←back to thread

290 points XzetaU8 | 4 comments | | HN request time: 0s | source
Show context
varbhat ◴[] No.44657995[source]
Until the time when Microsoft realises this and creates a privileged API just for Microsoft Recall so that It can see the screen.

Better switch to Linux. It's not perfect but I am sure that you will be fine using Linux(Unless you want to use Adobe Suite or Few Corporate applications which won't be used by many)

replies(5): >>44658202 #>>44658215 #>>44658253 #>>44658615 #>>44659205 #
glimshe ◴[] No.44658202[source]
Or most professional audio applications...
replies(3): >>44658390 #>>44658399 #>>44658470 #
louthy ◴[] No.44658390[source]
Bitwig works on Linux, but the problem I had was that my pro-audio soundcard [1] didn’t have supported drivers and I couldn’t get the open source drivers to work. I tried switching to a Dante based solution: none of the Dante based apps worked, so I tried AES67 (open source Dante), still no joy — I just could not get my Dante/AES67 AD/DA converters (which attach to everything in my studio) to be ‘seen’ on Linux.

So after weeks trying to get a high-channel count I/O solution working, I gave in, I found the best thing to do was to just get a M4 Mac Mini for my audio/studio work. And leave Linux for everything else. I was setup within an hour on macOS.

There’s unfortunately still too much resistance and it can cost $1000s trying to get to a working solution or ultimately in my case: a non-working solution. It cost me about $6000 trying various options — not all wasted, but still, not cheap to find out that nothing works.

[1] https://rme-audio.de/hdspe-madi-fx.html

replies(4): >>44658800 #>>44659695 #>>44661062 #>>44661378 #
1. dfedbeef ◴[] No.44659695{3}[source]
The reliable way to do this (I found) is check the kernel source tree first. The supported pro sound cards are, typically, kind of old. Because FOSS developers aren't just gifted hardware and documentation to write the drivers so they're a generation or two behind.

Counterintuitively; using the latest kernel can be more stable as bug fixes are merged.

RME does have a few supported cards (I use one) but they're mostly the ADAT ones. And the driver is in-tree.

replies(1): >>44660264 #
2. louthy ◴[] No.44660264[source]
> The reliable way to do this (I found) is check the kernel source tree first

Sure, in general that's good advice, but it becomes more complicated depending on the solution/situation...

I’d bought the RME card long before I was hoping to make it work with Linux. I'd been running Windows for a long time for work reasons, so I had my dev work and my music setup on the same computer (a 64 core Threadripper machine, with 128gb of RAM, and fast NVMe drives). A few months before, I'd sold my company, so for work at least I didn't need to be on Windows any more. Then I started getting random audio dropouts! Presumably because of all the crap Microsoft keep loading onto the OS with after every update.

The audio dropouts was the straw that broke the camel’s back. If a machine like that, with nothing else running on it other than my DAW, could start having audio dropouts, then you know something has gone horribly wrong with the OS.

That's why I wanted to get my existing RME card working on Linux. When I wasn’t able to use it, I then assumed I’d be able to get a network based protocol running (Dante/AES67). There was plenty of discussion about it online, it seemed viable, and it's a network, Linux can do networking! Also, I kinda like the idea of network based audio, I think it's likely to be more future proof.

So, I replaced one of my Ferrofish A32 Pro interfaces with a Ferrofish A32 Pro DANTE ($4300) [1]. It supports both Dante and AES67. I figure if I can’t get Dante running then the open protocol AES67 (with support in Linux-land) should work. That didn’t feel that risky. But no amount of finagling would make the interface appear via the virtual sound-card/router concept.

This had already taken weeks (maybe months) to not get anywhere, so I looked for a Class Compliant sound-card (or, one that definitely had Linux drivers) that could support the number of channels I needed (96 channels in and out), it also needed to support the AD/DA interfaces interfaces I already had (so connectivity via MADI or Dante/AES67), but there just wasn't anything. The only other sound-card out there was another RME interface.

