20 Aug
2013
20 Aug
'13
1:52 p.m.
Hi,
this is just another rt-kernel thread about RT performance.
I tried latest 3.10.6 with some issues so I decided to use a long term
release and I am now compiling 3.2.50.
So. Which kernel release are you running?
/r
20 Aug
20 Aug
2:38 p.m.
Hi Raffaele :)
On Tue, 2013-08-20 at 13:52 +0200, Raffaele Morelli wrote:
I tried latest 3.10.6 with some issues
Which kernel release are you running?
I made my last unfinished audio production using 3.8.4-rt2 x86_64 and
3.8.11-rt8 x86_64, both seemed to be ok.
Currently I'm running 3.8.13-rt14, but for daily non-audio work only, so
I don't know if there are issues for audio production, for non-audio
usage it's ok.
With 3.10.6-rt3 installed lightdm already hangs when checking the
password to log in. The non-rt 3.10.7 does work ok and 3.10.6 did work
ok too, but I didn't use/don't use the non-rt very often.
This is for my "main" Linux, I've got other installs running different
kernels too.
Regards,
Ralf
11:39 p.m.
On Tue, Aug 20, 2013 at 12:52 PM, Raffaele Morelli <
raffaele.morelli(a)gmail.com> wrote:
So. Which kernel release are you running?
Ex-Arch-audio "production" repo kernel: 3.6.11-rt25-1-rt #1 SMP PREEMPT RT
Tue Dec 25 00:18:17 CET 2012 x86_64 GNU/Linux
I've tuned this system extensively (manually testing IRQ's, various
blacklisted kernel modules, cpu scaling performance etc.
Details available here:
http://openavproductions.com/real-time-latency-tuning
The system runs at 0.3 ms without Xruns (with a CPU stress tester running
on both cores) on an Echo Indigo DJ PCMCIA card.
So long as the CPU can keep up the process load (and the clients are RT
safe) I don't think it will xrun.
I'm usually running at 5ms, just to reduce the CPU load...
HTH, -Harry
24 Aug
24 Aug
6:44 a.m.
Am 20.08.2013 23:39, schrieb Harry van Haaren:
On Tue, Aug 20, 2013 at 12:52 PM, Raffaele Morelli
<raffaele.morelli(a)gmail.com <mailto:raffaele.morelli@gmail.com>> wrote:
> So. Which kernel release are you running?
I've run 3.6.1-rt1 for some time, without any issue, yesterday I've
switched to 3.10.9-rt5 which run flawless up to now.
6:59 a.m.
Did you enable dynticks?
2013/8/24 hermann meyer <brummer-(a)web.de>
Am 20.08.2013 23:39, schrieb Harry van Haaren:
On Tue, Aug 20, 2013 at 12:52 PM, Raffaele Morelli <
raffaele.morelli(a)gmail.com> wrote:
So. Which kernel release are you running?
I've run 3.6.1-rt1 for some time, without any issue, yesterday I've
switched to 3.10.9-rt5 which run flawless up to now.
7:58 a.m.
Am 24.08.2013 06:59, schrieb Raffaele Morelli:
Did you enable dynticks?
No, as far I can see the /CONFIG_NO_HZ_FULL option isn't available, /
2013/8/24 hermann meyer <brummer-(a)web.de
<mailto:brummer-@web.de>>
Am 20.08.2013 23:39, schrieb Harry van Haaren:
On Tue, Aug 20, 2013 at 12:52 PM, Raffaele
Morelli
<raffaele.morelli(a)gmail.com <mailto:raffaele.morelli@gmail.com>>
wrote:
> So. Which kernel release are you running?
I've run 3.6.1-rt1 for some time, without any issue, yesterday
I've switched to 3.10.9-rt5 which run flawless up to now.
10:19 a.m.
On Sat, 2013-08-24 at 07:58 +0200, hermann meyer wrote:
Am 24.08.2013 06:59, schrieb Raffaele Morelli:
We should test it. Unfortunately I couldn't test it, since the kernel
with the rt3 patch didn't run on my machine, I'll test the new
patch/kernel ASAP.
Did you enable dynticks?
No, as far I can see the CONFIG_NO_HZ_FULL option isn't available
1:39 p.m.
On Sat, 2013-08-24 at 10:19 +0200, Ralf Mardorf wrote:
On Sat, 2013-08-24 at 07:58 +0200, hermann meyer
wrote:
> Am 24.08.2013 06:59, schrieb Raffaele Morelli:
>
> > Did you enable dynticks?
> >
> >
> >
> No, as far I can see the CONFIG_NO_HZ_FULL option isn't available
^^^^^^^^^^^^^^^
Oops, didn't read carefully, it's not _available_ anymore?
We should test it. Unfortunately I couldn't test
it, since the kernel
with the rt3 patch didn't run on my machine, I'll test the new
patch/kernel ASAP.
