Ultralite AVB
Moderators: MattKingUSA, khz
Re: Ultralite AVB
Got my 16A today, so could test myself. While initially i had the same problems as you (random glitches, channel movement etc), they immediately vanished after i removed the prebuffering (= sending in advance) of the iso out packets.
Note: The alsa usb driver does such prebuffering of iso out packaets for all devices that do not use an implicit feedback endpoint for synchronization, like e.g. the ultralite avb.
Note: The alsa usb driver does such prebuffering of iso out packaets for all devices that do not use an implicit feedback endpoint for synchronization, like e.g. the ultralite avb.
Re: Ultralite AVB
I tried removing the prebuffering from your script by setting PREBUFFER_PACKETS to 0 and it might have reduced the number of errors, but I'm certainly still getting them.
Re: Ultralite AVB
Unfortunately i was wrong. It worked, because i had hooked it up to a windows box, where it apparently initialized correctly and then reconnected it to the linux box without rebooting the device... The search continues...
-
- Established Member
- Posts: 59
- Joined: Sun Jan 14, 2018 8:38 pm
- Has thanked: 3 times
- Been thanked: 3 times
Re: Ultralite AVB
Just found out that I can switch the web interface into a mode where I can adjust the host buffer size and host safety offset.
Via JSON, there are parameters like:
Setting "host/os" from mac to win via instantly changes the device tab in the web interface.
Update:
Changing this didn't let me connect jack at all. Then dmesg had:. Using USB 3.0 let me connect again.
Via JSON, there are parameters like:
Code: Select all
"host/os":"mac",
"host/win/buffer_sizes_1x":"16:32:64:128:256:512:1024",
"host/win/safety_offsets_1x":"16:24:32:48:64:128:256",
"host/win/current_buffer_size_ 1x":128,
"host/win/current_safety_offset_1x":128,
"host/win/interrupt_capable":1,
"host/win/enable_interrupts":1,
Code: Select all
curl --data 'json={"value":"win"}' <Ultralite IP>/datastore/host/os
Update:
Changing this didn't let me connect jack at all. Then dmesg had:
Code: Select all
cannot submit urb 0, error -28: not enough bandwidth
Re: Ultralite AVB
More testing done. Like someone has written earlier, the device is extremly sensitive to gaps in the iso stream. My guess is it does not check the pointers of its internal ringbuffers for overflow/underflow. I run my test program for about an hour without any problems, with the irq of the host controller and test program pinned to the same processor core, which has been isolated from the rest of the system.
-
- Established Member
- Posts: 93
- Joined: Fri Aug 17, 2018 10:55 pm
- Has thanked: 36 times
- Been thanked: 5 times
Re: Ultralite AVB
Any ideas on how this issue might surmounted?Drumfix wrote:More testing done. Like someone has written earlier, the device is extremly sensitive to gaps in the iso stream...
Re: Ultralite AVB
Update:
will set the device into the windows driver mode where you have 64 channels at 44.1/48kHz, 32 channels at 88.2/96kHz and 24 channels at > 96kHz.
In this mode, the device is using interfaces 4 and 5 instead of 1 and 2 for streaming.
Using this mode i can run the device without the channels moving around with a modified version of my test program but i still get glitches after some time, so need to investigate more.
Code: Select all
curl --data 'json={"value":"win"}' <Ultralite IP>/datastore/host/os
curl --data 'json={"value":"USB2"}' <Ultralite IP>/datastore/host/type
In this mode, the device is using interfaces 4 and 5 instead of 1 and 2 for streaming.
Using this mode i can run the device without the channels moving around with a modified version of my test program but i still get glitches after some time, so need to investigate more.
-
- Established Member
- Posts: 59
- Joined: Sun Jan 14, 2018 8:38 pm
- Has thanked: 3 times
- Been thanked: 3 times
Re: Ultralite AVB
Looking at this thread http://www.motunation.com/forum/viewtop ... 14&t=65037 , there might even be other layers of problems or misconfigurations, like powersaving in particular, even on windows.
Have you guys with the new firmware tried setting (usb host controllers) tunables of powertop to "bad" (i.e. full freaking power consumption) ?
Have you guys with the new firmware tried setting (usb host controllers) tunables of powertop to "bad" (i.e. full freaking power consumption) ?
-
- Established Member
- Posts: 93
- Joined: Fri Aug 17, 2018 10:55 pm
- Has thanked: 36 times
- Been thanked: 5 times
Re: Ultralite AVB
I have seen this before in a few Windows DAW optimization guides, not just MOTU, although based on our experiences it's likely much more critical when using a MOTU interface! I've always changed the USB settings in Windows, but never bothered to before in Linux because my Focusrite 6i6 "just worked" without changing any settings.Musinux wrote:Looking at this thread http://www.motunation.com/forum/viewtop ... 14&t=65037 , there might even be other layers of problems or misconfigurations, like powersaving in particular, even on windows.
Have you guys with the new firmware tried setting (usb host controllers) tunables of powertop to "bad" (i.e. full freaking power consumption) ?
-
- Established Member
- Posts: 59
- Joined: Sun Jan 14, 2018 8:38 pm
- Has thanked: 3 times
- Been thanked: 3 times
Re: Ultralite AVB
https://www.gearslutz.com/board/music-c ... ce-87.html
Originally Posted by Rafter Man
There's a new update available. 1.2.7, but it appears the only changes are for Windows users adding the "Host Safety Offset" setting, aka USB Streaming Mode. Why only for Windows and not Mac as well?
--> The safety offset setting is not needed on OS X because Core Audio provides very stable timing. Our Mac driver uses a (very low) fixed safety offset.
Originally Posted by Rafter Man
There's a new update available. 1.2.7, but it appears the only changes are for Windows users adding the "Host Safety Offset" setting, aka USB Streaming Mode. Why only for Windows and not Mac as well?
--> The safety offset setting is not needed on OS X because Core Audio provides very stable timing. Our Mac driver uses a (very low) fixed safety offset.
In the pro audio driver readme, they say that 48 is a good starting point. I think my ultralite defaulted to 64. Have you tried increasing the safety offset?Drumfix wrote: Using this mode i can run the device without the channels moving around with a modified version of my test program but i still get glitches after some time, so need to investigate more.
Re: Ultralite AVB
I've tried setting the host safety offset to both extremes and to various values in between them, with no perceivable difference. This with the latest firmware.Musinux wrote:In the pro audio driver readme, they say that 48 is a good starting point. I think my ultralite defaulted to 64. Have you tried increasing the safety offset?
Re: Ultralite AVB
Has anyone else tried asking MOTU about this?
It looked like they were responsive over on gearslutz so I figured it might be worth pointing them to this thread. I'll report back if they're able to help.
It looked like they were responsive over on gearslutz so I figured it might be worth pointing them to this thread. I'll report back if they're able to help.
Re: Ultralite AVB
I got a suggestion to try "v1.2.4+201 (Release Date 2015-12-01)" and a reminder that MOTU doesn't officially support Linux which is about what I expected.