ... are there any workarounds to sleep
for very small time slices?
You can't sleep less than 1ms because that is the tick rate of the timer
interrupt. For now the only universal "solution" is to busy-wait.
If your soundcard's timer is supported by the ALSA timer API (currently
only emu10k1 and ymfpci), you can get very good resolution: 1 tick at
48000Hz is 21 usecs. Otherwise the ALSA timer resolution will only be
as good as the system timer.
i tested nanosleep(), usleep() and select(0, 0, 0, 0,
&timout) ...
The real solution, which is likely to end up in the mainline kernel, is
George Anzinger's high res timers patch:
http://sourceforge.net/projects/high-res-timers/
High res timers have been a hot topic on LKML lately.
Lee
7:50 p.m.
i'm trying to sleep for some very low time slices
... about 100 to 1000
us ...
but i can't get below about 1 ms ... are there any workarounds to sleep
for very small time slices?
i tested nanosleep(), usleep() and select(0, 0, 0, 0, &timout) ...
on a normal kernel, there's no way around this. the only possibility
is with a kernel running the hi-res timer patch. even then, its only a
possibility.
--p
8:53 p.m.
On Thu, 2005-05-12 at 17:50 -0400, Paul Davis wrote:
Why do you say it's only a possibility? AFAIK the high res timer patch
works fine on most hardware...
Lee
i'm trying
to sleep for some very low time slices ... about 100 to 1000
us ...
but i can't get below about 1 ms ... are there any workarounds to sleep
for very small time slices?
i tested nanosleep(), usleep() and select(0, 0, 0, 0, &timout) ...
on a normal kernel, there's no way around this. the only possibility
is with a kernel running the hi-res timer patch. even then, its only a
possibility.
10:46 p.m.
Why do you say it's only a possibility? AFAIK the
high res timer patch
works fine on most hardware...
the last person i know to try (it was some time ago) was jesse
chappell, and it messed up other things on his system in subtle but
noticeable ways.
the kernel may also impose limits on user space processes requesting
microsleeps, but i don't know if this is actually done with the HRT
patch.
--p
7194
days inactive
7194
days old
linux-audio-dev@lists.linuxaudio.org
4 comments
3 participants
participants (3)
-
Lee Revell -
Paul Davis -
Tim Blechmann