2:13 p.m.
Am 24.08.2013 10:21, schrieb Ralf Mardorf:
On Sat, 2013-08-24 at 10:19 +0200, Ralf Mardorf
wrote:
No, don't anymore, the opposite, . .
CONFIG_NO_HZ is available, but CONFIG_NO_HZ_FULL still isn't (if I
didn't overlook it in the config)
http://www.phoronix.com/scan.php?page=news_item&px=MTM2NTA
On Sat, 2013-08-24 at 07:58 +0200, hermann meyer
wrote:
> Am 24.08.2013 06:59, schrieb Raffaele Morelli:
>
>> Did you enable dynticks?
>>
>>
>>
> No, as far I can see the CONFIG_NO_HZ_FULL option isn't available
^^^^^^^^^^^^^^^
Oops, didn't read carefully, it's not _available_ anymore?
> We should test it. Unfortunately I couldn't
test it, since the kernel
> with the rt3 patch didn't run on my machine, I'll test the new
> patch/kernel ASAP.
2:25 p.m.
On Sat, 2013-08-24 at 14:13 +0200, hermann meyer wrote:
CONFIG_NO_HZ is available, but CONFIG_NO_HZ_FULL still
isn't
But it was. From a private mail a while ago:
On Tue, 2013-08-13 at 19:01 +0200, Joakim Hernberg wrote:
jack@tor /home/jack $ uname -a
Linux tor 3.10.6-rt3-1-rt #1 SMP PREEMPT RT Tue Aug 13 13:59:13 CEST
2013 x86_64 GNU/Linux
jack@tor /home/jack $ zgrep _HZ /proc/config.gz
CONFIG_NO_HZ_COMMON=y
# CONFIG_HZ_PERIODIC is not set
CONFIG_NO_HZ_IDLE=y
# CONFIG_NO_HZ_FULL is not set
CONFIG_NO_HZ=y
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
jack@tor /home/jack $
I was willing to test the kernel when it's enabled, but the kernel did
cause an issue on my machine already when it was disabled.
2:46 p.m.
Am 24.08.2013 14:25, schrieb Ralf Mardorf:
On Sat, 2013-08-24 at 14:13 +0200, hermann meyer
wrote:
Here is what I get
uname -a
Linux box 3.10.9-rt5 #1 SMP PREEMPT RT Fri Aug 23 11:14:14 CEST 2013
i686 GNU/Linux
zgrep _HZ /proc/config.gz
CONFIG_HZ_PERIODIC=y
# CONFIG_NO_HZ_IDLE is not set
# CONFIG_NO_HZ is not set
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
This is plain vanila +rt patch.
CONFIG_NO_HZ is available, but CONFIG_NO_HZ_FULL
still isn't
But it was. From a private mail a while ago:
On Tue, 2013-08-13 at 19:01 +0200, Joakim Hernberg wrote:
jack@tor /home/jack $ uname -a
Linux tor 3.10.6-rt3-1-rt #1 SMP PREEMPT RT Tue Aug 13 13:59:13 CEST
2013 x86_64 GNU/Linux
jack@tor /home/jack $ zgrep _HZ /proc/config.gz
CONFIG_NO_HZ_COMMON=y
# CONFIG_HZ_PERIODIC is not set
CONFIG_NO_HZ_IDLE=y
# CONFIG_NO_HZ_FULL is not set
CONFIG_NO_HZ=y
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
jack@tor /home/jack $
I was willing to test the kernel when it's enabled, but
the kernel did
cause an issue on my machine already when it was disabled.
3:03 p.m.
Quoting hermann meyer <brummer-(a)web.de>de>:
Am 24.08.2013 14:25, schrieb Ralf Mardorf:
There are some reasons for not having to set some of these high
resolution timers, especially if you are using MIDI through Jack. See
-
http://www.linuxmusicians.com/viewtopic.php?f=27&t=10058&p=44078#p4…
On Sat, 2013-08-24 at 14:13 +0200, hermann meyer
wrote:
Here is what I get
uname -a
Linux box 3.10.9-rt5 #1 SMP PREEMPT RT Fri Aug 23 11:14:14 CEST 2013
i686 GNU/Linux
zgrep _HZ /proc/config.gz
CONFIG_HZ_PERIODIC=y
# CONFIG_NO_HZ_IDLE is not set
# CONFIG_NO_HZ is not set
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
This is plain vanila +rt patch.
