Kernel-lowlatency vs kernel-rt
Ralf Mardorf
ralf.mardorf at alice-dsl.net
Sat Nov 3 09:06:00 UTC 2012
On Sat, 3 Nov 2012 01:19:53 +0100
ttoine <ttoine at ttoine.net> wrote:
> Ralf,
>
> If you don't need restricted drivers, -rt kernel is better than the
> -lowlatency. For audio, and for midi. If you need the restricted
> driver of, for example, nvidia or amd/ati graphic cards, -rt may not
> work. That's why there is the -lowlatency kernel.
>
> Toine
Hi Toine :)
yes, but I'm mistaken regarding to the issues on my machine. Randomly
the kernel-rt worked better for a while, but now I get xruns again too.
Seemingly sharing IRQs with a sound card is absolutely bad.
On Sat, 03 Nov 2012 01:30:40 +0100
"Kaj Ailomaa" <zequence at mousike.me> wrote:
> Did you compare with a lowlatency of the same kernel version?
> If not, the test is not very conclusive.
>
> If you like to see some good results from a lowlatency kernel, try
> 2.6.37, and perhaps also 2.6.38 and 2.6.39
Hi Kaj :)
with kernels < 2.6.39 (ALSA < 1.0.24) my card doesn't work and for 2.6
kernels at least the rtirq confg needs to be rewritten.
On Sat, 3 Nov 2012 01:43:58 +0100
ttoine <ttoine at ttoine.net> wrote:
> I am running 3.5.0.17 -lowlatency, I think from KXstudio repositories.
> Works great with Ubuntu 12.10.
It's the default kernel for Ubuntu Studio Quantal.
On Sat, 3 Nov 2012 09:45:06 +0800
Ho Wan Chan <smartboyhw at gmail.com> wrote:
> -lowlatency kernels aren't from KXStudio, it is in the official Ubuntu
> repositories, and we include it in our live images. Andy Whitcroft of
> Canonical is the main maintainer of the -lowlatency kernels, though I
> think our kernel team (mainly zequence) wants to maintain the precise
> kernels.
>
> -rt kernels are from the KXStudio repos. However I use -lowlatency
> kernel 3.5.0.17, and it works great.
Hi Ho :)
or we build our kernels simply ourself, see attachment.
Hi Len :)
you're right, but at the moment I use the RME HDSPe without ADAT, just
for a small audio sessions. Because I still get xruns using the
kernel-rt I guess I'm lost. There seems to be nothing I can do against
a shared IRQ. The current issue perhaps isn't caused by Linux, but by
my motherboard.
On Fri, 2 Nov 2012 21:56:36 -0700
"Len Ovens" <len at ovenwerks.net> wrote:
> Correct me if I am wrong, but are you not running two M-audio cards
> for midi as well? I think if I remember right, they are ice1712
> cards. Does the midi part of it need the ice1712 kernel driver? That
> is, would you still get midi if the ice1712 driver was black listed
> or unloaded? I'm just wondering if ALSA is using cpu time servicing
> 20 inputs at 48000 hrz for something you are not using. (24bits of
> silence is still 24 bits of throughput and the ice chips are 10
> channels each) I don't know how ALSA handles unused (unconnected)
> ports. I do know that jack and pulse take just as much time dealing
> with all the ports even the ones you don't happen to be using.
Two TerraTec EWX 24/96, but they are Envy24 too. A good idea. Btw. top
doesn't show the ATI graphics when running the MIDI/audio session.
The ATI graphics shares the IRQ with the RME. Some ohci device very
often is shown. Since nothing else is connected to USB at the moment,
it must be the mouse. I'll connect it to another USB port. Right now
I'm not using the TerraTec cards, removing the module on the quick
doesn't work:
spinymouse at qrc:~$ sudo rmmod snd-ice1712
spinymouse at qrc:~$ modprobe -l | grep 1712
kernel/sound/pci/ice1712/snd-ice1712.ko
kernel/sound/pci/ice1712/snd-ice17xx-ak4xxx.ko
kernel/sound/pci/ice1712/snd-ice1724.ko
I'll blacklist snd-ice1712. I also don't need firewire, wifi and
bluetooth. I don't know the modules names, but I'll find out their
names and unload or blacklist them.
Perhaps the mouse is the culprit. When I didn't own the RME,
but only the TerraTec's, they 1. didn't share their IRQs and 2. I used
a PS/2 mouse, so another USB slot, with another IRQ might improve
something.
spinymouse at qrc:~$ cat /proc/interrupts
CPU0 CPU1
0: 132 0 IO-APIC-edge timer
1: 3 5862 IO-APIC-edge i8042
7: 1 0 IO-APIC-edge parport0
8: 0 1 IO-APIC-edge rtc0
9: 0 0 IO-APIC-fasteoi acpi
14: 0 0 IO-APIC-edge pata_atiixp
15: 0 0 IO-APIC-edge pata_atiixp
16: 1 221477 IO-APIC-fasteoi ohci_hcd:usb2
17: 0 2 IO-APIC-fasteoi ohci_hcd:usb3,
ohci_hcd:usb5
18: 0 6549804 IO-APIC-fasteoi ohci_hcd:usb4,
ohci_hcd:usb6, snd_hdspm
spinymouse at qrc:~$ /etc/init.d/rtirq status
PID CLS RTPRIO NI PRI %CPU STAT COMMAND
62 FF 90 - 130 0.0 S irq/8-rtc0
806 FF 85 - 125 0.2 S irq/18-snd_hdsp
838 FF 85 - 125 0.0 S irq/20-snd_ice1
850 FF 85 - 125 0.0 S irq/21-snd_ice1
54 FF 80 - 120 0.0 S irq/19-ehci_hcd
55 FF 80 - 120 0.0 S irq/16-ohci_hcd
56 FF 79 - 119 0.0 S irq/17-ohci_hcd
58 FF 78 - 118 0.0 S irq/18-ohci_hcd
59 FF 78 - 118 0.0 S irq/17-ohci_hcd
60 FF 77 - 117 0.1 S irq/18-ohci_hcd
61 FF 75 - 115 0.0 S irq/1-i8042
24 FF 50 - 90 0.0 S irq/9-acpi
46 FF 50 - 90 0.0 S irq/22-ahci
232 FF 50 - 90 0.0 S irq/14-pata_ati
233 FF 50 - 90 0.0 S irq/15-pata_ati
274 FF 50 - 90 0.0 S irq/22-firewire
620 FF 50 - 90 0.0 S irq/7-parport0
780 FF 50 - 90 0.0 S irq/43-radeon
1104 FF 50 - 90 0.0 S irq/42-eth0
3 FF 1 - 41 0.1 S ksoftirqd/0
15 FF 1 - 41 0.1 S ksoftirqd/1
19: 0 2 IO-APIC-fasteoi ehci_hcd:usb1
20: 0 0 IO-APIC-fasteoi snd_ice1712
21: 0 0 IO-APIC-fasteoi snd_ice1712
22: 5 113762 IO-APIC-fasteoi ahci, firewire_ohci
42: 17275 24 PCI-MSI-edge eth0
43: 59 19419 PCI-MSI-edge radeon
Regards,
Ralf
PS: Thank you Len, I'll test if MIDI will work, after removing the envy
driver module and if I don't need the MIDI ports, I anyway could
blacklist the module.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: quantal-rt
Type: application/octet-stream
Size: 2706 bytes
Desc: not available
URL: <https://lists.ubuntu.com/archives/ubuntu-studio-devel/attachments/20121103/6e043ed1/attachment.obj>
More information about the Ubuntu-Studio-devel
mailing list