Intel Lunar Lake HD Audio & Other Sound Changes For Linux 6.4

[phoronix]

Phoronix: Intel Lunar Lake HD Audio & Other Sound Changes For Linux 6.4

Linux sound subsystem maintainer and SUSE engineer Takashi Iwai submitted all of the sound driver updates this week for the ongoing Linux 6.4 kernel merge window...

https://www.phoronix.com/news/Linux-6.4-Sound-Audio-Drivers

[caligula]

Does anyone know why there is so much driver development for audio? Apparently these audio codecs are designed for desktop & laptop PCs.

AFAIK Bluetooth offloads all processing to the headset. Most USB DACs simply use the standard USB audio class drivers. Motherboards typically have piece of shit quality audio via 3,5mm jacks, and almost nobody uses them. Also HTPC users typically offload the audio processing to a separate AVR receiver. So the computer only acts as a conduit, redirecting the DRM audio stream to the AVR gear. You can also decode most formats in software with minimal CPU load. E.g. decoding MP3s or AC3 uses less than 1% capacity of one core. It's the same for simple DSP effects like parametric equalizer or echo cancellation.

AFAIK there only used to be few common audio drivers for Sound Blaster / AC97 and other legacy hardware. Around 50 drivers in total for all PCI hardware ever built. Now, there are over 100 HD audio and ASoC drivers for hardware built during the last 5 or so years. This is so weird considering that most users listen to plain old stereo audio via Bluetooth or USB DACs.

[agd5f]

Does anyone know why there is so much driver development for audio? Apparently these audio codecs are designed for desktop & laptop PCs.

Modern audio hardware is a loosely defined collection of components (DMA engines, CODECs, AMPs, digital microphones, DSPs, etc.). As such you need a machine driver for each platform to enumerate and link all of these components together. Then each OEM can pick which combination of components they want to use in each SKU.

[caligula]

Modern audio hardware is a loosely defined collection of components (DMA engines, CODECs, AMPs, digital microphones, DSPs, etc.). As such you need a machine driver for each platform to enumerate and link all of these components together. Then each OEM can pick which combination of components they want to use in each SKU.

How is this any better? You need shit tons of drivers in the kernel. Still they're basically doing the same thing as classic Sound Blasters have done for the last 40 years.

[agd5f]

How is this any better? You need shit tons of drivers in the kernel. Still they're basically doing the same thing as classic Sound Blasters have done for the last 40 years.

Lower power, more flexible/customizable, easier to share component drivers? I'm not really an audio expert.