LinuxMusicians

Hello!

I'm thinking of getting a new audio interface. At the time I have a Line 6 UX1 and I can't really get below a latency of 34.8msec in Jack without getting xruns. I'll keep testing various things, but I'm starting to think it's a hardware limitation. I've been living with it for awhile now, and I've gotten my moneys worth out of it, so the thought of replacing it is getting more and more alluring.

From my limited researching, these are what I've found thus far:
Focusrite Scarlett 2i2 MK3
Focusrite Scarlett 8i6 MK3
Yamaha AG06
Behringer U-Phoria UMC404HD (or 204HD)

And I'm leaning mostly towards the Scarlett 8i6 MK3, seems like I'd be able to live with that one for some years. The MIDI input is a plus, even if it isn't really needed today.

My biggest criteria has been:
Latency below (~around) 10ms.
Not to expensive (student and limited needs really)
Linux support (duh...
2 inputs
headphones output

Any thoughts on why to go for one or the other? Or any other recommendations?

I appreciate the help! (and sorry for my first thread being a recommendation thread, I'll make sure for it to not be my last one!)

Hi gammtrall. Welcome to the forum.

I have found getting a low latency depends mostly on configuration, not hardware.

I have a PCI card and a USB 1.1 interface. The PCI card can use a buffer as small as 16 samples. This puts a huge strain on the CPU so I can't do much with that. I can use, for example, Carla and Tal NoiseMaker at this low latency but increasing the DSP load leads to Xruns.

The USB 1.1 (i.e. totally rubbish) interface can go as low as 64 samples. JACK will not start with a buffer smaller than that. That's the hardware limitation. Still that gives a latency of 1.3 ms at 48 kHz.

New hardware may not improve latency. It depends on the DSP load at which Xruns are happening.

gammtrall wrote:Hello!

I'm thinking of getting a new audio interface. At the time I have a Line 6 UX1 and I can't really get below a latency of 34.8msec in Jack without getting xruns. I'll keep testing various things, but I'm starting to think it's a hardware limitation. I've been living with it for awhile now, and I've gotten my moneys worth out of it, so the thought of replacing it is getting more and more alluring.

From my limited researching, these are what I've found thus far:
Focusrite Scarlett 2i2 MK3
Focusrite Scarlett 8i6 MK3
Yamaha AG06
Behringer U-Phoria UMC404HD (or 204HD)

And I'm leaning mostly towards the Scarlett 8i6 MK3, seems like I'd be able to live with that one for some years. The MIDI input is a plus, even if it isn't really needed today.

My biggest criteria has been:
Latency below (~around) 10ms.
Not to expensive (student and limited needs really)
Linux support (duh...
2 inputs
headphones output

There is no such thing as Linux support.

Most companies are even unwilling to do platform independency right.

Look for Class Compliant interfaces. They usually work. But only if they can be controlled completely without software.

- Scarlett Solo, 2i2, 2i4 are ok up to 2nd Gen. 3rd Gen seem to work but can't be controlled completely without Windows or Mac.
- MOTU UltraLite AVB was working with older firmware. The builtin mixer can be used with a Web Browser. But beware, they repleced the AVB by the similar looking AVB ES which cannot load older firmware. So it's useless until a firmware version is shipped that fixes the modulo-8 output channel hopping.

bluebell wrote:There is no such thing as Linux support.

That depends on what you mean by "support"... It might be enough to know that a given interface will work on Linux, and that there are people on this forum who have the same interface and can help if things go sideways.

I have the Behringer U-Phoria UMC204HD. I've not used it extensively for recording (yet) but my limited testing with Mixbus has been fine.

I use it as my (main) audio interface with Jack on the desktop generally as well and it's been great.

Michael Willis wrote:

bluebell wrote:There is no such thing as Linux support.

That depends on what you mean by "support"... It might be enough to know that a given interface will work on Linux, and that there are people on this forum who have the same interface and can help if things go sideways.

Linux support = "We make it platform independent". That's enough for my lowered expectations. But there are only some small interfaces that are "supported" in that way and some manufacturers like Focusrite seem to make huge steps backwards. They don't want to sell hardware that simply works. They want your data, they want you on their leash.

As someone working as a professional free/libre/open-source software developer for more than twenty years now, I can sympathize with the developers at manufacturers like Focusrite: they would likely be pleased to support Linux users, but can't justify devoting limited resources to it. I don't think we need to posit evil motives to explain their inability to support Linux: among other things, there isn't a well-defined target for exactly what "Linux support" looks like!

