Linux Sound Subsystem Also Seeing Many Fixes Driven by AI/LLMs

phoronix.com

33 pointsdboon3 comments

Yeah I am noticing that the best thing about AI/LLM is that many small bugs are able to be fixed. Just the pattern recognition alone seems to solve many things

they said something about VMs.

I've always wondered why VMs are so terrible at sound handling, even with a beefy machine.

Only way I've been able to get good sound from a VM is to pass though a USB controller and plug in an DAC

VMs don’t typically own their own physical memory, so you randomly page vault. CPUs can can be preempted and system calls and other instructions can fault for arbitrary lengths of time in the hypervisor. It’s very hard to get real time scheduling when so many sources of latency exist and are outside of your control.

Reply on news.ycombinator.com