So, that’s when I opted for a Class Compliant sound-card [2] for casual use on Linux ($324) and a new RME Digiface Dante sound-card ($1543) [3] that I could use with a newly purchased M4 Mac Mini ($3000). I also needed to replace another one of my Ferrofish A32 Pro interfaces with another Ferrofish A32 Pro Dante ($4300) to make the setup work.

I realise now that my earlier estimate of $6000 was wildly out, it cost $13467 to leave Windows and to get an alternative pro-audio setup working. There may well have been alternative approaches and I may well have missed a possible solution that could have either worked with the original RME card (which would cost nothing) or AES67 (that would still require me to replace 3 x Ferrofish A32 Pro interfaces, so would end up about the same cost), but I felt like I'd been pretty thorough.

I guess the reason I'm writing War & Peace here is that it's often not possible to know ahead of time whether any one setup might work. Drivers is one thing, but a pro-studio setup has more moving parts, and so if you don't know ahead of time whether any one setup will work, then it can be an expensive process to walk through the different options. And that's a problem that neither Windows nor Mac has. It's a real shame, because the stability of Linux should make it the best platform for pro-audio.

[1] https://www.ferrofish.com/a32pro-dante-converter-multimode/

[2] https://solidstatelogic.com/products/ssl-2-plus-mkii

[3] https://rme-audio.de/digiface-dante.html

replies(1): >>44662396 #
3. jancsika ◴[] No.44662396[source]
> it cost $13467 to leave Windows

I rather think it cost you $mac_mini to buy a mac mini, and $compulsion buying hardware for reasons I still do not understand.

Paul Davis has lurked here at least as long as you have, and it would have cost $0 just to ask if that card is currently supported in Linux.

I mean, for $13467 I bet I could buy a plane ticket to Shenzhen, hire a translator, and have them send an email to Collabora to quote a price to develop the firware/driver I can afford with the money I have left over.

replies(1): >>44662752 #
4. louthy ◴[] No.44662752{3}[source]
> Paul Davis has lurked here at least as long as you have, and it would have cost $0 just to ask if that card is currently supported in Linux

I didn’t need to ask him, I already owned it, I just needed to dual boot Linux to find out, it cost me $0.

You don’t seem to understand that a soundcard needs connecting to everything else in a studio, so there’s no such thing as just changing one thing and it not having a knock on effect (unless you’re really lucky, which I wasn’t).

You also don’t seem to know that once a setup is right, it can last a decade or more, so getting the right combo of gear to minimise friction in a studio is worth it over time, even if it is expensive upfront.

If it makes you feel a little less morally superior, I sold the original soundcard and the two replaced AD/DAs for ~$7200.

And, you still miss the point of the story completely: the point was that it’s too risky for anyone considering building a pro setup on Linux. Especially compared to Windows and macOS where everything is plug and play.

That doesn’t mean there aren’t pockets of success in Linux-land, but that it can be costly in money and time to get it right, and it might never work for the setup you have.

It is risk.

> for reasons I still do not understand.

That’s obvious.

But the reasons are:

* I wanted to move away from Windows because it was unstable and pissing me off

* I already owned a high-end PCI RME card that connected to three Ferrofish A32 Pro converters

* If I could install Linux and have the RME card work then I wouldn’t need to change my studio setup

* There’s no official or stable driver

* So, a change to the setup was required

* To try and future proof the setup I looked to modern protocols like Dante and AES67 as they are taking over pro studios and are much more flexible — I also thought there was a reasonable chance it would work on Linux

* I couldn’t get it working on Linux

* Time is not infinite

* Therefore I bought a Mac for audio

* To avoid the expense of a Mac Pro I had to switch from a PCI based soundcard to a USB based soundcard (which I could plug in to the Mac Mini)

* I still use Linux on the original machine (for dev work), but with a class compliant soundcard for casual use. It’s relatively trouble free, other than half of my usb ports don’t work, but you know, meh

* I haven’t needed to use Windows since. So I consider it a win.