Major kudos to those who have been able and willing to hack through the brush to make their devices work anyway!

I completely disagree.

tseaver wrote: they would likely be pleased to support Linux users, but can't justify devoting limited resources to it

Haha, nice joke.

The reason those companies don't support linux is their paranoia that giving away any information on how their devices work may either be used by a competitor -> advantage for them, or will show that they are infringing the rights of someone else. This is why they, provided they outsorced the driver development, give out that information only per NDA.

There was the "Linux driver project" https://blog.slyon.de/2007/09/30/linux- ... project-2/ (german language) which failed due to exactly this reasons.

tseaver wrote: there isn't a well-defined target for exactly what "Linux support" looks like!

There was and still is. See the Linux driver project.

if such a company goes bankrupt, my vertict: You got what you deserved!

Drumfix wrote: The reason those companies don't support linux is their paranoia that giving away any information on how their devices work may either be used by a competitor -> advantage for them, or will show that they are infringing the rights of someone else. This is why they, provided they outsorced the driver development, give out that information only per NDA.

I think it's even worse. Look at Focusrite. They actively redesign their stuff to make people registering their hardware. Look at the Mass Storage Device mode of the Gen3 Scarletts and read magazines where those Gen3 Scarletts are "tested": Writers praise Focusrite because it's so easy to register now. Sure those magazines cannot blame Focusrite for trying to lock-in and control their customers, else they won't sell ads to them anymore.

What seems to be the worst case for such a manufacturer? Selling interfaces and getting no personal data from their customers. It's much more fun for them having customers running on a leash. Their wet dreams: "Sorry, customer, you didn't install the recent version of our sniffingware on your PC, we have to disable your interface." That's a look into the future. Now it's like "Register and you get plugins worth of $250,000,000!"

What is your maximum price you want to invest in audio/MIDI interface?

khz wrote:What is your maximum price you want to invest in audio/MIDI interface?

Up to 1000 EUR if it has a good internal routing and is fully controllable with a browser, MIDI or hardware knobs.

My MOTU UltraLite AVB was at 649,- EUR and is nearly perfect. If MOTU releases new firmware that fixes the modulo-8 output and input channel hopping then it's perfect.

My suggestion would be, for example, an RME AIO.

As for the whole "linux support" deal. I realize I chose the wrong word. What I meant was "works under Linux".

merlyn wrote:Hi gammtrall. Welcome to the forum.

I have found getting a low latency depends mostly on configuration, not hardware.

I have a PCI card and a USB 1.1 interface. The PCI card can use a buffer as small as 16 samples. This puts a huge strain on the CPU so I can't do much with that. I can use, for example, Carla and Tal NoiseMaker at this low latency but increasing the DSP load leads to Xruns.

The USB 1.1 (i.e. totally rubbish) interface can go as low as 64 samples. JACK will not start with a buffer smaller than that. That's the hardware limitation. Still that gives a latency of 1.3 ms at 48 kHz.

New hardware may not improve latency. It depends on the DSP load at which Xruns are happening.

Thanks for the tips. I'll have time to try out a lot of things before I'm able to purchase a new interface, so I'll keep at it. The weird thing is that my DSP load seems to stay around 50% and I still get xruns. But I haven't had time to try out anything more lately.

bluebell wrote:Look for Class Compliant interfaces. They usually work. But only if they can be controlled completely without software.

- Scarlett Solo, 2i2, 2i4 are ok up to 2nd Gen. 3rd Gen seem to work but can't be controlled completely without Windows or Mac.
- MOTU UltraLite AVB was working with older firmware. The builtin mixer can be used with a Web Browser. But beware, they repleced the AVB by the similar looking AVB ES which cannot load older firmware. So it's useless until a firmware version is shipped that fixes the modulo-8 output channel hopping.

Thank you!

khz wrote:What is your maximum price you want to invest in audio/MIDI interface?

I'd like to stay around 200usd. However I'm willing to save up more, but beyond 400usd is too much.

The Behringer UMC204HD or UMC404HD will work out of the box on Linux. I have the UMC1820 and have been pleased with it.

GuntherT wrote: ↑Mon Feb 24, 2020 10:10 pm The Behringer UMC204HD or UMC404HD will work out of the box on Linux. I have the UMC1820 and have been pleased with it.

Thanks for the input (: