Hi,
I want to build my own audio recording system (mostly standard components). Genode seems a great and highly customizable alternative to blown up operating systems <insert one of the three main os here> with some extra packages. Also, it offers a low latency audio support on-top of alsa (good starting point).
So, the following random questions came into my mind:
1) Is the turmvilla scenario a good starting point? 2) What about the support for firewire-devices (e.g. audio interfaces)? 3) Does the e.g. Linux-kernel need to be patched to get lower latency? 4) Which PC architecture is supported (x86 & x86_64)? 5) What is the current state of the display-driver?
The general setup should be similar to this:
PC: Haswell CPU (i5, 2.6 GHz), 16 GB RAM, 512 GB SSD, Mini ITX Formfactor Equipment: Firewire Audio Interface Software: Linux-Kernel or base-hw, ALSA, FFADO, JACK?, Ardour4, ...
Thank you in advance.
Best regards Daniel
-- Daniel Krefft MSc. Wissenschaftlicher Mitarbeiter
Technische Universitaet Muenchen Fakultaet fuer Informatik Lehrstuhl fuer Betriebssysteme
Arcisstrasse 21 80333 Muenchen
Tel. +49 89 289 18791
daniel.krefft@...256... www.os.in.tum.de/personen/krefft
Hi Daniel,
I want to build my own audio recording system (mostly standard components). Genode seems a great and highly customizable alternative to blown up operating systems <insert one of the three main os here> with some extra packages. Also, it offers a low latency audio support on-top of alsa (good starting point).
as a note of caution, even though we have designed Genode's audio interfaces by paying attention to low latency as requirement, we haven't conducted real measurements and have not stressed those interfaces very intensively so far. But I agree that the current state already provides a nice playground.
* There are the 'Audio_in' and 'Audio_out' session interfaces, which make it fairly easy to chain audio components. * We have an audio driver ported from OpenBSD. * There is an audio mixer. * Several of the supported kernels provide hard priorities, which is useful to guarantee the timely scheduling of audio-processing components independent from unrelated (lower-priority) parts of the system.
So, the following random questions came into my mind:
- Is the turmvilla scenario a good starting point?
I don't think so. It may be good for inspiration but I would recommend you to build a scenario from scatch by including - step by step - only those components that you actually need.
For porting existing audio software, I would create a run script for each individual piece of software. So you can test and port each of part independently. Once you know that each of the individual parts work, you can integrate them into a bigger scenario.
- What about the support for firewire-devices (e.g. audio interfaces)?
There is no driver support for firewire yet.
- Does the e.g. Linux-kernel need to be patched to get lower latency?
Frankly, the Linux version of Genode is only meant as a development vehicle. It is great during development but for an actual deployment, I don't see the appeal of combining Genode with Linux. Have you considered to use of NOVA? This is the kernel that we use most often on x86 right now.
- Which PC architecture is supported (x86 & x86_64)?
Yes.
- What is the current state of the display-driver?
There are two display drivers for the PC platform available on Genode, a VESA driver (based of X.org's xf86emu) and the Intel graphics driver ported from the Linux kernel.
Cheers Norman