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
http://lists.linuxaudio.org/listinfo/linux-audio-user
CONFIG_NO_HZ is available, but CONFIG_NO_HZ_FULL
still isn't
But it was. From a private mail a while ago:
On Tue, 2013-08-13 at 19:01 +0200, Joakim Hernberg wrote:
jack@tor /home/jack $ uname -a
Linux tor 3.10.6-rt3-1-rt #1 SMP PREEMPT RT Tue Aug 13 13:59:13 CEST
2013 x86_64 GNU/Linux
jack@tor /home/jack $ zgrep _HZ /proc/config.gz
CONFIG_NO_HZ_COMMON=y
# CONFIG_HZ_PERIODIC is not set
CONFIG_NO_HZ_IDLE=y
# CONFIG_NO_HZ_FULL is not set
CONFIG_NO_HZ=y
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
jack@tor /home/jack $
I was willing to test the kernel when it's enabled, but
the kernel did
cause an issue on my machine already when it was disabled.
4:15 p.m.
On Sat, 2013-08-24 at 13:03 +0000, Barney Holmes wrote:
Thank you. I'll test it. I startet enabling HR timer a long time ago,
because I got less MIDI jitter, but nowadays I get many xruns, when
using another sound card, then I used in the past. Perhaps I get rid of
some xruns, if I disable hr timer.
25 Aug
25 Aug
6:51 a.m.
Am 24.08.2013 15:03, schrieb Barney Holmes:
Quoting hermann meyer <brummer-(a)web.de>de>:
Aha, found out why I haven't the CONFIG_NO_HZ_FULL option,
it is marked as 64bit only, but I run i686,
https://groups.google.com/forum/#!topic/linux.kernel/nq3HSPY3-Vg
Am 24.08.2013 14:25, schrieb Ralf Mardorf:
On Sat, 2013-08-24 at 14:13 +0200, hermann meyer
wrote:
Here is what I get
uname -a
Linux box 3.10.9-rt5 #1 SMP PREEMPT RT Fri Aug 23 11:14:14 CEST 2013
i686 GNU/Linux
zgrep _HZ /proc/config.gz
CONFIG_HZ_PERIODIC=y
# CONFIG_NO_HZ_IDLE is not set
# CONFIG_NO_HZ is not set
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
This is plain vanila +rt patch.
CONFIG_NO_HZ is available, but CONFIG_NO_HZ_FULL
still isn't
But it was. From a private mail a while ago:
On Tue, 2013-08-13 at 19:01 +0200, Joakim Hernberg wrote:
jack@tor /home/jack $ uname -a
Linux tor 3.10.6-rt3-1-rt #1 SMP PREEMPT RT Tue Aug 13 13:59:13 CEST
2013 x86_64 GNU/Linux
jack@tor /home/jack $ zgrep _HZ /proc/config.gz
CONFIG_NO_HZ_COMMON=y
# CONFIG_HZ_PERIODIC is not set
CONFIG_NO_HZ_IDLE=y
# CONFIG_NO_HZ_FULL is not set
CONFIG_NO_HZ=y
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000
jack@tor /home/jack $
I was willing to test the kernel when it's enabled, but
the kernel did
cause an issue on my machine already when it was disabled.
- it's marked 64-bit only at the moment - the
32-bit support patch is
small but did not get ready in time.
There are some reasons for not having to set some of
these high
resolution timers, especially if you are using MIDI through Jack. See
- http://www.linuxmusicians.com/viewtopic.php?f=27&t=10058&p=44078#p4…
24 Aug
24 Aug
2:41 p.m.
On Sat, 2013-08-24 at 14:13 +0200, hermann meyer wrote:
if I didn't overlook it in the config
It is in this config for 3.10.9-rt5:
https://aur.archlinux.org/packages/linux-rt/
When I'll build the kernel within the next days I'll enable it. Arch
users who'll build it too, please enable it for testing purpose, it's
disabled by default.
26 Aug
26 Aug
3:16 p.m.
On 08/20/2013 01:52 PM, Raffaele Morelli wrote:
Hi,
this is just another rt-kernel thread about RT performance.
I tried latest 3.10.6 with some issues so I decided to use a long term
release and I am now compiling 3.2.50.
So. Which kernel release are you running?
/r
3.2.48-rt69 but I'll compile 3.2.50 one of these days. Works good enough
for me.
Jeremy
3:19 p.m.
2013/8/26 Jeremy Jongepier <jeremy(a)autostatic.com>
On 08/20/2013 01:52 PM, Raffaele Morelli wrote:
3.2.50-rt70 works very good
Hi,
this is just another rt-kernel thread about RT performance.
I tried latest 3.10.6 with some issues so I decided to use a long term
release and I am now compiling 3.2.50.
So. Which kernel release are you running?
/r
3.2.48-rt69 but I'll compile 3.2.50 one of these days. Works good enough
for me.
Jeremy
4166
days inactive
4172
days old
linux-audio-user@lists.linuxaudio.org
17 comments
7 participants
participants (7)
-
Barney Holmes -
Diego Simak -
Harry van Haaren -
hermann meyer -
Jeremy Jongepier -
Raffaele Morelli -
Ralf Mardorf