[Marveric Beta] Result of testing some firewire devices

[Ronan Jouchet]

Hi,

Same good news here: I got to 8ms (possibly lower, I have to test) /
no xruns with the generic kernel, on my crappy test laptop + TI PCMCIA
card!
I'll do more testing with the -lowlatency and -realtime kernels, as
well as with a comparison with a ricoh chipset, but it looks like a
solid release is coming.

To the newcomers in this discussion, there is a great wiki page
guiding you to choose your kernel :
  https://help.ubuntu.com/community/UbuntuStudio/RealTimeKernel
And the usual preparation page :
  https://help.ubuntu.com/community/UbuntuStudioPreparation

Cheers,
Ronan

On Tue, Sep 28, 2010 at 7:37 AM, Ricardo Lameiro
<ricardolameiro at gmail.com> wrote:
> Hi Takashi et al
>
> I also tested Firewire devices on the Ubuntustudio Beta.
> My device is a Edirol FA-101.
> What I can say is that it worked out the box, just configuring it right in
> Jack.
> I had almost no XRUNS (ocasional XRUN when opening a jack app like Ardour,
> its normal) using the generic kernel.
> This is a great improvement since past setups, since it was almost mandatory
> to use a RT kernel with Firewire devices.
> True to be told, I was working at 16ms latency, maybe with the old stack, I
> could go to 8 ms and less XRUNS.
> There is also a point that most of users sometimes miss, the hardware...
> The Firewire chipset is a key element for the stability of a good and stable
> stream of data on the FW bus. this is not only related to Linux, but to all
> the operative systems, and it is recommended, by specialist the usage of
> Texas instruments Firewire chipsets (TI).
> Unfortunately, I have a Ricoh chipset on my laptop, but one of this days I
> will try to find a pcmcia card with a TI chipset.
>
> I think that reporting devices not working on the new stack should be
> reported not only to ubuntu, but also to upstream FFADO. If FFADO fixes some
> driver, then everyone will benefit from the changes.
>
> greetings
>
> 2010/9/28 Takashi Sakamoto <o-takashi at sakamocchi.jp>
>>
>> Hi,
>>
>> I test my firewire audio device, M-Audio Firewire Solo and MOTU 828 Mk2
>> on Ubuntu Studio Marveric Beta. And I found some changes from Marveric.
>>
>> 1. default jackd is version 2.
>> As you know. We can change jackd2 to jackd1 in synaptic exclusively.
>>
>> 2. New firewire stack (?) is default
>> The default kernel modules for firewire is firewire_ohci, firewire_core,
>> crc_itu_t as a result of running "lsmod".
>> In lucid, these are raw1394, ohci1394, ieee1394.
>> The developper of FFADO seem to call the former as "new firewire stack"
>> and latter as "old firewire stack".
>>
>> I read some documents of FFADO and they reccomend to use old firewire
>> stack. But it seems to be old for me. So I test my devices in both stacks.
>>
>> 2. both of my device work but...
>>
>> I can connect jackd2 to my M-Audio FW Solo with both stacks but to 828
>> Mk2 only with old firewire stack. I use "rmmod" and "modprobe" commands
>> much time in this operation.
>>
>> M-Audio Firewire Solo works fine in both stacks.
>> MOTU 828 Mk2 works in old stack but the sound is the worst.
>>
>> 3. new stack makes more FFADO XRUNs than old stack
>>
>> Then I use Renoise 2.6.6 (Beta) because this version can run under
>> jackd2 in the same device (M-Audio FW Solo). I played the same sample
>> music in both stacks. As a result, I got more FFADO XRUNs than old stack.
>>
>> You can see my report in launchpad.
>>
>> MOTU 828 Mk2 works but the sound is noisy, like buzzing.
>> https://bugs.launchpad.net/ubuntu/+source/libffado/+bug/649542
>>
>> New firewire stack makes more FFADO XRUNs than old firewire stack.
>> https://bugs.launchpad.net/ubuntu/+source/libffado/+bug/649583
>>
>> Please post to this mailing list if you know the other changes between
>> lucid and marveric.
>>
>> Thanks
>>
>> Takashi Sakamoto
>> o-takashi at sakamocchi.jp
>>