Once "kexts" are blocked fully, will pro-audio suffer on the mac?

[klong]

I've been having a bit of a back and forth with a random redditor. I've conceded that in terms of raw latency numbers etc, Windows ASIO drivers for systems like RME are apparently the best in the business. On mac, UA and other companies currently use "kernel extensions" which as we all know, we now need to go into deep security settings and disable security features to continue to use.

My question is - have other vendors already found a solution to this without compromising on latency etc? "Core Audio" inherently adds latency, does it not?

 

[JamesNorth]

Steinberg added the ASIO stack to Windows because it didn’t have something like Core Audio to begin with.

Is RME on Mac really slower? RME make no compromise speedy drivers like Presonus and don’t really have the need to have buffers for running upsampled analog emulation plugins.

The kext thing is just permission based - system vs user space. I’m sure all manufacturers will have the time to sort it out by the time it comes to be necessary.

Kexts are drivers like dlls on windows really. Windows just calls its kernel extensions a different name.

 

[mimp]

I think the RME explainer says basically yes, in terms of raw latency it will be worse with user mode/driverkit drivers as you’re more at the mercy of the OS, but in terms of system stability that’s a good thing. My experience was definitely worse with the driverkit versions for their hardware, but they’re still in active development so we’ll see (RME maintain both kernel extension and driverkit versions of their drivers so you can choose)

 

[JamesNorth]

That’s interesting insight - I hadn’t considered that latency would be *noticeably* worse by making this change.

I’d expect it was a theoretical but not noticeable change.

I guess there’s also the possibility that Apple does something with Core Audio to aid in this as things develop.

 

[Bear-Faced Cow]

I highly doubt that latency is going to be the issue. CoreAudio is a framework within the Mac OS. The drivers, whether in the user space or the system space are still being accessed in the same way by the CoreAudio framework through the hardware abstraction layer (HAL). Any change in latency is more indicative that the framework was bypassed or short circuited in some way rather than using the standard API calls, which developers have been known to do for the past 4 decades. This was often what made updating your OS a difficult and often scary task.

The more prominent issue will be sandboxing, which will most likely be prevalent in the case of UA because of the plug-ins more than the driver itself.

However, I wouldn’t go screaming that the sky is falling over the audio production industry since it took a considerable amount of time for developers to finally get AU sandboxing correct. Apple also less likely to put their music business in harm‘s way since they know there is a tie to their audio production software.

jord

 

[mimp]

To clarify, I didn’t get worse latency with driverkit, I got dropouts. Apparently they’re fixed by an OS update, but I’d already gone back to the kext by that point.